perm filename F81.IN[LET,JMC]1 blob
sn#632537 filedate 1982-01-04 generic text, type C, neo UTF8
COMMENT ⊗ VALID 00452 PAGES
C REC PAGE DESCRIPTION
C00001 00001
C00048 00002 ∂01-Oct-81 1556 CSD.TAJNAI at SU-SCORE (Carolyn Tajnai) Stanford Faculty Calendar
C00050 00003 ∂01-Oct-81 1605 CSD.TAJNAI at SU-SCORE (Carolyn Tajnai) MTC Qual
C00052 00004 ∂01-Oct-81 1628 SIS Colloq. Correction for Week of October 5th
C00053 00005 ∂02-Oct-81 0546 Chandra-at-OhioState <Chandrasekaran at RUTGERS> Proposal from Dr Teller
C00055 00006 ∂02-Oct-81 0748 CSD.TAJNAI at SU-SCORE (Carolyn Tajnai)
C00056 00007 ∂02-Oct-81 0922 RINDFLEISCH@SUMEX-AIM (SuNet) $ FIGURES FOR 11/750 FILE SERVER
C00060 00008 ∂02-Oct-81 0932 FFL
C00061 00009 ∂02-Oct-81 1410 RPG@Sail (SuNet)
C00064 00010 ∂02-Oct-81 1546 FFL
C00065 00011 ∂02-Oct-81 1547 FFL
C00066 00012 ∂02-Oct-81 1605 William Griffiths <CSD.GRIFFITHS at SU-SCORE> 390 course Spring 1980
C00070 00013 ∂02-Oct-81 1717 Jeffrey D. Ullman <CSD.ULLMAN at SU-SCORE>
C00071 00014 ∂02-Oct-81 2110 RPG@Sail (SuNet) David Luckham's wishlist
C00074 00015 ∂03-Oct-81 0017 LLW Conference Report
C00079 00016 ∂03-Oct-81 1113 CLT SEMINAR IN LOGIC AND FOUNDATIONS OF MATHEMATICS
C00080 00017 ∂04-Oct-81 0141 Feigenbaum at SUMEX-AIM call for LISP-oriented meeting
C00082 00018 ∂04-Oct-81 1138 RPG Lisp
C00084 00019 ∂04-Oct-81 1410 RPG Lisp
C00088 00020 ∂04-Oct-81 1523 RPG
C00089 00021 ∂04-Oct-81 1815 Feigenbaum at SUMEX-AIM Re: Lisp
C00090 00022 ∂05-Oct-81 1102 RPG Lisp timing
C00092 00023 ∂05-Oct-81 1215 ullman@Diablo (SuNet)
C00096 00024 ∂05-Oct-81 1347 SIS phone msg
C00097 00025 ∂05-Oct-81 1359 SL LISP, Vax, etc.
C00098 00026 ∂05-Oct-81 2047 V. Ellen Golden <ELLEN at MIT-MC> MACLisp Manuals
C00100 00027 ∂06-Oct-81 0915 PN Accounts
C00103 00028 ∂07-Oct-81 0835 FFL
C00104 00029 ∂07-Oct-81 1240 VRP via Ethernet host 50#301 Luca Cardelli
C00105 00030 ∂07-Oct-81 1545 Denny Brown <CSD.DBROWN at SU-SCORE> CS200
C00106 00031 ∂07-Oct-81 1955 Kanerva at SUMEX-AIM First meeting of the Context Advisory Council
C00107 00032 ∂08-Oct-81 0920 SIS Colloquium Notice of October 12 - 16, 1981
C00110 00033 ∂08-Oct-81 1232 RPG@Sail (SuNet) Lists
C00116 00034 ∂09-Oct-81 0223 pratt@Diablo (SuNet) Proposed budget changes
C00126 00035 ∂09-Oct-81 1329 FFL
C00127 00036 ∂09-Oct-81 1641 CLT SEMINAR IN LOGIC AND FOUNDATIONS OF MATHEMATICS
C00129 00037 ∂10-Oct-81 0313 Arthur Keller <CSD.KELLER at SU-SCORE> SAIL Bucks
C00132 00038 ∂10-Oct-81 1601 ARK S-1, SAIL, and JMC's comments
C00133 00039 ∂10-Oct-81 1855 Kanerva at SUMEX-AIM Thank you
C00134 00040 ∂10-Oct-81 2312 CLT Corrected announcement
C00135 00041 ∂11-Oct-81 0017 Lowell Wood <LLW at S1-A> Welcome!
C00137 00042 ∂11-Oct-81 0033 ME on TTY67 (at TV-120) 0033
C00138 00043 ∂11-Oct-81 0038 ME on TTY67 (at TV-120) 0038
C00139 00044 ∂11-Oct-81 1801 LGC Lunch Tomorrow
C00141 00045 ∂11-Oct-81 1917 Grosz at SRI-AI Re: Program Committee Chairman
C00149 00046 ∂12-Oct-81 0759 Nilsson at SRI-AI (Response to message)
C00150 00047 ∂12-Oct-81 0946 FFL CALL FROM DR. RUSSELL 415 654 7458
C00151 00048 ∂12-Oct-81 1011 JJW
C00152 00049 ∂12-Oct-81 1046 LGC
C00153 00050 ∂12-Oct-81 1049 JMC seminar
C00154 00051 ∂12-Oct-81 1111 CLT
C00155 00052 ∂12-Oct-81 1112 REG
C00157 00053 ∂12-Oct-81 1326 FFL
C00158 00054 ∂12-Oct-81 1726 David R. Cheriton <CSD.CHERITON at SU-SCORE> Winter quarter dept. seminars
C00159 00055 ∂12-Oct-81 2229 Don Walker <WALKER at SRI-AI> [Marvin Minsky <MINSKY at MIT-AI>: Program Committee Chairman]
C00162 00056 ∂12-Oct-81 2242 Don Walker <WALKER at SRI-AI> [Marvin Minsky <MINSKY at MIT-AI>:]
C00170 00057 ∂12-Oct-81 2344 JK EKL transportation
C00171 00058 ∂13-Oct-81 0029 energy at MIT-MC
C00172 00059 ∂13-Oct-81 0123 Jeff Rubin <JBR at S1-A>
C00174 00060 ∂13-Oct-81 0718 csl.jlh at SU-SCORE (John Hennessy) Computer facilities meeting
C00176 00061 ∂13-Oct-81 0806 Nilsson at SRI-AI (Response to message)
C00177 00062 ∂13-Oct-81 0831 Feigenbaum at SUMEX-AIM Re: Computer facilities meeting
C00181 00063 ∂13-Oct-81 1106 Irmgild Schack <CSD.IRMGILD at SU-SCORE> Faculty Lunch
C00182 00064 ∂13-Oct-81 1112 Baskett at PARC-MAXC Re: Computer facilities meeting
C00183 00065 ∂13-Oct-81 1116 FFL
C00184 00066 ∂13-Oct-81 1129 RWW reference
C00187 00067 ∂13-Oct-81 2050 RAH via S1-GATEWAY visit
C00188 00068 ∂13-Oct-81 2359 RAH via S1-GATEWAY visit2
C00189 00069 ∂14-Oct-81 0205 Jerry E. Pournelle <POURNE at MIT-MC> shackleton
C00191 00070 ∂14-Oct-81 0937 LOUNGO at RUTGERS Rutgers Computer Science Technical Reports
C00194 00071 ∂14-Oct-81 1506 John McCarthy <JMC at S1-A>
C00195 00072 ∂14-Oct-81 1702 Denny Brown <CSD.DBROWN at SU-SCORE> This week's CS200 prod
C00197 00073 ∂15-Oct-81 1528 JK ekl
C00199 00074 ∂15-Oct-81 2353 pratt@Diablo (SuNet) Dolphin timings
C00206 00075 ∂16-Oct-81 0812 Betty Scott <CSD.BSCOTT at SU-SCORE> [Scott at SUMEX-AIM: YOUR MAIL]
C00208 00076 ∂16-Oct-81 0859 FFL /
C00209 00077 ∂16-Oct-81 0915 FFL
C00210 00078 ∂16-Oct-81 0916 FFL
C00211 00079 ∂16-Oct-81 1000 RINDFLEISCH@SUMEX-AIM (SuNet) Re: Dolphin timings
C00214 00080 ∂16-Oct-81 1117 JEF via Ethernet host 50#12 summit meeting
C00215 00081 ∂16-Oct-81 1240 RPG@Sail (SuNet) Fairness
C00217 00082 ∂16-Oct-81 1301 Wiederhold at SUMEX-AIM Dick gabriels sail account.
C00218 00083 ∂16-Oct-81 1428 SIS Colloquium Notice for October 19 - 23, 1981
C00221 00084 ∂16-Oct-81 1605 FFL CIS Committee meeting
C00223 00085 ∂16-Oct-81 2230 RPG
C00224 00086 ∂16-Oct-81 2234 RPG
C00225 00087 ∂16-Oct-81 2320 SL TOB
C00226 00088 ∂17-Oct-81 0032 Lowell Wood <LLW at S1-A> Timing
C00233 00089 ∂17-Oct-81 1134 Jeffrey D. Ullman <CSD.ULLMAN at SU-SCORE> meeting
C00234 00090 ∂17-Oct-81 1340 Lowell Wood <LLW at S1-A> More On Privacy
C00239 00091 ∂17-Oct-81 1538 CSD.NOVAK@SU-SCORE (SuNet) Dolphin timings
C00246 00092 ∂17-Oct-81 1610 CLT
C00247 00093 ∂17-Oct-81 1750 pratt@SU-HPP-VAX (SuNet) Dolphin timings
C00252 00094 ∂17-Oct-81 2340 pratt@Diablo (SuNet) Fairness
C00262 00095 ∂18-Oct-81 0055 RPG@Sail (SuNet) For what it's worth
C00264 00096 ∂18-Oct-81 0858 JK
C00265 00097 ∂18-Oct-81 2141 pratt@Diablo (SuNet) For what it's worth
C00267 00098 ∂18-Oct-81 2254 RPG@Sail (SuNet) Several points:
C00271 00099 ∂18-Oct-81 2354 RPG
C00277 00100 ∂19-Oct-81 0938 RINDFLEISCH@SUMEX-AIM (SuNet) FYI - Other Lisp Timing Thrashes
C00285 00101 ∂19-Oct-81 0939 FFL
C00286 00102 ∂19-Oct-81 1000 JMC*
C00287 00103 ∂19-Oct-81 1000 JMC*
C00288 00104 ∂19-Oct-81 1046 pratt@Diablo (SuNet) Several points:
C00304 00105 ∂19-Oct-81 1205 Engelmore at SUMEX-AIM ARPA funding
C00306 00106 ∂19-Oct-81 1447 Jrobinson at SRI-AI Tinlunch readings for this Thursday
C00307 00107 ∂19-Oct-81 1525 Gene Golub <CSD.GOLUB at SU-SCORE> Faculty Appointment
C00308 00108 ∂20-Oct-81 0140 Jerry E. Pournelle <POURNE at MIT-MC> at last...
C00309 00109 ∂20-Oct-81 0900 RPG Terminal
C00310 00110 ∂20-Oct-81 1508 FFL
C00311 00111 ∂20-Oct-81 1543 Gene Golub <CSD.GOLUB at SU-SCORE> Re: Allen and Schwartz
C00312 00112 ∂21-Oct-81 2008 Denny Brown <CSD.DBROWN at SU-SCORE> My plans: Now to 1/82
C00314 00113 ∂21-Oct-81 2011 Irmgild Schack <CSD.IRMGILD at SU-SCORE> Schedule for J. Schwartz and F. Allen
C00316 00114 ∂21-Oct-81 2016 RINDFLEISCH@SUMEX-AIM (SuNet) [T. C. Rindfleisch <RINDFLEISCH>: System Industries Periphs for VAX]
C00320 00115 ∂21-Oct-81 2136 ullman@Diablo (SuNet) meeting
C00322 00116 ∂21-Oct-81 2212 Tom McWilliams <TM at S1-A> Schwartz talk and dinner at Louie's
C00324 00117 ∂22-Oct-81 0841 SIS Computer Science Colloquium Notice of October 26 - 30, 1981
C00327 00118 ∂22-Oct-81 1143 JMC
C00328 00119 ∂22-Oct-81 1529 Jrobinson at SRI-AI Tinlunch ALERT!
C00329 00120 ∂22-Oct-81 1335 RWW
C00330 00121 ∂22-Oct-81 2009 mogul@Diablo (SuNet)
C00342 00122 ∂23-Oct-81 0017 Robert Elton Maas <REM at MIT-MC>
C00343 00123 ∂23-Oct-81 1114 Grosz at SRI-AI Talk on Thursday Oct. 29 instead of tinlunch reading
C00347 00124 ∂23-Oct-81 1240 mogul@Diablo (SuNet) equipment
C00348 00125 ∂23-Oct-81 1514 Danny Hillis <DANNY at MIT-AI> Old NASA docs.
C00350 00126 ∂23-Oct-81 1545 JK proposal
C00351 00127 ∂23-Oct-81 1940 Mike Genesereth <CSD.GENESERETH at SU-SCORE> Chandra's proposal
C00355 00128 ∂24-Oct-81 2000 JMC*
C00356 00129 ∂24-Oct-81 2128 Lowell Wood <LLW at S1-A> `If Wishes Were Horses. . .`
C00357 00130 ∂25-Oct-81 0813 energy at MIT-MC
C00359 00131 ∂25-Oct-81 0906 Konolige at SRI-AI reply
C00360 00132 ∂25-Oct-81 1651 ME ftp
C00362 00133 ∂25-Oct-81 1656 ME ps
C00363 00134 ∂26-Oct-81 0021 RAH fiche
C00366 00135 ∂26-Oct-81 0850 CG
C00367 00136 ∂26-Oct-81 0923 C.S./Math Library <ADMIN.LIBRARY at SU-SCORE> NASA Reports
C00369 00137 ∂26-Oct-81 1005 FFL mail jmc,ffl
C00370 00138 ∂26-Oct-81 1150 TOB
C00372 00139 ∂26-Oct-81 1456 Irmgild Schack <CSD.IRMGILD at SU-SCORE> Advisory Committee (Stanford)
C00374 00140 ∂26-Oct-81 1528 FFL
C00375 00141 ∂27-Oct-81 0841 JK proposal
C00376 00142 ∂27-Oct-81 1007 FFL
C00377 00143 ∂27-Oct-81 1010 FFL
C00378 00144 ∂27-Oct-81 1159 Irmgild Schack <CSD.IRMGILD at SU-SCORE> Faculty Lunch
C00379 00145 ∂27-Oct-81 1406 RPG Re: Prolog
C00381 00146 ∂27-Oct-81 1631 RPG
C00382 00147 ∂27-Oct-81 2028 Gene Golub <CSD.GOLUB at SU-SCORE> Gripes
C00386 00148 ∂27-Oct-81 2233 Mike Farmwald <PMF at S1-A>
C00389 00149 ∂28-Oct-81 0859 ullman@Diablo (SuNet) There follows a version of the budget that reflects the discussions
C00390 00150 ∂28-Oct-81 0901 ullman@Diablo (SuNet)
C00395 00151 ∂28-Oct-81 1149 FFL
C00396 00152 We have to make some promises about deliverables.
C00399 00153 ∂28-Oct-81 1522 SIS Colloquium Notice - Week of November 2 - 6, 1981
C00402 00154 ∂28-Oct-81 2215 Marvin Minsky <MINSKY at MIT-AI> AAAI Corporate Fund-Raising Letter
C00405 00155 ∂29-Oct-81 0106 MINSKY@AI An idea.
C00413 00156 ∂29-Oct-81 0855 David R. Cheriton <CSL.DRC at SU-SCORE> Cuthbert Hurd
C00414 00157 ∂29-Oct-81 0858 Don Walker <WALKER at SRI-AI> Re: An idea.
C00416 00158 ∂29-Oct-81 0925 Aaai-Office at SUMEX-AIM Letter
C00419 00159 ∂29-Oct-81 0927 Nilsson at SRI-AI Re: An idea.
C00421 00160 ∂29-Oct-81 1150 Wiederhold at SUMEX-AIM Re: note from susan
C00422 00161 ∂29-Oct-81 1211 Marvin Minsky <MINSKY at MIT-MC> PROGRAM COMMITTEE CHAIRMAN
C00424 00162 ∂29-Oct-81 1212 RPG
C00426 00163 ∂29-Oct-81 1220 RPG Prolog contd.
C00428 00164 ∂29-Oct-81 1336 FFL
C00430 00165 ∂29-Oct-81 1620 Grosz at SRI-AI tinlunch reading
C00432 00166 ∂29-Oct-81 1911 RPG Meeting
C00433 00167 ∂30-Oct-81 0652 JK
C00434 00168 ∂30-Oct-81 1103 FFL
C00435 00169 ∂30-Oct-81 1534 JMM ekl at lots
C00436 00170 ∂30-Oct-81 1538 FFL
C00437 00171 ∂30-Oct-81 1657 Marvin Minsky <MINSKY at MIT-MC>
C00439 00172 ∂30-Oct-81 2013 Keith A. Lantz <CSL.LANTZ at SU-SCORE> New Systems References for the Comprehensive
C00443 00173 ∂31-Oct-81 0848 Engelmore at SUMEX-AIM Reminder: Deliverables
C00444 00174 ∂31-Oct-81 1117 CLT Guarneri
C00445 00175 ∂31-Oct-81 1632 Jeffrey D. Ullman <CSD.ULLMAN at SU-SCORE> ARPA proposal
C00446 00176 ∂01-Nov-81 1737 REM
C00447 00177 ∂01-Nov-81 2342 TOB computing
C00448 00178 ∂01-Nov-81 2347 TOB deliverables
C00449 00179 ∂02-Nov-81 0941 NAN p message
C00450 00180 ∂02-Nov-81 1000 JMC*
C00451 00181 ∂02-Nov-81 1118 NAN p message
C00452 00182 ∂02-Nov-81 1122 DCL
C00453 00183 ∂02-Nov-81 1327 FFL
C00454 00184 ∂02-Nov-81 1552 CG
C00455 00185 ∂02-Nov-81 1559 Denny Brown <CSD.DBROWN at SU-SCORE> Re: grader for cs206
C00456 00186 ∂02-Nov-81 1625 Irmgild Schack <CSD.IRMGILD at SU-SCORE> Meeting on Thursday, November 5
C00458 00187 ∂03-Nov-81 0958 SIS Correction to CS Colloq. Notice
C00459 00188 ∂03-Nov-81 1021 RPG Courses
C00460 00189 ∂03-Nov-81 1307 FFL
C00461 00190 ∂03-Nov-81 1414 FFL
C00462 00191 ∂03-Nov-81 1315 FFL
C00463 00192 ∂03-Nov-81 1547 Irmgild Schack <CSD.IRMGILD at SU-SCORE> Annual Faculty Reports
C00464 00193 ∂03-Nov-81 1605 JK ekl
C00465 00194 ∂03-Nov-81 1621 Jay Stewart <CSD.STEWART at SU-SCORE> Yes - I am in your course..
C00466 00195 ∂03-Nov-81 1629 FFL
C00467 00196 ∂03-Nov-81 2041 Denny Brown <CSD.DBROWN at SU-SCORE> cs200 lecture
C00468 00197 ∂04-Nov-81 0157 JMC
C00469 00198 ∂04-Nov-81 1311 FFL
C00470 00199 ∂04-Nov-81 1356 RPG
C00471 00200 ∂04-Nov-81 1418 Bob Engelmore <CSD.ENGELMORE at SU-SCORE> Deliverables
C00482 00201 ∂04-Nov-81 2108 SIS Colloquium Notice of November 9 - 13, 1981
C00485 00202 ∂04-Nov-81 2201 JMM Grader for CS206
C00487 00203 ∂05-Nov-81 0742 Grosz at SRI-AI tinlunch INSIDE today
C00488 00204 ∂05-Nov-81 0823 Shortliffe at SUMEX-AIM Re: Chandrasekaran at Ohio State
C00490 00205 ∂05-Nov-81 1411 FFL
C00491 00206 ∂05-Nov-81 1423 Denny Brown <CSD.DBROWN at SU-SCORE> Re: Grader for CS206
C00492 00207 ∂05-Nov-81 1612 SIS Updates to CS Colloquium Notice of
C00493 00208 ∂05-Nov-81 1707 CSD.ULLMAN@SU-SCORE (SuNet) equipment meeting
C00494 00209 ∂05-Nov-81 1747 JMC@Sail (SuNet) KA-10 as file server
C00495 00210 ∂05-Nov-81 2203 ME slow machine
C00496 00211 ∂06-Nov-81 0733 JK
C00498 00212 ∂06-Nov-81 0735 JK
C00499 00213 ∂06-Nov-81 0909 JK
C00500 00214 ∂06-Nov-81 1007 FFL
C00501 00215 ∂06-Nov-81 1111 FFL Message from Mr. Paolucci
C00502 00216 ∂06-Nov-81 1130 Bob Amsler <AMSLER at SRI-AI> TinLunch of Thursday November 12th - Automated Dictionaries
C00505 00217 ∂06-Nov-81 1230 Brian K. Reid <CSL.BKR at SU-SCORE> Finding the Potluck house
C00509 00218 ∂06-Nov-81 1234 Brian K. Reid <CSL.BKR at SU-SCORE> Re: Quarterly Potluck dinner!!!
C00510 00219 ∂06-Nov-81 1237 Ross Finlayson <RSF at SU-AI> Quarterly Potluck dinner!!!
C00512 00220 ∂06-Nov-81 1240 JK
C00513 00221 ∂06-Nov-81 1452 Carolyn Tajnai <CSD.TAJNAI at SU-SCORE> 1982 Annual Computer Forum Meeting
C00514 00222 ∂06-Nov-81 1530 Danny Berlin <CSD.BERLIN at SU-SCORE> Autumn 1980 Course Evaluations
C00517 00223 ∂06-Nov-81 1535 Carolyn Tajnai <CSD.TAJNAI at SU-SCORE> 1982 Annual Computer Forum Meeting
C00518 00224 ∂06-Nov-81 2352 RPG
C00519 00225 ∂07-Nov-81 0000 RPG@sail (suNet) Meeting
C00521 00226 ∂07-Nov-81 0823 pratt@Diablo (SuNet) Meeting
C00523 00227 ∂07-Nov-81 1442 Konolige at SRI-AI reading committee
C00524 00228 ∂07-Nov-81 1452 RPG@Sail (SuNet) Meeting
C00526 00229 ∂08-Nov-81 0856 JK
C00529 00230 ∂09-Nov-81 0023 RPG
C00530 00231 ∂09-Nov-81 0900 JMC*
C00531 00232 ∂09-Nov-81 1030 RWW VACATION
C00532 00233
C00541 00234 ∂09-Nov-81 1848 Bob Amsler <AMSLER at SRI-AI> "The Automated Dictionary" by Fox, Bebel and Parker
C00542 00235 ∂10-Nov-81 0912 JK
C00543 00236 ∂10-Nov-81 1021 JJW Samelength proof
C00544 00237 ∂10-Nov-81 1134 Irmgild Schack <CSD.IRMGILD at SU-SCORE> Faculty Lunch !Reminder!
C00545 00238 ∂11-Nov-81 0114 LGC Advice Taker
C00549 00239 ∂11-Nov-81 1000 FFL
C00550 00240 ∂11-Nov-81 1114 Danny Berlin <CSD.BERLIN at SU-SCORE> Course Evaluations for Aut 1980
C00553 00241 ∂11-Nov-81 1248 Gene Golub <CSD.GOLUB at SU-SCORE> [John McCarthy <JMC at SU-AI>:]
C00555 00242 ∂11-Nov-81 1249 Gene Golub <CSD.GOLUB at SU-SCORE> BELL RESPONSE
C00556 00243 ∂11-Nov-81 1446 Jeffrey D. Ullman <CSD.ULLMAN at SU-SCORE> equip.
C00558 00244 ∂11-Nov-81 1629 Irmgild Schack <CSD.IRMGILD at SU-SCORE> Vaughn Pratt Appointment
C00559 00245 ∂11-Nov-81 1650 Danny Berlin <CSD.BERLIN at SU-SCORE>
C00561 00246 ∂11-Nov-81 2151 ENGELMORE at SUMEX-AIM Re: equip.
C00564 00247 ∂12-Nov-81 0929 Konolige at SRI-AI meeting
C00565 00248 ∂12-Nov-81 0953 OPERATOR at SRI-AI TinLunch Today on "Automated Dictionaries" in EK242 at noon
C00566 00249 ∂12-Nov-81 1421 Jrobinson at SRI-AI Tinlunch reading for next week.
C00567 00250 ∂12-Nov-81 1511 Konolige at SRI-AI (Kurt Konolige)
C00568 00251 ∂12-Nov-81 1519 Jrobinson at SRI-AI Tinlunch abstract
C00602 00252 ∂12-Nov-81 1616 FFL
C00603 00253 ∂12-Nov-81 1614 SIS Computer Science Colloquium for November 16 - 20, 1981
C00606 00254 ∂12-Nov-81 2141 Brian K. Reid <CSL.BKR at SU-SCORE> Re: equip.
C00608 00255 ∂13-Nov-81 0849 Kanerva at SUMEX-AIM Cash problems
C00610 00256 ∂13-Nov-81 1056 Sharon.Burks at CMU-10A Your visit next week
C00612 00257 ∂13-Nov-81 1054 CLT driveway
C00613 00258 ∂13-Nov-81 1326 Konolige at SRI-AI meeting
C00614 00259 ∂13-Nov-81 2248 Danny Hillis <DANNY at MIT-AI>
C00618 00260 ∂13-Nov-81 2313 CLT SEMINAR IN LOGIC AND FOUNDATIONS OF MATHEMATICS
C00619 00261 ∂14-Nov-81 0031 Jerry E. Pournelle <POURNE at MIT-MC> report
C00621 00262 ∂14-Nov-81 0032 Jerry E. Pournelle <POURNE at MIT-MC>
C00622 00263 ∂14-Nov-81 1929 Lowell Wood <LLW at S1-A> Grubbing For Sustenance, cont'd.
C00630 00264 ∂15-Nov-81 0008 Lowell Wood <LLW at S1-A> Base Powering
C00634 00265 ∂15-Nov-81 0030 Lowell Wood <LLW at S1-A> Division Of Labor
C00637 00266 Even before I got this message, I was about to advocate asking Carolyn
C00639 00267 ∂15-Nov-81 2337 Rod Hyde <RAH at S1-A> We got the first four microfiche documents you sent. The next two,
C00643 00268 ∂17-Nov-81 1019 FFL
C00646 00269 ∂17-Nov-81 1121 REG SAIL Payers Meeting
C00647 00270 ∂17-Nov-81 1352 FFL
C00648 00271 ∂17-Nov-81 1424 FFL
C00649 00272 ∂17-Nov-81 1915 Oded Anoaf Feingold <OAF at MIT-MC> energy digest #3004
C00655 00273 ∂17-Nov-81 2239 Bob Engelmore <CSD.ENGELMORE at SU-SCORE> Partial Proposal Draft
C00689 00274 ∂18-Nov-81 2046 Gene Golub <CSD.GOLUB at SU-SCORE> Pratt
C00690 00275 ∂18-Nov-81 2102 Gene Golub <CSD.GOLUB at SU-SCORE>
C00691 00276 ∂19-Nov-81 0807 FFL
C00692 00277 ∂19-Nov-81 0944 NAN pmessage
C00693 00278 ∂19-Nov-81 0959 FFL
C00694 00279 ∂19-Nov-81 1002 FFL
C00695 00280 ∂19-Nov-81 1143 Jrobinson at SRI-AI Tinlunch today in EK242.
C00696 00281 ∂19-Nov-81 1304 FFL
C00697 00282 ∂19-Nov-81 1422 LGC Reading Kurt Konolige's Thesis
C00698 00283 ∂20-Nov-81 1145 Guy.Steele at CMU-10A Time/place of Common LISP Meeting
C00700 00284 ∂20-Nov-81 1200 Gene Golub <CSD.GOLUB at SU-SCORE> Lunch with C. Hurd
C00702 00285 ∂20-Nov-81 1332 Irmgild Schack <CSD.IRMGILD at SU-SCORE> Faculty Lunch
C00704 00286 ∂20-Nov-81 1354 Gene Golub <CSD.GOLUB at SU-SCORE> [Irmgild Schack <CSD.IRMGILD>: Faculty Lunch]
C00706 00287 ∂20-Nov-81 1858 CLT SEMINAR IN LOGIC AND FOUNDATIONS OF MATHEMATICS
C00707 00288 ∂20-Nov-81 2056 JK ekl
C00708 00289 ∂21-Nov-81 1259 JMC
C00710 00290 ∂23-Nov-81 0923 ullman@Diablo (SuNet) ARPA proposal
C00712 00291 ∂23-Nov-81 1734 Woody Bledsoe <ATP.Bledsoe at UTEXAS-20> Re: Turing award for Boyer and Moore
C00714 00292 ∂23-Nov-81 2023 CLT
C00715 00293 ∂24-Nov-81 1225 CSD.ULLMAN@SU-SCORE (SuNet) meeting
C00716 00294 ∂24-Nov-81 1248 CLT
C00717 00295 ∂24-Nov-81 1536 Tom Wadlow <TAW at S1-A> Kantrowitz article
C00719 00296 ∂24-Nov-81 2123 Marvin Minsky <MINSKY at MIT-AI>
C00720 00297 ∂24-Nov-81 2145 Vaughan Pratt <CSD.PRATT at SU-SCORE> eiuolcs
C00727 00298 ∂25-Nov-81 0050 Howard D. Trachtman <HDT at MIT-AI>
C00728 00299 ∂25-Nov-81 1025 Woody Bledsoe <ATP.Bledsoe at UTEXAS-20> [CS.DALE: Turing Award]
C00730 00300 ∂25-Nov-81 1411 JK disk allocation
C00732 00301 ∂26-Nov-81 0018 RWW thanksgiving
C00733 00302 ∂26-Nov-81 0035 Robert Elton Maas <REM at MIT-MC>
C00736 00303 ∂28-Nov-81 1150 Purger exceeding your disk quota
C00737 00304 ∂28-Nov-81 1418 RWW
C00738 00305 ∂30-Nov-81 0000 JMC*
C00739 00306 ∂30-Nov-81 1010 JJW Comprehensive Committee
C00740 00307 ∂30-Nov-81 1135 JEF via Ethernet host 50#12 equip proposal
C00741 00308 ∂30-Nov-81 1400 Gene Golub <CSD.GOLUB at SU-SCORE> Lunch on Tuesday
C00742 00309 ∂30-Nov-81 1425 FFL Call from Sarah Lippincott, New Yorker magazine
C00743 00310 ∂30-Nov-81 1433 Gene Golub <CSD.GOLUB at SU-SCORE> Monthly meeting
C00745 00311 ∂30-Nov-81 1542 Nancy Dorio <CSD.DORIO at SU-SCORE> pmessage
C00746 00312 ∂30-Nov-81 1616 FFL
C00747 00313 ∂01-Dec-81 1104 SIS Colloquium Notice of December 7 - 11, 1981
C00751 00314 ∂01-Dec-81 1131 FFL
C00752 00315 ∂01-Dec-81 1132 FFL
C00753 00316 ∂01-Dec-81 1443 CLT
C00754 00317 ∂01-Dec-81 1501 Gene Golub <CSD.GOLUB at SU-SCORE> Re: Selling Perseus computer time
C00755 00318 ∂01-Dec-81 1607 Jeffrey D. Ullman <CSD.ULLMAN at SU-SCORE> Re: meeting
C00758 00319 ∂01-Dec-81 2130 William G. Dubuque <WGD at MIT-MC> permutation groups breakthrough?
C00761 00320 ∂01-Dec-81 2351 JPM Meeting tonight
C00762 00321 ∂02-Dec-81 0641 Marvin Minsky <MINSKY at MIT-ML>
C00771 00322 ∂02-Dec-81 1214 FFL
C00772 00323 ∂02-Dec-81 1532 jjf@Shasta (SuNet) Lisp Proceedings
C00773 00324 ∂02-Dec-81 1535 FFL
C00774 00325 ∂02-Dec-81 1649 Robert Schreiber <CSD.SCHREIBER at SU-SCORE> Dinner with speaker
C00775 00326 ∂02-Dec-81 1800 JMC*
C00776 00327 ∂03-Dec-81 0300 LGC Disk Allocation
C00777 00328 ∂03-Dec-81 1008 pratt@Shasta (SuNet) benchmarks
C00781 00329 ∂03-Dec-81 1011 Gene Golub <CSD.GOLUB at SU-SCORE> ASSISTANTSHIP RENEWAL
C00782 00330 ∂03-Dec-81 1046 WOL Zohar is coming
C00784 00331 ∂04-Dec-81 1449 Jrobinson at SRI-AI TINLUNCH DEC. 10
C00785 00332 ∂04-Dec-81 1656 CG
C00788 00333 ∂06-Dec-81 0033 pratt@Shasta (SuNet) eyboards
C00807 00334 ∂07-Dec-81 0421 JPM Meeting on Friday
C00809 00335 ∂07-Dec-81 0818 CL.MOORE at UTEXAS-20 more on brown thumb
C00815 00336 ∂07-Dec-81 1012 Robert Schreiber <CSD.SCHREIBER at SU-SCORE> Visit by S. Wolfram
C00817 00337 ∂07-Dec-81 1021 FFL
C00818 00338 ∂07-Dec-81 1041 Gene Golub <CSD.GOLUB at SU-SCORE> Lunch on Tuesday
C00819 00339 ∂07-Dec-81 1939 Dave Waltz <DWaltz at BBN-TENEXD> AAAI-82 CONFERENCE PROGRAM COMMITTEE
C00823 00340 ∂07-Dec-81 1946 YM MTC Quals
C00824 00341 ∂07-Dec-81 2247 TOB
C00825 00342 ∂07-Dec-81 2256 TOB
C00826 00343 ∂07-Dec-81 2258 TOB lisp machines
C00827 00344 ∂08-Dec-81 0904 FFL
C00828 00345 ∂08-Dec-81 1006 CG
C00830 00346 ∂08-Dec-81 1131 CLT
C00831 00347 ∂08-Dec-81 1139 FFL mail jmc,ffl
C00832 00348 ∂08-Dec-81 1533 Jrobinson at SRI-AI Talk and Tinlunch on Thursday
C00835 00349 ∂08-Dec-81 1549 CG mtc qual
C00836 00350 ∂08-Dec-81 1650 JDH chess problem
C00839 00351 ∂08-Dec-81 1728 TOB what do you think
C00840 00352 ∂08-Dec-81 2039 Mike Slocum <MAS at S1-A>
C00842 00353 ∂08-Dec-81 2107 Mike Slocum <MAS at S1-A>
C00845 00354 ∂08-Dec-81 2320 Mike Slocum <MAS at S1-A> yes
C00846 00355 ∂09-Dec-81 0109 Jerry E. Pournelle <POURNE at MIT-MC> urgent
C00851 00356 ∂09-Dec-81 0126 Jerry E. Pournelle <POURNE at MIT-MC>
C00853 00357 ∂09-Dec-81 0133 Jerry E. Pournelle <POURNE at MIT-MC>
C00855 00358 ∂09-Dec-81 0838 FFL Reminder from Gene Golub's office
C00856 00359 ∂09-Dec-81 0853 Gene Golub <CSD.GOLUB at SU-SCORE> [Gene Golub <CSD.GOLUB>: Lunch with C. Hurd]
C00858 00360 ∂09-Dec-81 0953 FFL
C00859 00361 ∂09-Dec-81 1000 JMC*
C00860 00362 ∂09-Dec-81 1008 FFL
C00861 00363 ∂09-Dec-81 1230 Oded Anoaf Feingold <OAF at MIT-MC> bad mailing ...
C00863 00364 ∂09-Dec-81 1320 SIS Colloq. Schedule for Dec. 15th
C00864 00365 ∂09-Dec-81 1338 Gene Golub <CSD.GOLUB at SU-SCORE> Re: Lunch with C. Hurd
C00866 00366 ∂09-Dec-81 1354 Gene Golub <CSD.GOLUB at SU-SCORE> Re: Lunch with C. Hurd
C00868 00367 ∂09-Dec-81 1503 CG
C00870 00368 ∂09-Dec-81 1606 JMM MTC qual
C00871 00369 ∂09-Dec-81 1730 pratt@Shasta (SuNet) schedule
C00872 00370 ∂10-Dec-81 1010 Gene Golub <CSD.GOLUB at SU-SCORE> [SAMET at USC-ISIF: hello]
C00875 00371 ∂10-Dec-81 1013 DEK luncheon talk
C00879 00372 ∂10-Dec-81 1629 TW Comprehensive committee
C00880 00373 ∂10-Dec-81 2115 lantz@Shasta (SuNet) ARPA gateway
C00881 00374 ∂11-Dec-81 0230 Robert Elton Maas <REM at MIT-MC> Permutation group algorithm/program
C00883 00375 ∂11-Dec-81 0245 Robert Elton Maas <REM at MIT-MC> permutation-group algorithm
C00886 00376 ∂11-Dec-81 0309 Robert Elton Maas <REM at MIT-MC>
C00890 00377 ∂11-Dec-81 0415 Robert Elton Maas <REM at MIT-MC>
C00892 00378 ∂11-Dec-81 0856 Ichiki at SRI-AI 12/17 TINLUNCH MEETING
C00895 00379 ∂11-Dec-81 1222 CG
C00897 00380 ∂11-Dec-81 1215 David A. Moon <MOON at MIT-MC> LISPM Array Timings
C00899 00381 ∂11-Dec-81 1224 CG correction #23
C00900 00382 ∂11-Dec-81 1224 PJH via ROCHESTER cobol
C00902 00383 ∂11-Dec-81 1243 Robert Elton Maas <REM at MIT-MC>
C00904 00384 ∂11-Dec-81 1839 Mike Slocum <MAS at S1-A> final.f81
C00906 00385 ∂12-Dec-81 0749 Robert Elton Maas <REM at MIT-MC> Update on PERG
C00907 00386 ∂14-Dec-81 1056 FFL
C00908 00387 ∂14-Dec-81 1124 Gene Golub <CSD.GOLUB at SU-SCORE> Lunch
C00909 00388 ∂14-Dec-81 1534 Gene Golub <CSD.GOLUB at SU-SCORE> Lunch on Dec 15 (Second Notice)
C00911 00389 ∂15-Dec-81 0118 Robert Elton Maas <REM at MIT-MC> PERG and Rubic
C00913 00390 ∂15-Dec-81 0341 Robert Elton Maas <REM at MIT-MC>
C00914 00391 ∂16-Dec-81 1015 FFL
C00915 00392 ∂16-Dec-81 1109 Ichiki at SRI-AI TINLUNCH MEETING IN BUILDING -A-
C00917 00393 ∂16-Dec-81 1345 Gene Golub <CSD.GOLUB at SU-SCORE> Applicant
C00918 00394 ∂16-Dec-81 1631 Marilynn Walker <CSD.MWALKER at SU-SCORE> TGIF FOR DENNY BROWN
C00919 00395 ∂16-Dec-81 1828 Gene Golub <CSD.GOLUB at SU-SCORE> Applicant
C00921 00396 ∂17-Dec-81 0346 Mike Slocum <MAS at S1-A> Homework number 4.
C00923 00397 ∂17-Dec-81 1051 FFL
C00924 00398 ∂17-Dec-81 1303 Gene Golub <CSD.GOLUB at SU-SCORE> Sabbatical Leave
C00925 00399 ∂17-Dec-81 1338 FFL
C00926 00400 ∂17-Dec-81 1429 MAS
C00927 00401 ∂17-Dec-81 1624 Gene Golub <CSD.GOLUB at SU-SCORE> Faculty
C00928 00402 ∂17-Dec-81 1643 TW Comprehensive committee
C00931 00403 ∂17-Dec-81 1811 CLT
C00932 00404 ∂18-Dec-81 1104 CLT
C00933 00405 ∂18-Dec-81 1318 Gene Golub <CSD.GOLUB at SU-SCORE> Gift to Department
C00934 00406 ∂19-Dec-81 1505 Gifford at PARC-MAXC Recursive Programs as Functions in a First Order Theory
C00935 00407 ∂19-Dec-81 1606 Gene Golub <CSD.GOLUB at SU-SCORE> Wadler Visit
C00936 00408 ∂20-Dec-81 0238 Jerry E. Pournelle <POURNE at MIT-MC>
C00937 00409 ∂20-Dec-81 0952 JMM exams
C00938 00410 ∂20-Dec-81 1307 JMM
C00939 00411 ∂20-Dec-81 1321 RWW
C00940 00412 ∂20-Dec-81 1506 Gifford at PARC-MAXC PS: Recursive Programs as Functions in a First Order Theory
C00941 00413 ∂20-Dec-81 1736 RWG
C00942 00414 ∂21-Dec-81 1042 LGC cc of msg to REG on SAIL phone hookup
C00943 00415 ∂21-Dec-81 1231 RWW
C00944 00416 ∂21-Dec-81 1436 Guy.Steele at CMU-10A Results of Common LISP Meeting
C00947 00417 ∂21-Dec-81 1449 Guy.Steele at CMU-10A Correction: files on MIT-MC, not MIT-AI
C00949 00418 ∂21-Dec-81 1530 Gene Golub <CSD.GOLUB at SU-SCORE> Au Pair Girl
C00950 00419 ∂21-Dec-81 1608 vanMelle, Masinter Re: Results of Common LISP Meeting
C00952 00420 ∂21-Dec-81 2125 Gene Golub <CSD.GOLUB at SU-SCORE> Faculty meeting Jan 5, 1982
C00953 00421 ∂22-Dec-81 1000 JMC*
C00954 00422 ∂22-Dec-81 1317 Marvin Minsky <MINSKY at MIT-AI> AAAI Tutorial Chairman
C00956 00423 ∂22-Dec-81 1348 Ichiki at SRI-AI TINLUNCHES FOR 1/7 AND 1/14
C00957 00424 ∂23-Dec-81 0900 JMC*
C00958 00425 ∂23-Dec-81 1412 TW meeting time
C00959 00426 ∂24-Dec-81 0900 JMC*
C00960 00427 ∂24-Dec-81 0922 Gene Golub <CSD.GOLUB at SU-SCORE>
C00961 00428 ∂24-Dec-81 1832 Lowell Wood <LLW at S1-A> Forward Motion
C00963 00429 ∂25-Dec-81 0900 JMC*
C00964 00430 ∂26-Dec-81 1546 ZM
C00965 00431 ∂27-Dec-81 1042 Marvin Minsky <MINSKY at MIT-MC>
C00966 00432 ∂27-Dec-81 1656 TOB Clara Torda
C00967 00433 ∂27-Dec-81 1732 Lowell Wood <LLW at S1-A> Pearls Before Swine
C00969 00434 ∂27-Dec-81 1847 Lowell Wood <LLW at S1-A> Needed Item
C00979 00435 ∂28-Dec-81 1527 FFL
C00980 00436 ∂28-Dec-81 1528 FFL Call from student at H-P in Colorado Springs, John Romano
C00983 00437 ∂28-Dec-81 1623 FFL
C00984 00438 ∂28-Dec-81 1656 FFL
C00985 00439 ∂28-Dec-81 1906 TOB
C00987 00440 ∂29-Dec-81 1223 Mike Farmwald <PMF at S1-A>
C00988 00441 ∂29-Dec-81 1625 Gene Golub <CSD.GOLUB at SU-SCORE> Philip Wadler
C00989 00442 ∂29-Dec-81 2319 Mike Slocum <MAS at S1-A> Fall cs206 final...
C00990 00443 ∂30-Dec-81 1117 Guy.Steele at CMU-10A Text-file versions of DECISIONS and REVISIONS documents
C00991 00444 ∂31-Dec-81 1152 RPG
C00992 00445 ∂31-Dec-81 1235 FFL
C00993 00446 ∂01-Jan-82 1000 JMC*
C00994 00447 ∂01-Jan-82 1428 YK takasu + igarashi
C00995 00448 ∂01-Jan-82 1500 JMC*
C00996 00449 ∂01-Jan-82 1600 Guy.Steele at CMU-10A Tasks: A Reminder and Plea
C01000 00450 ∂03-Jan-82 1824 Gene Golub <CSD.GOLUB at SU-SCORE> Cancellation of tenured faculty meeting
C01001 00451 ∂04-Jan-82 1108 FFL
C01002 00452 ∂04-Jan-82 1147 Irmgild Schack <CSD.IRMGILD at SU-SCORE> Tuesday Faculty Luncheon
C01003 ENDMK
C⊗;
∂01-Oct-81 1556 CSD.TAJNAI at SU-SCORE (Carolyn Tajnai) Stanford Faculty Calendar
Date: 1 Oct 1981 1551-PDT
From: CSD.TAJNAI at SU-SCORE (Carolyn Tajnai)
Subject: Stanford Faculty Calendar
To: CSD-Faculty:
The new Stanford Faculty Calendars are available at the Bookstore in the
office. The new policy is for the faculty member to pick the calendar up
in person.
-------
∂01-Oct-81 1605 CSD.TAJNAI at SU-SCORE (Carolyn Tajnai) MTC Qual
Date: 1 Oct 1981 1605-PDT
From: CSD.TAJNAI at SU-SCORE (Carolyn Tajnai)
Subject: MTC Qual
To: JMC at SU-AI
cc: JMM at SU-AI, JJW at SU-AI, CSD.Yellin at SU-SCORE, CSD.Malachi at SU-SCORE
John, Malik, Weening, Yellin, and Malachi have signed up for the
MTC Qual that is scheduled for Monday, December 14.
Zohar said that you wanted to add a reference to the reading list.
It is imperative that I have the updated reading list by tomorrow.
It should have been ready the first of the quarter.
Carolyn
-------
What file is the current reading list for MTC? I have forgotten what
I wanted to add, and I think it will remind me.
∂01-Oct-81 1628 SIS Colloq. Correction for Week of October 5th
To: "@COLLOQ.DIS[INF,CSD]" at SU-AI
The Computer Systems Laboratory Seminar was listed incorrectly --
The following is the correction:
COMPUTER SYSTEMS LABORATORY SEMINAR --- Wednesday, October 7, 1981 at
4:15 p.m. in Terman 153. Kim Harris, of Forthright Enterprises, Palo Alto
will speak on ``FORTH''.
∂02-Oct-81 0546 Chandra-at-OhioState <Chandrasekaran at RUTGERS> Proposal from Dr Teller
Date: 2 Oct 1981 0836-EDT
From: Chandra-at-OhioState <Chandrasekaran at RUTGERS>
Subject: Proposal from Dr Teller
To: JMC at SU-AI
I had talked with Dr Teller a few weeks ago about an idea
of mine concening AI and nuclear safety. He had said he was
going to send the proposal for your comments and suggestions.
I'd be delighted to discuss with you the ideas involved -- some of the
stuff in it perhaps needs clarification. In any case
please let me know what your response is. Thanks.
(I use the Rutgers address for ARPANET communication).
-------
∂02-Oct-81 0748 CSD.TAJNAI at SU-SCORE (Carolyn Tajnai)
Date: 2 Oct 1981 0747-PDT
From: CSD.TAJNAI at SU-SCORE (Carolyn Tajnai)
To: JMC at SU-AI
In-Reply-To: Your message of 1-Oct-81 1635-PDT
I don't know the surce file, but I'll make a hardcopy of what I have.
-------
∂02-Oct-81 0922 RINDFLEISCH@SUMEX-AIM (SuNet) $ FIGURES FOR 11/750 FILE SERVER
Date: 2 Oct 1981 0917-PDT
From: Rindfleisch at SUMEX-AIM
Subject: $ FIGURES FOR 11/750 FILE SERVER
To: Equip at SU-HPP-VAX
cc: [SUMEX] at SUMEX-AIM, ETHERNET at SUMEX-AIM, DEV at SUMEX-AIM, GRP:
Following are some first cut figures for disk/tape equipment that might
be used on an 11/750 file server as collected by Bob Tucker. The
numbers are from Systems Industries and include their "standard"
Stanford discount. They do not include any GSA discounts or other
possible wheeling and dealing SI has been know to do when pressed to the
wall with alternative quotes from other vendors. Preliminary technical
specs look OK in terms of things like the controller being able to do
parallel seeks and not losing revolutions on transfers but these need to
be investigated in more detail.
Configuration:
1) 6 CDC 675 Mbyte disks (4.05 Gbytes total).
Formatted capacity = 512 Mbytes each
=> 3.07 Gbytes total formatted.
Interface, controller, & 1st drive $ 31,365
5 additional drives ($24,225 each) 121,125
--------
$152,490
These are UNIBUS devices. S/I still has no
CMI interface for the VAX 11/750. When
available (July 82?), it will be one card
that plugs in to replace the Unibus interface.
It will cost about $8000 for an upgrade. The
transfer rate will remain at 1.2 Mbytes/sec.
Installation and 3 months prime time maint.
for the interface, controller, and 1st drive
runs $2190 and $1335 for additional drives.
Delivery: 30 to 45 days
2) 2 tri-density STC tapes - 800/1600/6250 BPI
at 125 IPS.
Interface, controller, & 1st drive $ 27,285
1 additional drive ($17,600 each) 17,600
--------
$ 44,885
S/I currently supplies only the Unibus interface.
A CMI interface will be available at some
unspecified time in the future.
Installation and 3 months prime time maint.
for the interface, controller, and 1st drive
runs $1665 and $1260 for additional drives.
Delivery: 30 days
TOTAL FILE SERVER:
VAX 11/750 $ 34,000
Disk System 152,490
Tape System 44,885
--------
$231,375
-------
∂02-Oct-81 0932 FFL
To: JMC, FFL
Call from TV Network that there was something the master with master controls
during the first ten minutes of your lecture. So they ask for an instant
replay of anything vital which you said then. I assume it included announce-
ment of the cost and method of securing classroom materials.
∂02-Oct-81 1410 RPG@Sail (SuNet)
Date: 02 Oct 1981 1404-PDT
From: Dick Gabriel <RPG at SU-AI>
To: equip at DIABLO, (sunet) at Sail, csd.genesereth at SU-SCORE
Mike Genesereth and I have done a head count, finally, of the Lisp community
as it exists at Stanford CSD and CSL. I have included the groups: HPP,
Formal Reasoning, Verification, and Zohar.
The results are a bit surprising: there are 60 Lisp users that we can identify.
Of these there are 36 who use a Lisp other than Interlisp. These will eventually
become Common Lisp users. 24 are InterLisp users. Of the entire Lisp community,
36 are HPP and 24 are non-HPP. Within HPP 24 use InterLisp and 12 non-InterLisp.
InterLisp Common Lisp
Hpp 24 12
Non-Hpp 0 24
Total 24 36
HPP 36
Non-HPP 24
I think that the responsible thing for us to do is to equip the Lisp community
according to this ratio; if individual groups want to supply otherwise,
then they may.
This means that of the 15 lisp machines, we should buy 9 Symbolics machines
and 6 Dolphins. 3 Symbolics machines will be for HPP primary use and
6 for non-HPP primary use. All the Dolphins will be for HPP primary use.
I know that this wmight not sit well with everyone on the committee,
but I am willling to listen to arguments against equity.
About future Lisp computing needs. When a large processor capable of running Lisp
well (e.q. an S-1) arrives, we should consider re-distributing things, though
I imagine that there will continue to be a need for Lisp cycles of all types.
-rpg-
∂02-Oct-81 1546 FFL
To: JMC, FFL
Dr. Bob Russell called re your speaking at the Conference on Cybernetics.
It is presently set for January 15, 1982, 4 to 6:30 p.m. He wants to
confirm the date with you. 415 654 7458. If date not good, is any
other Friday in January better.
∂02-Oct-81 1547 FFL
To: JMC, FFL
Sarah would like to have the paper from Stanford which gives her tuition
at David for the winter quarter, please.
∂02-Oct-81 1605 William Griffiths <CSD.GRIFFITHS at SU-SCORE> 390 course Spring 1980
Date: 2 Oct 1981 1605-PDT
From: William Griffiths <CSD.GRIFFITHS at SU-SCORE>
Subject: 390 course Spring 1980
To: jmc at SU-AI
My name is Bill Griffiths. I took 9 units of 390 from you during the Spring
quarter 1980 ( the work was supervised primarily by Carolyn Talcott ).
I intended to register TGR this quarter in order to study for the qualifying
exams to be given in February. When I Attempted to file for candidacy in order
to register TGR I discovered 2 problems. The first was that the grade for the
390 units Spring 1980 was an N. The second was that I didn't have any 293 units
on my record.
We had an interchange of mail (I believe during the Autumn quarter 1980)
regarding the N grade. The grade was to be changed to an A, but apparently
never got recorded.
The second problem is due to some bad advice I got at the end of the Winter
quarter of last year regarding the additional units I needed in order to
complete the degree requirements. Unfortunately 6 of the units I registered
for in the following Spring quarter should have been 293. As it stands I
have sufficient units, but none in 293. I had planned to do the programming
project after I completed the Qualifying Exams. I didn't realize the project
had to be done uder the auspices of 293.
What I would like to do, if you are agreeable, is to retain the N grade, change
the course 390 to 293, and do the project under the continuing 293. The project
I am planning to do is turn the work on pattern recognition that I have been
doing here (including the 390 units with you) into a program. Under this
arrangement I would be able to register TGR this quarter and you could replace
the N grade after satisfactory completion of the programming project. This
would save me a considerable amount of money and my academic program could be
completed as I had planned. Carolyn Tajnai assures me that the changes could be
made if you OK it. If you would like we could discuss it some afternoon next
week.
Thanks for your time.
Bill Griffiths
-------
∂02-Oct-81 1717 Jeffrey D. Ullman <CSD.ULLMAN at SU-SCORE>
Date: 2 Oct 1981 1714-PDT
From: Jeffrey D. Ullman <CSD.ULLMAN at SU-SCORE>
To: JMC at SU-AI
In-Reply-To: Your message of 2-Oct-81 0056-PDT
alas, you did.
-------
∂02-Oct-81 2110 RPG@Sail (SuNet) David Luckham's wishlist
Date: 02 Oct 1981 2108-PDT
From: Dick Gabriel <RPG at SU-AI>
Subject: David Luckham's wishlist
To: equip at DIABLO, (sunet) at Sail
I asked David what he needed to continue his work wrt Lisp after
I saw the CSL wishlist. I found it hard to believe that he wanted
Vaxes etc if he were still using Lisp heavily. Cheriton seemed to
think David was cognizant of things and that Vaxes was what he wanted.
A 2060 allows the Verification group to use MacLisp in a slightly
faster, more flexible environment than on SAIL. He also mentioned a
Lisp machine. In my accounting of Lisp users his group was counted,
and I assume he would use a Lisp Machine quite nicely for his work.
In my previous message about Lisp I talked about a 3/2 ratio of Symbolics
machines to Dolphins as the target for the ARPA re-equipment proposal.
I believe, though, that HPP might need further Dolphin enhancement,
unless there is some motion towards Common Lisp from InterLisp there
in the future. I think that given budget constraints HPP should acquire
these additional Dolphins with other money since the equipment committee
must address a larger community of Lisp users. If I had my real choice
I would have asked for 15 Dolphins and 24 Symbolics Machines, totalling
$3.1mbucks.
Someone mentioned eliminating the S-1 from the budget. This certainly, then,
has to mean that Lisp Machines replace the S-1, and in that case I would
have to ask for 39 Lisp machines and Dolphins over 3 years.
∂03-Oct-81 0017 LLW Conference Report
To: pourne at MIT-MC
CC: LLW at SU-AI, JMC at SU-AI
∂02-Oct-81 0054 JMC
To: pourne at MIT-MC
CC: LLW at SU-AI
pourne[f81,jmc] Note to Jerry about report
Inevitably, much of what the second meeting had to do was to
repeat points raised at the first meeting. It seems to me that there
are two points that should be emphasized in the report.
1. The anti-tank crowbar system has potential to give the
President non-nuclear alternatives in some situation where he
would otherwise have to use the neutron bomb. The extent to which
this should be emphasized probably depends on the degree of optimism
of the DARPA sponsors of the research. Lowell could probably tell
you about that.
In a psychological environment
in which men of vision and good will are looking for justifications
for space activities, it may be an important point.
[John is surely right on this point. Also, as was pointed out in the
pertinent discussion, it's the tactical, not the strategic, defense
pocket which is the truly bottomless one. The DARPA program in this
area is called Assault Breaker, and has a great deal of Congressional
attention as well; most of it focuses on more mundane tricks than
multi-hundred km range spears. Making the case for pre-deploying in
space (vs launching as needed from the ground) is the trick in the
whole tactical business, the more so when in-space survivability is
carefully considered.]
2. I think we should suggest that the President announce that
our astronaut programs will accept a limited number of foreigners
who meet qualifications for trainees. The foreign man in the street,
even in poor Asian countries, was much more impressed and interested
by the Shuttle and Voyager than by any amount of communication satellites
for education and earth resources surveys. The Indian student has
as much imagination as an American, and is just as inclined to be
fascinated by space, even though the sober types think he ought
rather to be interested in stamping out illiteracy.
The possibility of becoming an astronaut, even if he is one of the
999 out of 1000 who never seriously considers applying, can be more
inspiring to a young man of the third world
than the possibility of being a Marxist revolutionary.
The great interest in the Shuttle and Voyager, the spectaculars
that extend the possibilities of humanity as compared to the
narrow material benefits of Landsat, etc. was confirmed to me
by an American sitting next to me on the plane to Europe. He
described the excitement in Singapore and the congratulations
he received as an American at the time of the Shuttle landing.
However, we should admit foreigners as individuals in the
American way rather than as representatives of their governments
in the Soviet way. The President should consider announcing
this at Cancun. Applications would be accepted directly from
individuals rather than through governments.
[Another excellent suggestion, moreover most timely!]
[Lowell]
∂03-Oct-81 1113 CLT SEMINAR IN LOGIC AND FOUNDATIONS OF MATHEMATICS
To: "@LOGIC.DIS[1,CLT]" at SU-AI
FIRST MEETING: Tues. October 6, 4:15-5:30, Room 383-N (3rd. floor, Math.)
SPEAKER: Prof. Helmut Schwichtenberg
University of Munich
TITLE: "Complexity of normalization in the pure typed lambda calculus"
∂04-Oct-81 0141 Feigenbaum at SUMEX-AIM call for LISP-oriented meeting
Date: 4 Oct 1981 0135-PDT
From: Feigenbaum at SUMEX-AIM
Subject: call for LISP-oriented meeting
To: jmc at SU-AI, zm at SU-AI, rpg at SU-AI
cc: engelmore at SUMEX-AIM, rindfleisch at SUMEX-AIM
(Dick, would you redistribute this to Dave Luckham, and the heads of any
LISP-based research projects I may have missed--except Binford who is going
for his own VAX?)
The LISP community (with the possible exception of Dave) seems to be close
to consensus on what it needs in the new equipment proposal. What we need
now is a meeting of heads of LISP-based projects to adjust numbers and
nail down the consensus. How about Tuesday afternoon at 4:15?
The meeting should include Dick Gabriel, who has been helping us get our
act together, Bob Engelmore, who underestands the ARPA picture, and
a few others like Tom Rindfleisch who help us to see the big picture.
Ed F.
-------
∂04-Oct-81 1138 RPG Lisp
I assume you saw the message from EAF about Lisp? I'd like
to talk to you about the situation, possibly tomorrow afternoon?
One main thing thing is that EAF seems to think TOB shouldn't be included.
Well, trivially that's true because he's in europe right now, but EAF
thinks that TOB is getting his own Vax... The Vax is intended for NSF work
(the workbench), and the ARPA work is separate. Druffel wants ACRONYM work
to flow from Stanford to Hughes, probably via the Lisp Machine route, so
TOB probably needs 1 of them. To be honest, TOB doesn't understand the Lisp
issues wrt Vax and may have cut himself off in the bargain.
Any thoughts welcome.
-rpg-
∂04-Oct-81 1410 RPG Lisp
To: EAF at SU-AI, JMC at SU-AI, ZM at SU-AI, DCL at SU-AI
CC: csd.genesereth at SU-SCORE, SL at SU-AI,
engelmore at SUMEX-AIM, rindfleisch at SUMEX-AIM
Ed,
I agree that such a meeting is necessary, and I don't want to
give the impression of torpedoeing any HPP plans. As I said, and as I
continue to say, my interest is in serving the community as it exists now
and as it will develope in the future. I am perfectly happy to buy 15 Dolphins
for HPP as long as everyone else is served well.
There is one comment I'd like to make, mainly because I don't understand
the realities and need some clarification. I don't understand why a TOB
representative should not attend the meeting. I understand that TOB is getting
a Vax, but I understood this to be for NSF contract needs. There are the
ARPA contract needs that have not been adequately addressed, I think. Hand-eye
(eye mainly) is heavily into ACRONYM use, and I understand that Druffel wants
ACRONYM to move to Hughes, and via the vehicle, possibly, of Lisp Machines.
Moreover, since we have all agreed (I thought) that a Vax is a
poor vehicle for Lisp, and since we know TOB's crew are heavy Lisp users,
I would have guessed that the committee would recommend to TOB that he think
more realistically about his groups needs in the light of our deliberations.
However, Tom is in Europe at the moment, so Sid Liebes is acting
as head. Perhaps we should invite him?
Unfortunately, I found out today that my head count was not accurate,
essentially because I miscounted Hand-eye Lisp usage: I failed to count 5
Hand-eye Lisp users. Gio, who does not use department Lisp cycles, has 4 InterLisp
users that I did not include (because they don't use department resources
for them: I talked to Jerry Kaplan who said that they were perfectly happy
with using SRI-AI [a 2060] for Lisp and the Vax would run a PASCAL based
database system.) I will add them to my list on this round too, for fairness.
InterLisp Common Lisp
Hpp 24 12
Non-Hpp 4 29
total 28 41
HPP 36
non-Hpp 33
∂04-Oct-81 1523 RPG
∂04-Oct-81 1513 JMC lisp
I will be available tomorrow afternoon and also Tuesday at 4:15 for
Feigenbaum's meeting. How many LISP users did you count in the
Formal Reasoning Group?
Not all these people are in the Formal Reasoning group (I think),
but they are non-HPP. Not all are ARPA (but neither are the 36 HPP
I counted).
rpg jmc clt lgc jk cg rww rsf jjw moses
Malik
∂04-Oct-81 1815 Feigenbaum at SUMEX-AIM Re: Lisp
Date: 4 Oct 1981 1812-PDT
From: Feigenbaum at SUMEX-AIM
Subject: Re: Lisp
To: RPG at SU-AI, EAF at SU-AI, JMC at SU-AI, ZM at SU-AI, DCL at SU-AI
cc: csd.genesereth at SU-SCORE, SL at SU-AI
In response to the message sent 04 Oct 1981 1410-PDT from RPG at SU-AI
Re TOB, I thought he had essentially gotten ARPA to agree to his own
project's VAX, independent of other community needs.
Ed
p.s. nobody has yet said whether they would attend that meeting on Tuesday.
-------
∂05-Oct-81 1102 RPG Lisp timing
∂05-Oct-81 1016 Engelmore at SUMEX-AIM Lisp timing
Date: 5 Oct 1981 1013-PDT
From: Engelmore at SUMEX-AIM
Subject: Lisp timing
To: RPG at SU-AI
Mail-from: ARPANET host USC-ISI rcvd at 5-Oct-81 1008-PDT
Date: 5 Oct 1981 1005-PDT
Sender: OHLANDER at USC-ISI
Subject: Re: Lisp timing project
From: OHLANDER at USC-ISI
To: Engelmore at SUMEX-AIM
Message-ID: <[USC-ISI] 5-Oct-81 10:05:24.OHLANDER>
In-Reply-To: Your message of 4 Oct 1981 1249-PDT
Bob,
The MRAO covered approx. $266K for McCarthy which appears to cover the
Lisp timing effort. At least, the proposal shows funds for Gabriel and
a research assistant for Lisp timing. In other words we funded everything
that McCarthy asked for. One problem that arises, however, is that the
Arpa Order still has not been signed. It has only been in PM for 2 months.
It should be signed this week and dated for 1 Oct. Will let you know.
Regards, Ron
Dick,
Last msg was supposed to contain the enclosed.
Bob
-------
∂05-Oct-81 1215 ullman@Diablo (SuNet)
Date: 5 Oct 1981 11:59:22-PDT
From: ullman at Diablo
To: equip
Here is the current list of equipment the committee wishes
to place in the proposal to ARPA. I have broken it down
both by year and by the area (if any) to which the items
will be dedicated. SAIL stand for the AI community minus
HPP. The assumption is that the "Dolphins" would at first
be used by the whole AI community, but with the arrival of
the S1, would become dedicated to HPP use. Not shown on the
list are the NA vax from NSF/other and the 75 McSUN termi-
nals we would build from the Stauffer grant.
1982 1983 1984 HPP SAIL SYST GENL
VAX 250 250
S1 650 325 325
"Dolphins" 600 300 900
SUN 140 180 250 150 420
McSUN 100 100
Foonly 225 225
Ethernet 50 50 100
ERL Local Net 30 30 60
Printers 30 60 30 120
Local file stores 50 50 100
SUVAXen 180 90 90
TOTALS 1030 995 1230 990 325 550 1390
GRAND TOTAL = $3,255,000
It is doubtful that we are going to get this much for equip-
ment over 3 years. Therefore, the committee must prioritize
the items on the list. Since it is possible that we shall
have to pay maintenence out of the 3 million we expect, I
suggest that everything over 2.5M must receive a priority.
Therefore, would you each give me a list of the 2.5M worth
of equipment you think should be bought without question,
and a rank ordering of all equipment over that amount (about
.75M worth).
I'm serious about needing these lists from everyone, since I
intend to devote next Friday's meeting to hammering out the
details of the proposal. No fair waffling and waiting to
see everyone else's list. However, you can, if you wish,
send the list to me directly, rather than to equip, and I'll
summarize what I receive.
October 5, 1981
∂05-Oct-81 1347 SIS phone msg
John, Marvin Minsky returned your call. Susan
∂05-Oct-81 1359 SL LISP, Vax, etc.
To: RPG at SU-AI, EAF at SU-AI, JMC at SU-AI
Regarding MAIL copied to me Oct 4 concerning "Lisp", Vax, etc., I
would like to attend meeting. However, I will not be able to represent
details of views and opinions of Binford. He is in weekly contact,
and is scheduled to return Oct. 24. - Sid
∂05-Oct-81 2047 V. Ellen Golden <ELLEN at MIT-MC> MACLisp Manuals
Date: 5 October 1981 23:44-EDT
From: V. Ellen Golden <ELLEN at MIT-MC>
Subject: MACLisp Manuals
To: jmm at SU-AI
cc: jmc at SU-AI, rpg at SU-AI
As we know, the sad Maclisp Manual (i.e. 3 parts, costing $4.25 or some such)
is sad, but available from
Publications
Laboratory for Computer Science
545 Technology Square
Cambridge, MA 02139
Prepayment (or purchase order) is required.
The Lisp Machine Manual (which serves to describe Lisp on the Lisp Machine,
and is quite up to date) is available from the AI Lab (same address) for
10 or 12 dollars (inflation...). Send network mail to PUBLICATIONS@AI
for details on the Lisp Machine Manual (also known as "the CHINE UAL" since
the title wraps around the book, and that is what is on the front).
Please phone them and ask if 50 are available. Say that the number may be
more than 50 and ask how many are available. Verify price and ask if they
could ship airfreight if prepaid.
∂06-Oct-81 0915 PN Accounts
To: RPG at SU-AI
CC: csd.hill at SU-SCORE, JMC at SU-AI, REG at SU-AI
Susan Hill asked me to ask you to come up with percentages as to how
much of your SAIL account should be paid by Ralph, how much by John, and
how much by S-1. It should probably be by roughly how much disk space you
use for each project.
Thanks,
Peter
∂07-Oct-81 0835 FFL
To: JMC, FFL
Petrita DeCima called you again.
Mr. Ito from Japan stopped by to see you. He will call you or stop
again later.
∂07-Oct-81 1240 VRP via Ethernet host 50#301 Luca Cardelli
A while ago Rod Burstall prodded me about the possibility of having
Luca Cardelli, an Italian graduate student finishing up at Edinburgh, spend
a year or so as a post-doc at Stanford. I have a letter of recommendation for
him from Rod which speaks highly of him. Is this something Stanford in general
and one of us in particular should pursue?
∂07-Oct-81 1545 Denny Brown <CSD.DBROWN at SU-SCORE> CS200
Date: 7 Oct 1981 1542-PDT
From: Denny Brown <CSD.DBROWN at SU-SCORE>
Stanford-Phone: (415) 497-2274
Subject: CS200
To: CSD-Faculty: ;
cc: sl at SU-AI, CSD.ENGELMORE at SU-SCORE, CSD.HBROWN at SU-SCORE
I have had very little response concerning CS200. This is your best forum
for introducing yourselves and your projects to new students. I have
Vaughn Pratt scheduled for tomorrow, and Terry Winograd has volunteered for
a November date. I would like to announce next week's speaker at the
meeting tomorrow. Who wants it? -Denny
-------
∂07-Oct-81 1955 Kanerva at SUMEX-AIM First meeting of the Context Advisory Council
Date: 7 Oct 1981 1954-PDT
From: Kanerva at SUMEX-AIM
Subject: First meeting of the Context Advisory Council
To: McCarthy at SU-AI
The first meeting will be this Friday at 9 a.m. in Cedar Hall.
You should have received the official invitation some time ago. Even
if such councils can be but seremonial, I sincerely hope that you are
interested in what we are doing, can attend, and will advise.
- Pentti Kanerva
-------
∂08-Oct-81 0920 SIS Colloquium Notice of October 12 - 16, 1981
To: "@COLLOQ.DIS[INF,CSD]" at SU-AI
Date Place Person
Day Event From
Time Title
-------------------------------------------------------------------------------
10/12/81 Math380C Helen Yee
Monday Numerical Analysis NASA-Ames Research Center
4:15p.m. Seminar ``Stability Analysis of Implicit Difference
Approximations of Hyperbolic Initial Boundary
Value Problems With Applications To Fluid
Dynamics''.
10/13/81 Jordan 041 Clark Thompson
Tuesday Computer Science UC Berkeley
4:15 p.m. Colloquium ``VLSI Modelling and The Sorting Problem''.
10/14/81 Terman 153 Ellen V.B. Lapham, Robin J. Jijour, and
Wednesday Computer Systems Charles Kellner, of Syntauri Corp, Palo Alto
4:15 p.m. Laboratory Seminar ``The alphaSyntauri Synthesizer, A New Concept
in A Musical Instrument Product''.
10/16/81 MJ301 C. Mohan
Friday Database Seminar University of Texas at Austin
3:15 p.m. ``Strategies for Enhancing Concurrency and
Managing Deadlocks''.
∂08-Oct-81 1232 RPG@Sail (SuNet) Lists
Date: 08 Oct 1981 1230-PDT
From: Dick Gabriel <RPG at SU-AI>
Subject: Lists
To: equip at DIABLO
Here are my lists, which are the paths for getting under $3m and
for getting to a guaranteed ARPA approval. I have adjusted the benefits
to each group a bit, since the buy of lisp stations will be more widely
available, I think, do to a less coincidentally group specific mix
of actual equipment bought (i.e. probably more Lisp Machines and less
Dolphins). The second list, though, has the advantage of purchasing the
file server this year, however, I think we can start getting the file
server right away and pay for most of it in fiscal 1983.
Path of least flush:
1982 1983 1984 HPP SAIL SYST GENL
VAX 250 250
S1 650 325 325
"Dolphins" 600 300 650 250
SUN 70 90 125 75 210
McSUN 100 100
Foonly 225 225
Ethernet 50 50 100
ERL Local Net 30 30 60
Printers 30 60 30 120
Local file stores 50 50 100
TOTALS 780 905 1105 650 575 385 1180
GRAND TOTAL = $2,790,000
Path of most flush:
1982 1983 1984 HPP SAIL SYST GENL
VAX 250 250
S1 650 325 325
"Dolphins" 350 300 450 200
SUN 70 70 70 105 105
McSUN 100 100
Foonly 225 225
Ethernet 50 50 100
ERL Local Net 30 30 60
Printers 30 60 30 120
Local file stores 50 50 100
TOTALS 755 910 730 450 525 415 1075
GRAND TOTAL = $2,395,000
∂09-Oct-81 0223 pratt@Diablo (SuNet) Proposed budget changes
Date: 9 Oct 1981 02:16:56-PDT
From: pratt at Diablo
To: equip
Subject: Proposed budget changes
I propose four changes to Jeff's most recent budget (Oct 5). I think these
changes are as much as we should concede to ARPA in budget reductions on a
first pass. It is never good practice to ask for too little; despite any
real or imagined threats from ARPA to take matters into their own hands if we
don't do our own budget trimming, I think we should aim to ask for a little
more than they'll give us.
The overall effect on the budget of my four changes shifts $140K from year 1
to year 2, and eliminates an additional $180K from year 1.
Change 1: Exchange year-1/year-2 Dolphin expenditures
I doubt the wisdom of purchasing a large number of "Dolphins" in the first
year. Even for machines where the department had accumulated several years
worth of experience and found them entirely satisfactory, I would be hesitant
to see $600K of year 1's limited budget spent on this one category of
machine. For machines that are by department standards complete mysteries it
seems inappropriate to spend two-thirds of the year's budget on a pig in a
poke. For example suppose these machines turn out during the first year of use
to be unusable due to inadequate performance? Will we be able to get our
money back from the manufacturer, or have we just lost $600K down the drain?
I would urge caution in purchasing these machines until a better picture
emerges as to the role they can play in meeting departmental needs. An
appropriate budget reflecting such caution would allocate $300K for "Dolphins"
in year 1, increasing to $600K in year 2.
Change 2: Balance Dolphin/Sun expenditure in year 1
By the same token I would urge caution in purchasing Suns, which have also not
yet been tested thoroughly in the department, though less so than Dolphins
since the Sun terminal in my office has been used pretty heavily by myself
and students during the past three months, certainly more than the Dolphins
presently in HPP. Since the Suns meet at least as urgent a need as Dolphins,
namely the terminal function, editing, graphics, and non-Lisp computing
(Pascal and C programs) it would seem appropriate to spend at least as much
on Suns. I remain unconvinced that the departmental need for Lisp cycles
is as great as the combined need for terminals, editing, graphics, and
Pascal/C programming; however as one who would rather program in Lisp than
either Pascal or C (for reasons spelled out in my various papers on Lisp) I am
sufficiently eager to be able to take advantage of Lisp machines that I will
not dwell on this question, and simply propose that the expenditure on Suns
in year 1 equal that for "Dolphins."
Year 1 of my proposal then reads like Jeff's, but with "Dolphins" cut back to
$300K, and Suns increased to $300K. Year 1's budget is thereby reduced by
$140K.
Change 3: More generic placeholders for year 2
For years 2 and 3 we have been using machines to hold places, i.e. we really
are budgeting for generic machines rather than particular models. I'm
disinclined to distinguish between Lisp and non-Lisp personal computers that
far in the future until the Lisp supply picture for the various personal
computers firms up. For example, if Stars arrive with Interlisp in year 2 then
Dolphins are out. The same goes for Suns, which are very likely to have
a good Common Lisp by then. Thus rather than budget separately for
Lisp and non-Lisp machines, it would make more sense to budget generically for
personal computers. Whether these firm up into on-the-desk or per-session
machines can be left undetermined for now.
I suggest that the budget for personal computers in year 2 be $620K, namely
Jeff's original figure (sum of "Dolphins" and Suns) plus the $140K
transferred from year 1. If Suns turn out to be a disaster, this amount can
be spent entirely on "Dolphins," which is more than Jeff's budget allocated
for them. In any event, I think we should expect to see $620K being spent on
machines all of which can support Lisp; this is $20K more for personal
machines with Lisp capability than Jeff's proposal called for. If the Lisp
community would like to see even more than this spent in year 2, I for one
would support this item as the first place to expand the budget if the
committee decides to be a little bolder with ARPA.
Change 4: Prioritize and infer a cut: SU-Vax's
Jeff asked for priorities. Of all the items on the list, the least urgent
is the set of SU-Vax's. The reason everyone is so eager to pay
$180K for them is simple human nature: how could anyone pass up such a
bargain? Just like the item you bought in Filene's bargain basement, when you
get it home you wonder why you bought it. I have yet to hear a plausible
application for SU-Vaxes that is consistent with DEC's terms. Even if a
plausible reason did emerge, I don't see how it could move the SU-Vax's up from
the bottom of the priority list. Removing this dubious item gets the first
year's bottom line down to an easily digested $710K.
I believe that no further adjustments to the rest of the budget, including
years 2 and 3, should be needed after making this easy cut. In fact I am
inclined to think that now the budget is just a trifle too lean.
Conclusion:
Here is the effect of these four changes on Jeff's budget. I object to the
territorialism implicit in the HPP/SYS/SAIL/GEN columns and have therefore
omitted them.
1982 1983 1984
VAX 250
S1 650
"Dolphins" 300
Suns 300
Personal Computers 620 250 (this line not in Jeff's budget)
McSUN 100
Foonly 225
Ethernet 50 50
ERL Local Net 30 30
Printers 30 60 30
Local file stores 50 50
TOTALS 710 1135 1230
GRAND TOTAL = $3,075,000
∂09-Oct-81 1329 FFL
To: JMC, FFL
Dan Friedman, U. of Indiana called. He would like you to call him
this afternoon, please. Try 812 339 l760 first, or 812 337 4885.
∂09-Oct-81 1641 CLT SEMINAR IN LOGIC AND FOUNDATIONS OF MATHEMATICS
To: "@LOGIC.DIS[1,CLT]" at SU-AI
SPEAKER: Prof. Lou vna den Dries, Stanford
TITLE: "Exponentiation"
TIME: Tues. October 6, 4:15-5:30
PLACE: Room 383-N, 3rd. floor Math. Bldg.
∂10-Oct-81 0313 Arthur Keller <CSD.KELLER at SU-SCORE> SAIL Bucks
Date: 10 Oct 1981 0313-PDT
From: Arthur Keller <CSD.KELLER at SU-SCORE>
Reply-to: ARK@SU-AI
Subject: SAIL Bucks
To: REG at SU-AI
cc: CSD.DBrown at SU-SCORE, JMC at SU-AI
The S-1 project decided to pull out and reduce everyone to guests.
Since they are going to stop paying essentially, I don't think that they
should be allowed to maintain a large list of guests.
The following are the list of S1 guests listed as such in USERS.DAT:
OTA, JMB, WRB, HWC, SJC, JRD, CFE, CWF, CBF, CEG, PLH, DH, RLH, RAH, AJH,
RTJ, JPK, PBK, RK, JLM, LRM, JCM, CJN, WCN, RKS, JGP, RFP, CAR, LWR, JBR,
WHR, PMS, MSS, JRS, RLS, LAN, GLS, TJS, JT, TAW, RKW, DLW, CDW, LCW, LLW.
This is 45 users and amounts to 22.5 aliquots! If they want to push, I
think we can push back. We don't have to have any guests if we don't want
to.
Arthur
-------
I think the contacts with the S-1 people are valuable for Stanford,
and we should continue the S-1 users assuming they pay for their
level of use. The implication in ARK's message that the S-1 people
have somehow behaved discourteously in deciding to set up their
own WAITS rather than pay our prices is incorrect. Whether they
made an optimal decision or not is their affair. What it tells us
is that usage is more price sensitive than we might have assumed
and that CSD-CF must run a more economical operation. Not everyone
requires a large staff to run a time-sharing system, and those brave
enough to use Foonlys instead of DEC-20s seem to be rewarded by
fortune. Context should also take warning from this.
∂10-Oct-81 1601 ARK S-1, SAIL, and JMC's comments
To: JMC at SU-AI, REG at SU-AI, CSD.DBrown at SU-SCORE
ARK - I see nothing wrong with the S-1 people *paying* for their level of
use. But perhaps we should either require them to up their allocation to
1 aliquot per account, or flush some accounts. Having set up their own
system does not enter into my opinion. My point is that most of their
usage is at the guest level, and that seems unfair. At 25 blocks each,
they are *not* paying for their level of service.
∂10-Oct-81 1855 Kanerva at SUMEX-AIM Thank you
Date: 10 Oct 1981 1245-PDT
From: Kanerva at SUMEX-AIM
Subject: Thank you
To: McCarthy at SU-AI
Thank you for coming to the meeting Friday morning. - Pentti
-------
∂10-Oct-81 2312 CLT Corrected announcement
To: "@LOGIC.DIS[1,CLT]" at SU-AI
SEMINAR IN LOGIC AND FOUNDATIONS OF MATHEMATICS
SPEAKER: Prof. Lou van den Dries, Stanford
TITLE: "Exponentiation"
TIME: Tues. October 13, 4:15-5:30
~~~~~~~~~~~~~~~~
PLACE: Room 383-N, 3rd. floor Math. Bldg.
∂11-Oct-81 0017 Lowell Wood <LLW at S1-A> Welcome!
Date: 11 Oct 1981 0015-PDT
From: Lowell Wood <LLW at S1-A>
Subject: Welcome!
To: JMC at S1-A, jmc at SU-AI
CC: LLW at S1-A, JBR at S1-A, TM at S1-A, PMF at S1-A, JMB at S1-A
Welcome to S1-A, John, the WAITS System from which (system-wide) purges
have been forever banished! Jeff will establish you as a formal user
when he sees this message, but you can start moving your manuscripts over
here for safekeeping (at least until you can get the SAIL disc storage
charges back down to rational levels). Distinguished defectors such as
yourself are always granted political asylum in S1-Land (and, because
of the Project's great debt to you, we won't even let the news get about
that fell circumstance have driven John McCarthy off the computer
system of his own AI Lab)! Lowell
∂11-Oct-81 0033 ME on TTY67 (at TV-120) 0033
You get it automatically, but there are a few things you have to type
specially, like ESC, BREAK, CLEAR and CALL. The first 3 of these you get
by typing ESC I followed by, respectively, α, β, and ε. CALL is αZ. αZ
you get by typing ESC I αZ. ESC I ? will give some help.
∂11-Oct-81 0038 ME on TTY67 (at TV-120) 0038
Right. ← and _ get interchanged. Use _ to get ← and vice versa.
Also, if you say TTY SAIL at S1, then I think _ will appear correctly
instead of as x (for ↑X). This interchange is because MRC thinks SD
is only used to talk to ITS.
∂11-Oct-81 1801 LGC Lunch Tomorrow
Shall we meet at your office at 11:00 or 11:30 tomorrow morning
for lunch with Castaneda?
By the way, you once asked me a question about "logics of action"
or the "logic of ought" or something like that. Castaneda is one
of the main philosophical proponents of such a logic. I think there
are some sections (or chapters) of his book devoted to that topic.
∂11-Oct-81 1917 Grosz at SRI-AI Re: Program Committee Chairman
Date: 11 Oct 1981 1904-PDT
From: Grosz at SRI-AI
Subject: Re: Program Committee Chairman
To: MINSKY at MIT-AI, tenenbaum at SRI-KL, newell at CMU-10A,
To: reddy at CMU-10A, webber at BBND, Aaai-Office at SUMEX-AIM,
To: bledsoe at UTEXAS-20, nilsson at SRI-AI, walker at SRI-AI,
To: grosz at SRI-AI
cc: bobrow at PARC, jmc at SU-AI, simon at CMUA,
cc: sussman at MIT-AI, dwaltz at BBND, woods at BBND
Marvin --
This is an attempt to respond to all of your messages at once.
My recollection is that the list we constructed was of candidates for
program chair, not possible new AAAI exec council members (which I
don't think we need to worry about right now). It certainly would be
useful to pass the list on to whoever becomes program chair since I'm
sure everyone on the list would make a fine committee member as well.
I don't have a copy of that list, but think that Bonnie does.
I would like to ask you to look over the list and consider some
changes in the list of choice for program chair. I'll give you some
specific reactions to your list, but first need to give my overall
view of what's required.
AAAI is still in its formative stages which I think creates two
special needs that we should keep in mind when choosing the program
chair and new members of the program committee. First, we want to do
our best to insure that the committee represents as wide a spectrum of
AI interests as possible, not only of particular subject areas, but
also of approaches to AI. It seems especially important that the chair
be open-minded about this (a point I think Nils raised at that very
late night meeting in Vancouver). Second, it will take even more time
and effort on the part of the chair than would normally be the case
(and the plain vanilla program chair job involves plenty of work)
because the procedures to be followed still are in a state of flux.
So, we need someone willing to put in a good deal of time on this.
The latter point raises an issue on which Bob Balzer and I disagree.
His recollection seems to be that the choice of papers last year
('80) was fine and that the procedures were acceptable to most people.
My memory is that there was a long and heated discussion at the open
AAAI meeting about how the program committee was chosen, how they chose
the papers, etc. This discussion included several good suggestions
for modifications, suggestions Tenenbaum and Balzer assured folks would
be taken into consideration for the next AAAI program. I am willing to
be corrected on this score if other people remember differently (I was
not able to attend the open AAAI meeting at Vancouver -- maybe something
else was decided there and I just need to be brought up to date), but if
I am correct, I think we need to make sure we've got a program chair who
will take the whole job of the program committee into serious consideration.
Finally, it is my recollection that the bylaws give the AAAI executive
council responsibility for choosing the new members of the program
committee, but allow for the possibility of passing that on to the
program chair. I would like to see us (the exec committee) involved in
the choice of new members at least this time around. The main reason
is to insure a fair coverage of the field on various levels. This is
something I think N minds can do a whole lot better than one (no
matter how fair minded that one is). Again, AAAI is in its formative
stages. I have confidence that if we get a well-rounded committee set
up, it will continue itself as well-rounded (and equal confidence that
if we set up a committee biased in some direction, the committee will
continue in that vein).
As for particulars. I have no quarrel with Dave Waltz but am
pessimistic about his saying "yes" because of the time involved. My
interactions with Goldstein and Lenat leave me uneasy about how broad
a spectrum of approaches would be tolerated, but I'd like to hear from
others who know them better than I. Unless Hewitt has changed a lot, I
find it doubtful that he'd put in the time required to reorganize. I
don't know Luckham well enough to say anything about him.
On the issue of industrial support, it's not clear to me what
$25K buys that $5K doesn't other than ads in AI Magazine -- are they
really worth $5K an issue? If not, are there some tax incentives or
are we going to give the sponsors free admission to tutorials or some
other goodies?
On the issue of Lou Robinson: did we get a list of tasks he was going
to perform (I notice he is reminding us of what we're supposed to be
doing for the next meeting)? what's the state of our finances now?
Barbara
p.s. I've cc'd all the people listed in the latest AAAI as on
the exec council for whom I could construct net addresses. McDermott
and Rieger are the two left out.
-------
∂12-Oct-81 0759 Nilsson at SRI-AI (Response to message)
Date: 12 Oct 1981 0800-PDT
From: Nilsson at SRI-AI
Subject: (Response to message)
To: JMC at SU-AI
cc: NILSSON
Well, I won't be interested in Chinese Restaurants for a while. How
about meeting later this week. I'm pretty filled up thru Thurs, but
anytime Fri. is fine. Should ∪Stan attend also? --Nils
-------
Any time Friday afternoon would be fine. Let me suggest 1:30 for
definiteness. It would be good if Stan were there too.
∂12-Oct-81 0946 FFL CALL FROM DR. RUSSELL 415 654 7458
To: JMC, FFL
He wishes to confirm the date of January l5 for the conference on Computer
Models of the Mind. If the date is not acceptable, please let him know at
once as he is proceding on that assumptions in making plans.
Panelists will be Joseph Weizenbaum, Hubert Dreyfuss, and Adele Goldberg.
The honorarium will be $250. It will last from 4 p.m. through the evening.
In two weeks will send you questions and theme.
∂12-Oct-81 1011 JJW
∂12-Oct-81 0142 JMC
I fixed spelling of Weyhrauch but had to unprotect eklman.tex.
----------
Thanks. I ran TEX on it again so now EKLMAN.PRE has the
corrected spelling.
∂12-Oct-81 1046 LGC
Ok, see you there, then. -- Lew
∂12-Oct-81 1049 JMC seminar
To: "@PROVE.LIS[F81,JMC]"
Starting Tuesday Oct 12 at 3pm, there will be a seminar on
proving properties of programs in EKL and/or FOL. We need to
develop our technique, extend the class of properties that
can be proved, recommend improvements in the provers and
develop an environment of axioms for CS206 and CS258.
∂12-Oct-81 1111 CLT
(1) Tuesday Oct 12 = Tuesday Oct 13
(2) SIMPLIFY (2)
(3) FALSE
(4) The proving problem has been solved
∂12-Oct-81 1112 REG
Before I go off and make a revision to the budget, I'd like some clear
guidelines about what you (and Ed, and the users) think the budget ought
to pay for.
I can distinguish between ``operation'' and ``development.'' But do you
want all development eliminated? Elimination of all development means
that we won't replace the data disc with RAM; we won't bother to finish
the MEIS; we won't work on McSun; we won't extend the Ethernet
anywhere. On the other hand, if you want and permit development of
such things as want and which we can't buy outside, then it seems that
nearly all of Sun qualifies.
The only thing that I'm doing that I'll admit the general CSD users aren't
interested in is the multiple micro-processor system. We can eliminate
that, and save $30K to $40 K.
∂12-Oct-81 1326 FFL
To: JMC, FFL
Please call Dan Friedman at 812 339 9333.
∂12-Oct-81 1726 David R. Cheriton <CSD.CHERITON at SU-SCORE> Winter quarter dept. seminars
Date: 12 Oct 1981 1721-PDT
From: David R. Cheriton <CSD.CHERITON at SU-SCORE>
Subject: Winter quarter dept. seminars
To: CSD-Faculty: ;
I have been placed in charge of arranging dept. seminars
for the winter quarter. Please send suggestions of people
you think would be of interest. To save money, non-local
people are more attractive if we can share travel expenses
with other places.
-------
∂12-Oct-81 2229 Don Walker <WALKER at SRI-AI> [Marvin Minsky <MINSKY at MIT-AI>: Program Committee Chairman]
Date: 12 Oct 1981 2218-PDT
From: Don Walker <WALKER at SRI-AI>
Subject: [Marvin Minsky <MINSKY at MIT-AI>: Program Committee Chairman]
To: bobrow at PARC-MAXC, buchanan at SUMEX-AIM, erman at USC-ISIF, feldman at SUMEX-AIM,
jmc at SU-AI, simon at CMU-10A, gjs at MIT-AI, dwaltz at BBND, woods at BBND
This is the message from Marvin that prompted Barbara's reply
---------------
Mail-from: ARPAnet host MIT-AI rcvd at 10-Oct-81 2103-PDT
Date: 10 October 1981 23:06-EDT
From: Marvin Minsky <MINSKY at MIT-AI>
Subject: Program Committee Chairman
To: MINSKY at MIT-AI, tenenbaum at SRI-KL, newell at CMU-10A,
reddy at CMU-10A, webber at BBND, Aaai-Office at SUMEX-AIM,
bledsoe at UTEXAS-20, nilsson at SRI-AI, walker at SRI-AI,
grosz at SRI-AI
cc: balzer at USC-ISIE
Here is ordered list of Program Committee Chairman candidates.
Please comment, add, blackball, etc. soon, or I shall start
asking them in that order.
1. WALTZ
2. GOLDSTEIN
3. LENAT
4. HEWITT
5. LUCKHAM
WITH BLEDSOE AS A HIDDEN RESERVE IF ALL ELSE FAILS.
Actually, this should go to Executive Committee, but I seem to have
no list of that committee.
DAVID BARSTOW has been suggested for Tutorial Chairman, and
I think thatc would be fine. Any other suggestions?
--m Marvin Minsky
---------------
-------
∂12-Oct-81 2242 Don Walker <WALKER at SRI-AI> [Marvin Minsky <MINSKY at MIT-AI>:]
Date: 12 Oct 1981 2233-PDT
From: Don Walker <WALKER at SRI-AI>
Subject: [Marvin Minsky <MINSKY at MIT-AI>:]
To: bobrow at PARC-MAXC, buchanan at SUMEX-AIM, erman at USC-ISIF, feldman at SUMEX-AIM,
jmc at SU-AI, simon at CMU-10A, gjs at MIT-AI, dwaltz at BBND, woods at BBND,
webber at BBND
Mail-from: ARPAnet host MIT-AI rcvd at 11-Oct-81 1047-PDT
Date: 11 October 1981 13:41-EDT
From: Marvin Minsky <MINSKY at MIT-AI>
To: MINSKY at MIT-AI, TENENBAUM at SRI-KL, Allen.Newell at CMU-10A,
Raj.Reddy at CMU-10A, FEIGENBAUM at SUMEX-AIM,
AAAI-OFFICE at SUMEX-AIM, BLEDSOE at UTEXAS-20, Walker at SRI-AI,
NILSSON at SRI-AI, Grosz at SRI-AI
Here is a draft of corporate beg letter. Please send me
1. Comments on strategy and tactics of letter.
2. Suggestions for corporate target individuals. I should think
that we can surely aim the letter at some high-level personal contact
in each firm!
* * * * *
Dear ----- (to be personalized)
This is a call for help for the field of research and development
called Artificial Intelligence, which plays an important role in many
aspects of the Computer industry including
Computer languages
System programming techniques
Automatic Programming
Advanced Programming
Image Understanding
Robotics
New computer architectures
Natural language systems
Knowledge-based "expert" systems
Data management systems
And this field makes central discoveries for our basic scientific
understanding of computation.
In 1979, many of the most prominent computer scientists in the U.S.
formed a scientific and professional society to serve this exciting
field, named the American Association for Artificial Intelligence.
This, the "AAAI", serves as the principal communications link between
industry and this research community. It does this three ways:
CONFERENCES: the annual AAAI meeting is the central event in the field.
On alternate years it is combined with the International Joint
Conference on Artificial Intelligence.
TUTORIALS: The tutorial program sponsored by the AAAI provide timely,
concentrated looks at important current research developments.
PUBLICATION: distributing a publication called the AI Magazine.
I would like to see the AAAI continue and extend these services. But
I do not see how to do this, given only the membership dues of the
professional members of the AAAI. We need additional support from the
industrial sectors that benefit from these services.
Therefore, I would like to offer you the opportunity to link your
company directly with this dynamic and rapidly expanding field, by
supporting the AAAI. This will help the continued development of
those areas mentioned above, that could be key factors in your own
company's growth and future expansion. We are soliciting
contributions in one of the following categories:
Industrial Sponsor ---- $25,000 annually.
Industrial Affiliate --- $5,000 annually.
Industrial Participant - $1,000.
We will formally acknowledge all corporate support at our conferences.
We will acknowledge the support of Industrial Sponsors in the masthead
of the AAAI publication, the "AI Magazine," as well as through a free
full page ad in that publication for each sponsor. We will also list
the names of industrial sponsors and affiliates in other Association
materials such as on our press release stationery letterhead and our
conference brochures -- which receive wide national and international
distribution.
Please join me in support of a unique and important organization as it
takes its first steps. Please take a minute of your time to fill out
and mail the enclosed card indicating the amount of your contribution
and the pledge category.
Cordially,
Marvin Minsky
ANY SUGGESTIONS ABOUT:
signature
the "participant" category?
etc?
Finally, it may be overkill, but many companies will have already
profitable activities that have AI origins, but they don't realize it.
For instance, Texas Instruments uses in its production facilities
their own variant of Scheinman's "VICARM" robot. This design
originated in a contract that the MIT AI lab. gave to Victor, many
years ago, to make a small robot arm for AI research. In turn, that
was suggested by his earlier work at Stanford AI. Similarly, Kurzweil
was an MIT Ai student, though I claim no direct connection with his
Reading Machine ideas, now a XEROX-owned company. The point is, if I
can have a collection of such corporate successful applications, I can
add each to the corresponding letter, and the "beg" will make its
point felt where it counts!!!
-- marvin
---------------
-------
∂12-Oct-81 2344 JK EKL transportation
To: JMC, JMM
seems to have succeeded : I executed the entire proof of
Ramsey's theorem at SCORE. I was not able to save a core image
since I don't have enough space there yet.
∂13-Oct-81 0029 energy at MIT-MC
Date: 13 October 1981 03:25-EDT
From: energy at MIT-MC
Sender: OAF1 at MIT-MC
To: energy
RMS@MIT-AI 10/13/81 00:00:15
To: ENERGY at MIT-AI
I recall seeing an article a year ago about a "slow breeder" reactor,
supposedly operational in the US, that had various desirable qualities.
Does any one know more about this?
∂13-Oct-81 0123 Jeff Rubin <JBR at S1-A>
Date: 13 Oct 1981 0123-PDT
From: Jeff Rubin <JBR at S1-A>
To: LLW at S1-A, jmc at SU-AI
∂13-Oct-81 0119 LLW S1-A FTPery
To: JMC
CC: LLW, JBR
∂11-Oct-81 0203 JMC via SU-AI ftpable files
The updated copy of SHACKL is now on this machine. FTPing it requires my
password. Is there or should there be a way of making a file FTPable
without a password? Of course, it is no real trouble to keep the master
copy on SAIL from which files are FTPable without password unless they are
otherwise protected.
[John: JBR handles all password matters, with the general guidance from me
that we don`t want every Net hacker in existence playing with our file
system. He'll respond to all suggestions gracefully, I expect. Lowell]
[Gracefully speaking, there is no way to allow FTP without login that does
not also force everyone to specifically protect all sensitive files. --jeff]
∂13-Oct-81 0718 csl.jlh at SU-SCORE (John Hennessy) Computer facilities meeting
Date: 13 Oct 1981 0715-PDT
From: csl.jlh at SU-SCORE (John Hennessy)
Subject: Computer facilities meeting
To: CSD.FEIGENBAUM at SU-SCORE, jmc at SU-AI
cc: baskett at PARC-MAXC, csd.ullman at SU-SCORE
As per suggestions by Ed and John, I am trying to organize a smaller
meeting to try to organize the equipment proposal to Arpa.
I have agreed to act as co-PI and have asked Forest to attend as a senior
systems rep. and Jeff to attend as department rep.
We could meet at the same time the committee was meeting (10AM Friday);
I am also free Thursday mornings and anytime Wednesday.
John
-------
∂13-Oct-81 0806 Nilsson at SRI-AI (Response to message)
Date: 13 Oct 1981 0806-PDT
From: Nilsson at SRI-AI
Subject: (Response to message)
To: JMC at SU-AI
cc: NILSSON
ok, 1:30 Friday it is. -Nils
-------
∂13-Oct-81 0831 Feigenbaum at SUMEX-AIM Re: Computer facilities meeting
Date: 13 Oct 1981 0826-PDT
From: Feigenbaum at SUMEX-AIM
Subject: Re: Computer facilities meeting
To: csl.jlh at SU-SCORE, CSD.FEIGENBAUM at SU-SCORE, jmc at SU-AI
cc: baskett at PARC-MAXC, csd.ullman at SU-SCORE
In response to the message sent 13 Oct 1981 0715-PDT from csl.jlh at
SU-SCORE (John Hennessy)
Dear John,
You are taking what could have been a simple solution and
complicating it again. I have spoken at length with Jeff Ullman about
the process of arriving at an ARPA proposal and have strong feelings
about it.
The suggestion made by John McCarthy was that you, he, and I get
together to frame and write the proposal. Specifically, Jeff was not
included as a departmental representative. The thought was that three
co-PIs, representing the major ARPA project interests, Systems, SAIL,
and HPP, get together and decide what will be asked of ARPA.
I object to the inclusion of Forest in the group. All of us have met
for many many hours with the bigger group to try to frame something
reasonable. Forest has not once (unless I am mistaken) seen fit to
spend his time in this way. Therefore, as a matter of principle, I
would regard his presence at our small meeting as unwarranted.
(Symmetrically, I would feel the same about my presence if I had sent
Doug Lenat to represent the HPP all these many weeks, and suddenly
injected myself into the process at this stage.)
So, I would like to get back to John McCarthy's original suggestion
that three of us meet to do this. I am leaving for Japan on Friday
for 17 days, so a quick meeting Wednesday or Thursday is useful.
Wednesday afternoon is the only really substantial block of
continuous time that I have. I can meet on Thursday morning before
10am (start anytime); or can cancel my lunch appt. on Thursday to
make a block of time available between noon and 2:30 on Thursday.
I await McCarthy's response re selecting a time.
Ed
-------
∂13-Oct-81 1106 Irmgild Schack <CSD.IRMGILD at SU-SCORE> Faculty Lunch
Date: 13 Oct 1981 1040-PDT
From: Irmgild Schack <CSD.IRMGILD at SU-SCORE>
Subject: Faculty Lunch
To: CSD-Faculty: ;
This is just a reminder that there will be a faculty luncheon
at noon today in conference room 146.
Irmgild
-------
∂13-Oct-81 1112 Baskett at PARC-MAXC Re: Computer facilities meeting
Date: 13 Oct 1981 11:01 PDT
From: Baskett at PARC-MAXC
Subject: Re: Computer facilities meeting
In-reply-to: csl.jlh's message of 13 Oct 1981 0715-PDT
To: csl.jlh at SU-SCORE (John Hennessy)
cc: CSD.FEIGENBAUM at SU-SCORE, jmc at SU-AI, csd.ullman at SU-SCORE
Friday at 10 is good for me.
∂13-Oct-81 1116 FFL
To: JMC, FFL
Don Perkel of Biological Sciences called. 7-1663. Dean Wessels asked
him to check with you about distribution of software.
∂13-Oct-81 1129 RWW reference
can i have a reference to
1) your minimalization scheme and
2) abstract syntax
thanks Richard
also the 91 function
thanks again
Richard
∂13-Oct-81 2050 RAH via S1-GATEWAY visit
i'll be around wednesday. Most of the time Mike & I will be up on the
hill stringing joists, but I'ld like to talk about it. How about late
afternoon- say 4, or in the evening.
4 will be fine or dinner and beyond
∂13-Oct-81 2359 RAH via S1-GATEWAY visit2
ok, see you around 4.
∂14-Oct-81 0205 Jerry E. Pournelle <POURNE at MIT-MC> shackleton
Date: 14 October 1981 05:02-EDT
From: Jerry E. Pournelle <POURNE at MIT-MC>
Subject: shackleton
To: jmc at SU-AI, llw at SU-AI
Having read the Shackleton proposal, I like it. can I help?
Suggestion: this is thew kind of thing we could get a LOT of
science fictin writers--all of them probably--behind.
The L-5 Society already exists to do some of this; it ccould
serve as a seed money collector although the project des
probably need to have its own existence.
I expect the big TV shows would be interested.
OMNI Foundation would probably comew up with a few tens of
thousands as seed money. Plus publicity. Including on the OMNI
tv program with Nero oops Peter Ustinov.
∂14-Oct-81 0937 LOUNGO at RUTGERS Rutgers Computer Science Technical Reports
Date: 14 Oct 1981 1223-EDT
From: LOUNGO at RUTGERS
Subject: Rutgers Computer Science Technical Reports
To: erman at USC-ISIB, shortliffe at SUMEX-AIM, dreifus at RUTGERS,
bennett at SU-SCORE, mittal at RUTGERS, chandrasekaran at RUTGERS,
jsmith at RUTGERS, deolankar at RUTGERS, wilkins at SRI-KL, bruce at BBNA,
webber at BBND, friedland at SUMEX-AIM, plondon at USC-ISIB, erm at MIT-AI,
rdg at SU-AI, pressburger at SCI-ICS, csd.gardner at SU-SCORE, fagan at SUMEX-AIM,
fikes at PARC-MAXC, jmc at SU-AI, clancey at SUMEX-AIM, krd at MIT-AI,
hamilton.es at PARC-MAXC, cs.amsler at UTEXAS-20, chinguyen.es at PARC-MAXC,
lisa at UTEXAS-11, kwh at MIT-AI, utgoff at RUTGERS, turock at RUTGERS,
ecg.rich at DEC-MARLBORO
cc: loungo at RUTGERS, dolese at RUTGERS
The newest technical reports are available for access via FTP with
user account <anonymous> with any password. The file names are:
<library>publ-order-form.doc - This contains the
list of our newest technical reports.
<library>online-tecrpts.doc - This contains an
abstract of each of the reports.
If you wish to order copies of any of these reports please send mail
via the ARPANET to LOUNGO@RUTGERS. Thank you!!
-------
∂14-Oct-81 1506 John McCarthy <JMC at S1-A>
Date: 14 Oct 1981 1503-PDT
From: John McCarthy <JMC at S1-A>
To: jmc at SU-AI
sri friday 1:30,sept tax
∂14-Oct-81 1702 Denny Brown <CSD.DBROWN at SU-SCORE> This week's CS200 prod
Date: 14 Oct 1981 1656-PDT
From: Denny Brown <CSD.DBROWN at SU-SCORE>
Stanford-Phone: (415) 497-2274
Subject: This week's CS200 prod
To: CSD-Faculty: ;
Here are some stats about CS200 this quarter.
Dates filled: 7
Dates open: 3 (10/29, 11/5, 12/10)
Full Profs scheduled: 1 (Mike Flynn)
Systems talks schedld: 4 (Flynn, Pratt, Lantz, Hennessy)
AI talks scheduled: 1 (Winograd)
Theory talks scheduled: 0
NA talks scheduled: 1 (Oliger)
If you don't like the (lack of) coverage of your favorite topic,
you'd better volunteer soon. Only three dates left...
-Denny
-------
I will talk on 11/5 in CS200 if it's still open.
∂15-Oct-81 1528 JK ekl
define and ∀e have been fixed.
∂15-Oct-81 2353 pratt@Diablo (SuNet) Dolphin timings
Date: 15 Oct 1981 23:35:40-PDT
From: pratt at Diablo
To: equip
Subject: Dolphin timings
I have benchmarked two Lisp programs on the Stanford Dolphins. The result was
an eye-opener for me, but came as no surprise to HPP-er Gordon Novak, who
has ported a large natural language system, ISAAC, from the U. of Texas
2060 Interlisp to Dolphin Interlisp. Gordon cited a ratio of about 28
between these two machines for all phases of Interlisp use, namely
compilation, loading, and execution. I asked him for an ISAAC demo, so he
proceeded to load his already compiled system, which he told me had taken 50
minutes to compile on the Dolphin. We chatted about features of Interlisp to
while away the 10 minutes it took the Dolphin to load his demo. He then ran
ISAAC on a number of physics problems stated in English. Problem 1 took 45
seconds. Gordon said that when running interpreted on the Dolphin this
problem had taken 180 seconds, as against 6.4 seconds running interpreted on
the 2060. He did not have timings for compiled code on the 2060, though 1-2
seconds is a reasonable guess. The remaining problems all showed similar
behavior.
Gordon commented, "This sort of performance really changes your style of using
computers. You arrange to work on other things while your program is loading,
or to go out for lunch while it is compiling."
My observations were in close agreement with Gordon's, despite my programs
being much smaller and also more CONS-intensive. One program took derivatives
of algebraic expressions. Taking the derivative of 3x↑2+ax↑2+ax+5 in Franz
Lisp on Diablo took 3.4 milliseconds, 4.8 milliseconds in a toy Lisp
system on the Sun terminal, and 6.6 milliseconds in UCI-Lisp on Sumex's KI-10.
On the Dolphin it took 160 milliseconds. (All these timings were obtained by
running the problem a large number of times, bringing the measured time up to
the order of many seconds.)
Another program converted logical formulas into disjunctive normal form. One
fairly large formula took 21 milliseconds in Franz Lisp, 29 milliseconds on the
Sun, and 1 second on the Dolphin.
Yesterday a contingent from Xerox visited Stanford to discuss Dolphins in
general and Dolphin Interlisp in particular. One topic was how to port
Interlisp programs to the Dolphin. Another was a breakdown of Interlisp
performance on the Dolphin; for example CONS was described as taking around 800
milliseconds. CONS and function-call-and-return were identified as performance
bottlenecks; CAR and CDR however were described as being fast, although no
figures were given for them. Masinter is optimistic about being able to make
30% or better improvements in the performance of CONS and call-and-return;
meanwhile one should expect particularly bad performance from programs that
make excessive use of CONS and function calls.
One conclusion from all this is that the cost-effective way to get Interlisp
cycles for those users for whom address space is not a problem is to buy a
2060, which gets you the equivalent of nearly 30 Dolphins, not to mention the
advantage of having more than one Dolphin's worth of power on those occasions
when you aren't competing with 30 other Interlisp users. I readily concede
that users running up against address space limitations have a problem.
However it is far from clear that Dolphins are the solution to that problem for
any but the most patient of users. If one is to believe Dick Gabriel's
claim that a Vax 11/780 becomes unusable with two Franz Lisp users, one must
also believe that a Dolphin becomes unusable as soon as it has more than 1/25
of an Interlisp user.
The other conclusion is that no one but a dyed-in-the-wool Interlisp enthusiast
is going to make serious use of Interlisp on the Dolphins. I am a Lisp
enthusiast myself, but not to the extent that I would bother to use a Lisp as
slow as that on the Dolphins.
Symbolics 3600's sound considerably more attractive in the light of these
Dolphin measurements. I am very anxious to see these measurements taken on the
3600.
Vaughan
∂16-Oct-81 0812 Betty Scott <CSD.BSCOTT at SU-SCORE> [Scott at SUMEX-AIM: YOUR MAIL]
Date: 16 Oct 1981 0809-PDT
From: Betty Scott <CSD.BSCOTT at SU-SCORE>
Subject: [Scott at SUMEX-AIM: YOUR MAIL]
To: JMC at SU-AI
Permission confirmed - John
John, would you please confirm permission to send your salary rate to
LLL.
Thanks,
Betty
---------------
Mail-from: ARPANET site SUMEX-AIM rcvd at 15-Oct-81 1157-PDT
Date: 15 Oct 1981 1152-PDT
From: Scott at SUMEX-AIM
Subject: YOUR MAIL
To: CSD.BSCOTT at SU-SCORE
Mail-from: ARPANET host SU-SCORE rcvd at 15-Oct-81 1126-PDT
Mail-from: ARPANET site S1-A rcvd at 15-Oct-81 1123-PDT
Date: 15 Oct 1981 1121-PDT
From: Paula Bauman <PJB at S1-A>
Subject: John McCarthy
To: csd.scott at SU-SCORE
CC: PJB at S1-A
John McCarthy asked me to request from you a letter indicating a new
salary rate, effective September 1, 1981, since his rate of pay here
at Lawrence Livermore Laboratory is based on his current rate of pay.
Please address letter to: Lawrence Livermore Laboratory
P. O. Box 808 L-276
Livermore, CA 94550
-------
---------------
-------
∂16-Oct-81 0859 FFL /
To: JMC, FFL
I think you did get something from Teller. I will go through the in box.
∂16-Oct-81 0915 FFL
To: JMC, FFL
There is a large notebook with information about nuclear regulation
which was sent to you from Dr. Teller's office.
∂16-Oct-81 0916 FFL
To: JMC, FFL
Are you having a committee meeting here this morning?
∂16-Oct-81 1000 RINDFLEISCH@SUMEX-AIM (SuNet) Re: Dolphin timings
Date: 16 Oct 1981 0912-PDT
From: T. C. Rindfleisch <RINDFLEISCH at SUMEX-AIM>
Subject: Re: Dolphin timings
To: pratt at SU-HPP-VAX, equip at SU-HPP-VAX
cc: RINDFLEISCH at SUMEX-AIM
In-Reply-To: Your message of 15-Oct-81 2336-PDT
There are indeed anomalies in benchmarked performance of the Dolphins so
far. Some programs run quite well and others quite poorly. There are
some known problems (such as need to speedup function calls and no
microcoded support for CONS) for which fixes are forthcoming. The
causes of other problems are still not known. Many of the programs
being run on our Dolphins are different from those at PARC which were
used to tune and prioritize developments for Dolphins. Getting this
experience was one of the main reasons Xerox was anxious to get Dolphins
in use at Stanford in the first place.
The Xerox Dolphin wizards have just been made aware of our performance
data recently and have been very responsive in committing to analyze the
causes. They need some time to do this carefully. I believe we are
still in an information collecting mode and I would encourage as much
testing of the Dolphin performance as possible together with working
with Xerox to see what the real limiting (not just current) problems
are. The performance comparisons Vaughan Pratt described in his earlier
msg are worthwhile and provocative but I think it is premature to reach
action conclusions. This same prudence must also apply when we can
finally benchmark other Lisp workstation candidates.
Tom R.
-------
∂16-Oct-81 1117 JEF via Ethernet host 50#12 summit meeting
To: JMC at SU-AI, RSE at SU-AI, csl.jlh at SU-SCORE
I would like to get the three of you together to attempt a prioritization
of the items on our equiment budget. Alternatives, based on what
I know of your schedules are Monday or Tues. mornings, or Wednesday evening.
In the latter case, you are invited to my place for the meeting.
∂16-Oct-81 1240 RPG@Sail (SuNet) Fairness
Date: 16 Oct 1981 1234-PDT
From: Dick Gabriel <RPG at SU-AI>
Subject: Fairness
To: equip at DIABLO, CSD.NOVAK at SU-SCORE,
CSD.GENESERETH at SU-SCORE
Vaughn's little benchmark leaned heavily on the
embarassment, CONS, on the Dolphin. I suspect that the Dolphin
can do better, and so JMC and I did the Takeuci function on SAIL, a Dolphin,
and an F2 (Foonly).
First, here is Takeuci:
(defun tak (x y z)
(cond ((not (< y x))
z)
(t (tak (tak (1- x) y z)
(tak (1- y) z x)
(tak (1- z) x y))))))
On the Dolphin, (TAK 18. 12. 6.) took 11.195 seconds compiled.
On the Foonly it took (MacLisp) 4.1 seconds compiled.
On SAIL (MacLisp) .83 seconds compiled.
So, SAIL (a KL-1090 [i.e. about 80% of a 2060]) was 13.5 times faster,
whereas, on Vaughn's little example, SAIL was 260 times faster, if I
am to believe the 160 millisecond figure.
-rpg-
∂16-Oct-81 1301 Wiederhold at SUMEX-AIM Dick gabriels sail account.
Date: 16 Oct 1981 1244-PDT
From: Wiederhold at SUMEX-AIM
Subject: Dick gabriels sail account.
To: jmc at SU-AI
cc: reg at SU-AI, pn at SU-AI, rpg at SU-AI, csd.hill at SU-SCORE
We have to drastically reduce S-1 charges at sail etc.
Since Dicks s-1 work is not being done at Sail and his account is held there
for LISP work could you pick up his charges there.
Thanks Gio
-------
∂16-Oct-81 1428 SIS Colloquium Notice for October 19 - 23, 1981
To: "@COLLOQ.DIS[INF,CSD]" at SU-AI
Date Place Person
Day Event From
Time Title
-------------------------------------------------------------------------------
10/19/81 MJ 146 Robert Blum
Monday PhD Oral Stanford University
12:00 p.m. ``Discovery and Representation of Causal
Relationships
from a Large Time-Oriented Clinical Database''.
10/20/81 MJH352 Horst Wesch
Tuesday Medical Journal Institut fur Nuclearmedizin, Heidelberg, Germany
1:30 p.m. Club ``MIDAS (Modular Interactive Database System)''.
10/20/81 Bldg. 200-30 Richard Weyhrauch
Tuesday Concurrency and AI Stanford University
2:30 p.m. ``The Computer Individual''.
10/20/81 Jordan 041 Jean-Pierre Barass
Tuesday Computer Science SLIGOS
4:15 p.m. Colloquium ``A Remote File System in APL''.
10/21/81 Terman 153 To be announced
Wednesday Computer Systems
4:15 p.m. Laboratory Seminar
10/23/81 MJ301 Jack Swartz
Friday Database Research N.Y.U.
3:15 p.m. Seminar To be announced
∂16-Oct-81 1605 FFL CIS Committee meeting
To: JMC, FFL
Monday, Oct. 19, 3:30 p.m. in CIS Conference Rm. ERL, to draw up policy
statement. I put on your calendar.
∂16-Oct-81 2230 RPG
∂16-Oct-81 2224 JMC comparison
I didn't know SAIL was slower than 2060 for LISP. Perhaps we should
try SCORE on the Takeuchi example for comparison.
SCORE varied more than SAIL, with some .80's and some .84's,
while SAIL was always .83. The 2060 has the B CPU while we have
the original CPU. Ralph and Marty have often discussed the pro's and
con's of getting the B CPU.
-rpg-
I assume this was MACLISP on SCORE?
∂16-Oct-81 2234 RPG
∂16-Oct-81 2233 JMC
I assume this was MACLISP on SCORE?
Yes.
∂16-Oct-81 2320 SL TOB
∂16-Oct-81 2319 JMC
When is Tom expected back?
SL - Nominally Oct. 24; possibly a few days later.
∂17-Oct-81 0032 Lowell Wood <LLW at S1-A> Timing
Date: 16 Oct 1981 2231-PDT
From: Lowell Wood <LLW at S1-A>
Subject: Timing
To: JMC at SU-AI, RAH at S1-A, minsky at MIT-AI,
danny at MIT-AI
CC: LLW at S1-A
Gentlemen:
It has been very aptly remarked that, in politics, timing is everything.
Getting (more than) one's fair share of the pie is, by definition,
politics. Getting a lunar colony into existence will require at least our
fair share of the pie. It therefore behooves us to time the announcement
of our proposal for a lunar colony project carefully and well, lest we
dribble out the word to the ever greater advantage of the undertaking's
natural (and acquired) enemies and the eventual boredom of its (present
and potential) friends. (A nominally hush-hush project along not totally
dissimilar lines into which Rod and I have been co-opted is presently
trashing its political prospects due to the inability of its principals to
refrain from ego-gratifying babbling about what it's doing and why; as a
result, its enemies are mobilizing, forewarned and forearmed, and its
friends are walking away from completely unnecessary early-time battles in
disappointment and disgust. It's rather sad to watch.)
We're going to be pitching our lunar colony proposal to human beings, and
we should keep foremost in mind that humans 1) want to back winners, and
2) react better to surprises than non-surprises (in that those naturally
opposed to a surprise proposal are disorganized and ineffective, and those
that are a priori favorable to it are interested and enthused). If we
handle this development intelligently (rather than immaturely), we'll 1)
set up the project to be a prima facie winner when it first surfaces
publically, by canvassing in confidence for financial and political
backing of tellingly great magnitudes prior to ***any** public
announcement, and 2) finally go public in a *really* big and completely
synchronized way, compared to which Jerry's (undoubtedly very
well-intentioned) suggestions will be quite puny: full-page annoucements
of the opening of the project's Terra Operations Center headed by
Shackleton's call-to-arms to potential volunteers in all the major
newpapers reinforced by favorable editorials, opinion-makers interviewed
in arranged `news' stories hymning the glories of the proposal, and
massively co-sponsored resolutions almost immediately introduced (with at
least tacit Administation backing) in both Houses of Congress directlng
NASA to fully cooperate and supply launch services and spare hardware
(e.g., RL-10s) in exchange for first crack at all scientific and technical
data gathered/generated, etc, etc. We'll thus get the personnel, hardware
and launch vehicles which we need to get the work well underway before the
opposition can mobilize, *if* we do the job right. Otherwise, the
proposal seems to me likely to merely become `walking dead'--it won't lay
down and die right away, but it will be clear to everyone that it has
*zero* future; the present age isn't friendly to Baltimore Gun Clubs.
Momentum is crucial in getting a project of this magnitude and audacity to
its goal, and, like virginity, once it's gone there's no retrieving it.
Therefore, let's plan and do our initial work in privacy, so that one of
our vital assets isn't thrown away for naught.
Lowell
I entirely agree with your last. In fact my reply to Pournelle's message
was the following paragraph. Would you like to add weight to it by one of
your own, suggesting that the project be entirely kept out of the press or
official circles or even any organization that might prematurely formulate
an attitude? The other possibility is to let well enough alone if you
think Jerry can't help talking. I have no evidence on that point one way
or the other.
All the suggestions you have made are worth thinking about. Rod Hyde
and Danny Hilles are gathering facts relevant to a cost estimate at
the moment. My intuition says that publicity should be timed for
when it can do the most good. We may have a better estimate in a
few days or weeks. We'll keep you informed.
∂17-Oct-81 1134 Jeffrey D. Ullman <CSD.ULLMAN at SU-SCORE> meeting
Date: 17 Oct 1981 1132-PDT
From: Jeffrey D. Ullman <CSD.ULLMAN at SU-SCORE>
Subject: meeting
To: engelmore at SUMEX-AIM, jmc at SU-AI, csl.jlh at SU-SCORE
As JLH will be out of town this week, let me propose 10AM, Mondaay
the 26th as a time to meet.
-------
Any chance of getting together this weekend?
∂17-Oct-81 1340 Lowell Wood <LLW at S1-A> More On Privacy
Date: 17 Oct 1981 1339-PDT
From: Lowell Wood <LLW at S1-A>
Subject: More On Privacy
To: jmc at SU-AI
CC: LLW at S1-A, RAH at S1-A, minsky at MIT-AI,
danny at MIT-AI
∂16-Oct-81 2244 John McCarthy <JMC at SU-AI>
Date: 17 Oct 1981 0044-PDT
From: John McCarthy <JMC at SU-AI>
To: llw at S1-A
I entirely agree with your last. In fact my reply to Pournelle's message
was the following paragraph. Would you like to add weight to it by one of
your own, suggesting that the project be entirely kept out of the press or
official circles or even any organization that might prematurely formulate
an attitude? The other possibility is to let well enough alone if you
think Jerry can't help talking. I have no evidence on that point one way
or the other.
All the suggestions you have made are worth thinking about. Rod Hyde
and Danny Hillis are gathering facts relevant to a cost estimate at
the moment. My intuition says that publicity should be timed for
when it can do the most good. We may have a better estimate in a
few days or weeks. We'll keep you informed.
[John: Jerry's personal reliability re private info is regrettably of no
moment. I'm told by several people that between dozens and hundreds of
space groupies, fanatics, etc., with Net access read the contents his MC
mailbox on a daily basis; sending him mail is like publishing the item in
the newspaper. We should keep him informed, but by telephone and with the
continuing admonition to keep the whole matter under his hat. Rod and I
will be tracking down the needed older NASA documents in Washington this
coming week. I suspect that, if we arrange things right, total project
costs might be only the start-up ones, and could thus be an order of
magnitude smaller than we guestimated at Larry's: ***if*** we can get the
Congress to direct NASA's donation of launch services and `excess'
hardware, and ***if*** we can contract with the media to sell footage for
a continuing series of specials dealing with the most exciting, human
interest-intensive happening of the decade, Rod and I think that we might get
the project launched onto a viable track for something of the order of $10
M--just what's needed to get the TerOpC (pronounced `terops', of course)
started and the all-volunteer staff in place and working productively
toward the first launch. Ten megabucks I quite confidently expect that we
could raise in a quarter's intensive effort; a hundred gets a *lot*
harder, being of the university multi-year endowment drive scale (and we
have no alumni, no in-place operation, no 70% incremental tax bracket,
etc.). Lowell]
∂17-Oct-81 1538 CSD.NOVAK@SU-SCORE (SuNet) Dolphin timings
Mail-from: ARPANET host SU-SCORE rcvd at 17-Oct-81 1535-PDT
Date: 17 Oct 1981 1532-PDT
From: Gordon Novak <CSD.NOVAK at SU-SCORE at SUMEX-AIM>
Subject: Dolphin timings
To: equip at SU-HPP-VAX, RPG at SU-AI, CSD.GENESERETH at SU-SCORE at SUMEX-AIM,
CSD.HBROWN at SU-SCORE at SUMEX-AIM, SCHOEN at SUMEX-AIM,
CSD.GERRING at SU-SCORE at SUMEX-AIM, CSD.NOVAK at SU-SCORE at SUMEX-AIM
I would like to clarify some of the statements attributed to me in
Vaughan Pratt's message concerning Dolphin timings.
First, let me state that the figures given for my program are those for
the first attempt at running it. It appears that relatively simple
changes (e.g., declaring global variables as such at top level) may
make substantial improvements in execution time. The figures are
also preliminary in the sense that the Dolphin's performance is
still being tuned, and large improvements are likely. Stanford has
benefitted from a close working relationship with Xerox PARC, and has
gotten early versions of the Dolphin as a result of this relationship.
We should be careful that we don't damage this relationship by bad-mouthing
the Dolphin's performance based on experience with the early versions;
we should be especially careful that statements to people outside the
local community are properly qualified, so that we are fair to
Xerox.
I did not find a ratio of performance of 28 "for all phases of Interlisp
use", as Vaughan says; rather, the figure was 28 for interpretive
execution (with the caveats mentioned above), 22 for compilation,
and 15 for loading of symbolic files. Ratios for compiled code are still
to be determined, as is the effect of making global variable declarations.
My comparisons were made against an unloaded 2060 (UTEXAS-20 late at night).
An unloaded 2060 is nice for benchmarks, since it is a stable environment,
but in a sense it is unfair to the Dolphin. The reason for this is that
performance of a machine like the 2060 degrades much faster than linearly
with incresing numbers of users. With more users, the working set of each
user gets smaller, and performance degrades rapidly due to page swapping.
A factor of 28 between a Dolphin and an unloaded 2060 does not translate into
equivalent performance between a Dolphin and a 2060 with 28 Interlisp
users; I would guess that the number of users a 2060 could support with
equivalent performance is more like 10 or 15.
When the Dolphin is used in its intended role as a personal machine (for
one or two users), slow compilation and slow loading of files are not such
a significant problem. One typically leaves one's Interlisp system up on
the machine for long periods of time; restarting from where one left off
takes only a few minutes. Interlisp easily and quickly recompiles the
few individual functions which have been changed during a session.
Address space limitations are a problem for virtually everyone in HPP,
and for many others in the A.I. community. Last week, there was a demo
of SRI's system for natural language access to databases; it runs as
three separate forks (each nearly full) which communicate with each
other via messages on "print" files. This breaking up of a program
to get around address limitations is not only clumsy, but hurts
performance as well. Getting more 2060 power isn't what we need.
CONS takes 800 microseconds (not milli), and this includes garbage
collection time.
Finally, I should emphasize that raw Lisp execution speed is not the
only important criterion. Someone who does a lot of big searches needs
fast execution. Personally, I find that I spend much more time
writing programs than executing them. The programming environment
of Interlisp more than makes up for slow execution on the Dolphin.
I am now using the Dolphin for serious work, and plan to continue.
--- Gordon
-------
∂17-Oct-81 1610 CLT
i am off to santa cruz, don't know if i'll get back for supper, so
don't count on it.
∂17-Oct-81 1750 pratt@SU-HPP-VAX (SuNet) Dolphin timings
Mail-from: SU-NET host SU-HPP-VAX rcvd at 17-Oct-81 1733-PDT
Date: 17 Oct 1981 17:33:16-PDT
From: pratt@Diablo at Sumex-Aim
To: CSD.GENESERETH@SU-SCORE at SUMEX-AIM, CSD.GERRING@SU-SCORE at SUMEX-AIM,
CSD.HBROWN@SU-SCORE at SUMEX-AIM, CSD.NOVAK@SU-SCORE, CSD.NOVAK@SU-SCORE at SUMEX-AIM,
RPG@SU-AI, SCHOEN@SUMEX-AIM, equip@SU-HPP-VAX
Subject: Dolphin timings
My apologies for appearing to be bad-mouthing the Dolphin. I'll try to get my
tests and my remarks both more neutral and more accurate in future.
Gordon is quite right about not nettling Xerox. Clearly the goal should be to
constructively encourage Xerox to improve the Dolphin in those areas where
it hurts the most. Equally clearly, this need not imply a commitment
on anyone's part to buy Dolphins in significant volume if it is eventually
concluded that they are not sufficiently effective Lisp tools. Both further
Interlisp development and further benchmarking (or just plain experience with
InterlispD), including getting programs adjusted to the Dolphin environment,
are needed before decisions of this sort can be made.
My apologies for getting Gordon's compile-time and load-time ratios wrong, I
misunderstood what he had told me about this. (The perils of being a
reporter!) My "800 milliseconds" for CONS was of course a typo for 800
microseconds, .8 seconds for a CONS would be nonsense even on an 8080. (I
wonder about that 800 microseconds. My DERIV program performed 305,000
conses and took 800 seconds. If conses take 800 microseconds then that only
accounts for 250 seconds. Hence either CONS does NOT dominate the
computation time of DERIV or CONS takes more like 2.4 milliseconds on the
Stanford Dolphins.)
I agree heartily with Gordon's statement that writing time is more important
than running time. That's why I prefer to write in a standard Lisp style,
without spending time worrying about how to transform my programs to eliminate
consing. If when I use a cons-intensive style my programs slow to a crawl then
I feel justified in lodging a performance complaint.
Gordon's upbeat remarks about using the Dolphin as a Lisp machine (which seemed
much more positive than when I was talking to him on Thursday) would appear
to contradict Dick Gabriel's feeling that a Vax 11/750 would be "unusable" on
performance grounds as a personal Lisp machine. It would be nice to get some
agreement from the Lisp community as to what level of performance is tolerable
on a personal computer. (As a Lisp user myself, I lean more to Dick's position
than Gordon's - nice environment notwithstanding, I think I'd find the Dolphin
slower than I could bear.)
Vaughan
∂17-Oct-81 2340 pratt@Diablo (SuNet) Fairness
Date: 17 Oct 1981 23:36:10-PDT
From: pratt at Diablo
To: equip, genesereth@score, novak@score
Subject: Fairness
(The following is in response to the Takeuc(h?)i benchmark from RPG.)
Here are the criteria I have been using to date in choosing Lisp
benchmarks.
1. The benchmark should solve a problem whose computer solution is frequently
called for.
2. The programming style used should be one that helps make programs more
writable, readable, maintainable, and portable.
3. Subject to criterion 2, the benchmark should implement an efficient
solution to the problem it solves.
Criterion 1 is to avoid the complaint commonly made about some benchmarks
that they do not test real world problems. Criterion 2 attempts to live up
to standards recommended by software engineers based on relative costs of
people and hardware, a ratio that programmers continue to have difficulty
believing in and adjusting to. Criterion 3 supplements criterion 1 by weeding
out programs that would not arise in practice on account of being too
inefficient. (For the most realistic benchmarks, criterion 3 should receive
the same degree of emphasis as in typical programming; since this emphasis is
highly variable the criterion may safely be interpreted somewhat vaguely.)
Customer benchmarks can afford to meet criterion 1 more narrowly than general
benchmarks, in that "wide use" can be construed to mean "heavily used by the
customer." Thus for HPP purchases it makes sense to use HPP programs as
benchmarks, at least up to a point. Current HPP programs give an accurate
estimate for immediate applications, but this accuracy starts to drift as
programs, programmers, and programming styles change. Thus for long-range
predictions, programs of general utility, in particular programs that have
found frequent application over a period of many years in many places and can
be expected to continue to show up as subroutines in many programs, make for
more satisfactory benchmarks.
A machine that requires benchmarks not meeting criterion 1 in order to look
good is not tuned for real world problems. If a programming style that
defeats criterion 2 is needed for a machine not to look like a dog compared
to other machines then that machine is not tuned for the economics of today's
computing milieu. Defeating criterion 3 to reduce the performance gap
between machines is like holding Olympic track events in molasses.
With these criteria in mind, I'd like to defend myself against the objections
that have been raised about one of my Lisp benchmarks consing excessively.
That it was considered excessive surprised me since I thought I had
implemented differentiation in a pretty standard Lisp style, and pretty
efficiently at that, meeting all three of my criteria.
To see what happened in the "real world" I went over to Macsyma and took the
derivative of the same expression used in my benchmark, 3*x*x+a*x*x+b*x+5.
Macsyma performed around 300 conses, of which somewhere between 150 and 200
appeared to be attributable to actually taking the derivative, the rest being
due to parsing and other overhead. My program performed only 61 conses, the
lower number probably being attributable to my not attempting any
simplification of the result.
Conclusion: I see no sustainable objection to my benchmark.
I might add that I chose it, along with two other benchmarks, purely using
my three criteria. I had no prior expectations that it would exercise one
part of Lisp more than another, although I also did not expect that it would
serve as a universal benchmark. There is no such thing as a universal
benchmark; at best you can only hope to have a broad range of representative
benchmarks. This is why I had three benchmarks solving problems from
three distinct areas, algebraic manipulation, logic, and sorting. Lack of time
has prevented me from covering yet more territory, and I am grateful for all
contributed benchmarks from other sources. However if your contribution
comes with the remark that I am being unfair in not having a sufficiently
broad range of benchmarks (as did the Takeuchi benchmark) I will be rather
piqued; I just don't have the resources to produce a sufficiently
representative range of benchmarks on my own.
I do not think that one should strive for fairness in benchmarking by
trying to distribute problems according to how well a particular machine
performs. Fairness does not mean that everyone should get a prize, but rather
that one judge using methods that lead to informative and accurate judgments.
If it turns out that a set of benchmarks representing a suitably diverse
range of applications runs poorly on a given machine in comparison to other
machines, I don't consider it fair to then attempt to put that machine in a
good light by looking specifically for benchmarks that favor that machine,
any more than I would consider it fair to look for benchmarks that make the
machine perform poorly relative to other machines.
In the case of the Takeuchi benchmark I get the feeling that it was chosen more
because it did no consing at all and was therefore likely to perform better
on the Dolphin than because of any consideration of representativeness.
Whether or not this was actually the case, I can at least raise the technical
objection that this benchmark fails my criteria 1 and 3. (Criterion 3 fails
because there is a closed-form expression for Takeuchi's function, permitting
it to be computed in constant time rather than in what appears to be
exponential time.)
One way to come up with a benchmark that meets my three criteria but that
should do as well as Takeuchi in making Dolphins look good would be to
implement ASSOC straightforwardly. This would not make me any happier about
maintaining a spirit of representativeness, but at least it would dispose of
my technical objections to the Takeuchi benchmark.
Incidentally, the Takeuchi benchmark consumes 2.1 seconds using Franz Lisp on
Diablo. (RPG's timings were .83 for Sail, 4.1 for the Foonly, and 11.2 for
the Dolphin.) For what it's worth a C version of it ran in 1.35 seconds on
the Vax and 1.9 seconds on the Sun.
Vaughan
∂18-Oct-81 0055 RPG@Sail (SuNet) For what it's worth
Date: 18 Oct 1981 0053-PDT
From: Dick Gabriel <RPG at SU-AI>
Subject: For what it's worth
To: equip at DIABLO
DERIV coincidentally was a very CONS intensive program. TAK
is function call intensive and has no CONSing of any kind.
By `fairness' I meant that it is rare in a `natural' Lisp program
that CONSing is done in such high percentages. Of course, CONSing
in such low percentages as in TAK is rare as well. Benchmarks
are worthles unless there is accompanying commentary on what they
test.
If TAK in C on the SUN and VAX are interesting, how about TAK in
FAIL on SAIL?:
.436 seconds
By the way, I didn't bum to any great extent, and I used 2 stacks.
-rpg-
∂18-Oct-81 0858 JK
∂17-Oct-81 2028 JMC problem with ekl
I'm not sure whether it's my problem or yours. CYCLE.PRF[F81,JMC] contains
the proof in question. My object was to rewrite so as to eliminate the
occurrence of drec and so obtain the recursion equation for append. Why
doesn't the last step work?
--------------
You were trying to use line 51 in rewriting which was a conditional :
unless the rewriter can verify somehow that the antecedent , list(cdr u)),
is true, it cannot use it the way you want. Seems that you should put more
sort info in.
∂18-Oct-81 2141 pratt@Diablo (SuNet) For what it's worth
Date: 18 Oct 1981 21:40:31-PDT
From: pratt at Diablo
To: RPG@Sail, equip@DIABLO
Subject: For what it's worth
From: Dick Gabriel <RPG at SU-AI>
Subject: For what it's worth
To: equip at DIABLO
DERIV coincidentally was a very CONS intensive program. TAK
is function call intensive and has no CONSing of any kind.
By `fairness' I meant that it is rare in a `natural' Lisp program
that CONSing is done in such high percentages.
My Macsyma data didn't sway you then? Are you saying that Macsyma is
an unnatural Lisp program or a rare one?
If TAK in C on the SUN and VAX are interesting, how about TAK in
FAIL on SAIL?:
.436 seconds
Using pretty straightforward assembly language on the Sun I measured .70
seconds. I'm surprised the Sail/Sun gap is so small, I thought KL-10's were
supposed to be blindingly fast. I certainly wasn't expecting the Sun to be 62%
of a KL-10 for a function-call-intensive benchmark!
∂18-Oct-81 2254 RPG@Sail (SuNet) Several points:
Date: 18 Oct 1981 2246-PDT
From: Dick Gabriel <RPG at SU-AI>
Subject: Several points:
To: equip at DIABLO
1. Macsyma IS a natural program: it did a lot of CONSing, but
I doubt that the percentage of CONSing to other things is anywhere
near as high as in VRP's DERIV program. This is because MACSYMA does a lot
of stuff while taking derivatives (for example, displaying the answer).
I write much Lisp code, and my code certainly does not mention several
conses per line, as Pratt's DERIV function does. The Macsyma derivative
is not Pratt's.
Any benchmark such as APPEND and DERIV is pathological. I never see such
code AS THE NORM. I see: type testing, MEMQ's, EQ's, COND's, lambda-binding,
array/vector access/assignment, function calls. The programs I deal with
do a lot of CONSing, but not at the rate that Pratt's does. If his DERIV
is natural, then the Dolphin is 260 times slower than SAIL. No one else
reports that. QED.
2. The `art' of benchmarking is subtle. For example, what does it
mean to ``time'' a benchmark? The assembly language code I wrote was
loaded in MacLisp, and I used the timing mechanism I always do to
time things (to be consistent). The mechanism for measuring times
on SAIL is the RUNTIM UUO, which is known to measure a quantity related
to actual EBOX execution time. I'm not sure what it measures, since it
appears to count as EBOX time for my job the code run at interrupt level
while I am active (such as interrupts for character input from any user).
It may count cache filling time. Recall that the memory on SAIL consists
of 256k of 900 nanosecond memory and 1 meg of 3 microsecond memory. Until
the cache is filled I'm losing. Of course, I get charged for the execution
of RUNTIM. With a benchmark so short, this is significant.
The timing methodology counted several Lisp function calls, so
I eliminated that, redid it, and got 380 milliseconds. Even at that
I don't know what I measured. When I write up the Lisp Evaluation
results, I will read the system code, study the hardware, and try to
relate my results to some reality. Now that you know what I know
about my measurements, what did Pratt measure?
-rpg-
∂18-Oct-81 2354 RPG
∂18-Oct-81 2257 JMC your latest message
Any comments on Chandrasekar proposal?
I finished it a little while ago. There are several problems I see
with it. Some technical, one political. I'll state my political one
and let you decide whether it matters. I wanted to think about it over
night, but here are some off the top of my head thoughts:
1. I'm not sure they can make it work the way they say they can,
though as a scientific endeavor I think they will do ok. Assume that
they can sort of make it work. With so many foes of nuclear power and AI around,
can we stand to have something scary like AI perform an serious error in a
nuclear setting? Or is this never intended to directly enter a real
situation for a long time?
2. I'm not sure they can make it work because of the decomposition assumption.
Though an NPP can be broken down according to their hierarchical theory, the
malfunctions may cross those boundaries at will. The blackboard mechanism
is a poor communication device for those situations. More later.
3. It seems that they need a minimum of a constraint system to model the
plant. I think they need to consider rates of change in sensor readings,
that they need to model some real-valued things and essentially have
a physical model at all times to answer Teller's question of what would
happen if... For instance, I understand that some hydrogen gas
was created in TMI; could their mechanisms deal with that? How about
predicting it? The rates of changes of things can interact in ways that
their discrete methods cannot handle.
The kinds of facts that can appear on the blackboard are quite simple.
One might think that some specialist(s) or other might deduce a complex
fact such as a conditional, or a causal relationship for the
later WWHI phase. Perhaps they would want to post some equations
on the board and have some system try to solve them. I don't know
enough about power plants to tell how complex they need to get.
4. I suspect that they will need to consider competing alternatives more
than they seem to believe, where there is more or better evidence for one
thing or several, but no single theory evolving. Some qunatitiative reasoning
here and in 3 is necessary.
5. They seem to want to have 2 models for the system. One for diagnosis, one
for prediction. Both sound as if they might be partially procedural. I'd rather
see a single model stating facts and quantitative relations among things which
can then form the basis for the various types of reasoning. But, verifying the
facts seems crucial to the success. Do you want a typo melting down a plant?
Do you want Jerry Brown finding the typo?
6. Their system seems to be a subset of the one Paul Martin proposed,
and his had some of the robustness wrt competing theories that I
mentioned above. However, his implementation was very slow, so I wonder
about response time (maybe).
7. On the other hand, if they just want to deduce some of the simpler
things about the plant, then I think they'll do ok. As unpleasant as
the KE approach might seem to our style, for highly structured domains
it seems to work well. I'm not sure this domain fits the bill.
8. As far proposals for work in this field go (knowledge engineering)
I think they know what they are talking about. If you trust that this
technology can work in this domain, I think they can pull it off as well
as anyone. I'll make inquiries about the group tomorrow.
-rpg-
Many thanks. I await tomorrow.
∂19-Oct-81 0938 RINDFLEISCH@SUMEX-AIM (SuNet) FYI - Other Lisp Timing Thrashes
Date: 19 Oct 1981 0926-PDT
From: Rindfleisch at SUMEX-AIM
Subject: FYI - Other Lisp Timing Thrashes
To: Equip at SU-HPP-VAX
cc: [SUMEX] at SUMEX-AIM, ETHERNET at SUMEX-AIM, DEV at SUMEX-AIM, GRP:
1 18 Oct Masinter at PARC-MAXC some more races out of the past
2 18 Oct Masinter at PARC-MAXC timings - fyi
1 -- ************************
Mail-from: ARPANET host PARC-MAXC rcvd at 18-Oct-81 1249-PDT
Date: 18 Oct 1981 10:12 PDT
From: Masinter at PARC-MAXC
Subject: some more races out of the past
To: Rindfleisch@sumex-aim
---------------------------
Mail-from: Arpanet host MIT-MC rcvd at 26-FEB-81 2243-PST
Date: 26 Feb 1981 14:42:52-PST
From: CSVAX.fateman at Berkeley
To: CSVAX.jkf@Berkeley, jlk@mit-mc, lisp-forum@mit-mc, rz@mit-mc
Cc: CSVAX.fateman@Berkeley
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
| | UCILISP | INTERLISP | MACLISP |Franz/VAX|
|-------------+---------+-----------+---------+---------|
| Interpreter | 57.0 | 26.0 | 22.8 | 65.0 |
|-------------+---------+-----------+---------+---------|
| Compiler | 2.90 | 15.0 | 0.69 | 1.1 ** |
|-------------+---------+-----------+---------+---------|
Times are for (TAK 4 2 0), where TAK is an interesting function
defined by Mr. Ikuo Takeuchi.
(DEFUN TAK (X Y Z)
(COND ((GREATERP X Y)
(TAK (TAK (SUB1 X) Y Z)
(TAK (SUB1 Y) Z X)
(TAK (SUB1 Z) X Y) ))
(T Y) ))
(**) 5.3 with (1- x) etc [no other declarations, so greaterp is closed comp.]
4.1 with local function declaration (fast subroutine call)
1.1 with > open compiled
times on a VAX 11/780 at Berkeley, Feb. 26, 1981
------------------------------------------------------------
2 -- ************************
Mail-from: ARPANET host PARC-MAXC rcvd at 18-Oct-81 1249-PDT
Date: 18 Oct 1981 09:55 PDT
From: Masinter at PARC-MAXC
Subject: timings - fyi
To: LispGroup↑, Rindfleisch@sumex-aim
Reply-To: Masinter
---------------------------
Mail-from: Arpanet host MIT-MC rcvd at 1-MAR-81 2221-PST
Date: 2 March 1981 00:55-EST
From: Charles Frankston <CBF at MIT-MC>
Subject: timings
To: CSVAX.fateman at BERKELEY
cc: LISP-FORUM at MIT-MC, masinter at PARC-MAXC, RWS at MIT-XX,
guttag at MIT-XX
It is rather obvious that the timings you distributed are wall times for
the Lisp Machine, whereas the Vax and MC times count only time spent
directly executing code that is considered part of Macsyma. Ie. the
Vax and MC times exclude not only garbage collection, but operating system
overhard, disk i/o and/or paging, time to output characters to terminals, etc.
I submit comparing wall times with (what the Multics people call) "virtual
CPU" time, is not a very informative excercise. I'm not sure if the Lisp
Machine has the facilities to make analagous measurements, but everyone
can measure wall time, and in some ways thats the most useful comparison.
Is anyone willing to try the same benchmarks on the Vax and MC with just
one user on and measureing wall times?
Also, are there yet any Lisp machines with greater than 256K words? No
one would dream of running Macsyma on a 256K word PDP10 and I presume that
goes the same for a 1 Megabyte Vax. The Lisp Machine may not have a time
sharing system resident in core, but in terms of amount of memory needed
for operating system overhard, the fanciness of its user interface
probably more than makes up for that. I'll bet another 128K words of
memory would not be beyond the point of diminishing returns, insofar
as running Macsyma.
Lastly, the choice of examples. Due to internal Macsyma optimizations,
these examples have a property I don't like in a benchmark. The timings
for subsequent runs in the same environment differ widely from previous
runs. It is often useful to be able to factor out setup times from a
benchmark. These benchmarks would seem to run the danger of being
dominated by setup costs. (Eg. suppose disk I/O is much more expensive
on one system; that is probably not generally interesting to a Macsyma user,
but it could dominate benchmarks such as these.)
I would be as interested as anyone else in seeing the various lisp systems
benchmarked. I hope there is a reasonable understanding in the various
Lisp communities of how to do fair and accurate, else the results will be
worse than useless, they will be damaging.
------------------------------------------------------------
-------
∂19-Oct-81 0939 FFL
To: JMC, FFL
Call from Prof. Moses office at MIT asking if you have your recommendation
for Patrick Winston. Said they need it within two days.
∂19-Oct-81 1000 JMC*
continue ekl
∂19-Oct-81 1000 JMC*
syllabus
∂19-Oct-81 1046 pratt@Diablo (SuNet) Several points:
Date: 19 Oct 1981 10:45:40-PDT
From: pratt at Diablo
To: RPG@Sail, equip@DIABLO
Subject: Several points:
[Seems to me benchmarking generates more debate than information. -vrp]
From: Dick Gabriel <RPG at SU-AI>
1. Macsyma IS a natural program: it did a lot of CONSing, but
I doubt that the percentage of CONSing to other things is anywhere
near as high as in VRP's DERIV program. This is because MACSYMA does a lot
of stuff while taking derivatives (for example, displaying the answer).
The measurements I made of Macsyma's differentiator did not include the work
done during display, nor during parsing. Nor should it if you are just
comparing differentiation programs. What is the "lot of stuff" Macsyma does
WHILE taking derivatives?
I write much Lisp code, and my code certainly does not mention several
conses per line, as Pratt's DERIV function does.
Ok, let's see your version of DERIV. I'll be interested to see how you manage
to use fewer conses per line. No fair merely spreading the code over more
lines.
The Macsyma
derivative is not Pratt's.
Your argument here seems to be that because you do something Macsyma does it
too.
We can resolve the question of what Macsyma does by looking at the Macsyma
code for DIFF, a copy of which I have requested from Jeff Golden. (Maybe I
should have asked Fateman, on the principle that one goes to the Soviet
embassy to make casual inquiries about US military secrets. I heard Moses
was rather upset that Fateman had ported Macsyma to Franz Lisp.)
2. ...[on the meaning of time]...
.
.
.
Now that you know what I know
about my measurements, what did Pratt measure?
Depends on the machine, but in all cases my wristwatch is the final authority.
Sun and Dolphin: wristwatch time over a large number of runs. (On the
Dolphin this seems to agree with the time returned by
InterlispD's TIME function.)
Vax: user cpu time as returned by the 'times' kernel call.
The user cpu time has the following two properties:
(1) I have observed little variation of this parameter
over a wide range of system loads, cache usage
considerations notwithstanding.
(2) For very low system loads I have seen it
come to within 90% or so of wristwatch time.
These two observations together imply that the 'times'
kernel call is a reliable indicator of user cpu time.
∂19-Oct-81 1205 Engelmore at SUMEX-AIM ARPA funding
Date: 19 Oct 1981 1159-PDT
From: Engelmore at SUMEX-AIM
Subject: ARPA funding
To: JMC at SU-AI, TOB at SU-AI, ZM at SU-AI, DCL at SU-AI,
Wiederhold at SUMEX-AIM
cc: csd.BSCOTT at SU-SCORE
Ron Ohlander called this morning to say that the ARPA order had been
signed and backdated to October 1, and that he was hand carrying it
over to Navelex. He suggested I call John Machado in a few days to
get an estimate of when the money will arrive and, more importantly,
authorization to start charging to the new contract immediately.
I'll call Machado on Friday.
Bob
-------
Thanks Bob.
∂19-Oct-81 1447 Jrobinson at SRI-AI Tinlunch readings for this Thursday
Date: 19 Oct 1981 1447-PDT
From: Jrobinson at SRI-AI
Subject: Tinlunch readings for this Thursday
To: tlgrp:
Frazier, "Constraints, Control and Strategies in Sentence Comprehension."
25 copies now on Barbara's cabinet.
-------
∂19-Oct-81 1525 Gene Golub <CSD.GOLUB at SU-SCORE> Faculty Appointment
Date: 19 Oct 1981 1518-PDT
From: Gene Golub <CSD.GOLUB at SU-SCORE>
Subject: Faculty Appointment
To: CSD-Faculty: ;
cc: csd.broder at SU-SCORE
There is a possibility that Jack Schwartz and Fran Allen would be interested
in positions at Stanford.
They will be visiting on Friday, October 23; Jack speaks in the database
seminar at 3:15. If you would like to speak to Jack or Fran, please let me
know at your earlist convenience.
GENE
-------
∂20-Oct-81 0140 Jerry E. Pournelle <POURNE at MIT-MC> at last...
Date: 20 October 1981 04:37-EDT
From: Jerry E. Pournelle <POURNE at MIT-MC>
Subject: at last...
To: MINSKY at MIT-MC, jmc at SU-AI, llw at SU-AI
Delivering aoubt 120 pages of COUNCIL stuff to Gould for
mailing; I drive it down to Downy tomoroow, and after that I get
to wait for everyone to play about with it.
it si is stilll pretty rough, but there is so much of it I
wasn't abole to do more. At that I boiled it from about twice
the size.,
Sigh.
JEP
∂20-Oct-81 0900 RPG Terminal
To: JMC, LGC, JK
Ralph indicated that only 2 terminals could be installed in 360 due
to wiring limitations. That seems consistent with my inspection.
Since Lew isn't there too often (yet) is this a problem (yet)?
-rpg-
∂20-Oct-81 1508 FFL
To: JMC, FFL
Please call Miro Todorovich. 212 864 0645.
∂20-Oct-81 1543 Gene Golub <CSD.GOLUB at SU-SCORE> Re: Allen and Schwartz
Date: 20 Oct 1981 1540-PDT
From: Gene Golub <CSD.GOLUB at SU-SCORE>
Subject: Re: Allen and Schwartz
To: JMC at SU-AI, GHG at SU-AI
cc: CSD.GOLUB at SU-SCORE
In-Reply-To: Your message of 19-Oct-81 1909-PDT
That seems like a good idea. May I join you? GENE
-------
Fine. We'll plan for it then.
∂21-Oct-81 2008 Denny Brown <CSD.DBROWN at SU-SCORE> My plans: Now to 1/82
Date: 21 Oct 1981 1331-PDT
From: Denny Brown <CSD.DBROWN at SU-SCORE>
Stanford-Phone: (415) 497-2274
Subject: My plans: Now to 1/82
To: bboard at SU-SCORE, csd.bbd at SUMEX-AIM,
CSD-Faculty: ;,
CSD-Secretaries: ;
I will resign as Associate Chairman to accept a job at Teknowledge
as soon as a replacement can be found. Starting now,
I will be 80% at Stanford, and 20% at Teknowledge.
Plan: MWThF at Stanford, T at Teknowledge.
50% Stanford, 50% Teknowledge as soon as possible.
100% Teknowledge by Jan. 1.
My immediate Stanford priorities are:
1. Find a replacement.
2. Prepare information for him/her.
3. Train the new members of my old staff. I.e. Susan, Marilynn,
Nancy, and a replacement for Jake's old job.
4. CS107
5. Everything else.
This means that much of the "everything else" work will not get done.
Gene will be asking for suggestions, both about particular candidates
and general qualities.
It is not yet time for a "good-bye" message.
-Denny
-------
∂21-Oct-81 2011 Irmgild Schack <CSD.IRMGILD at SU-SCORE> Schedule for J. Schwartz and F. Allen
Date: 21 Oct 1981 1508-PDT
From: Irmgild Schack <CSD.IRMGILD at SU-SCORE>
Subject: Schedule for J. Schwartz and F. Allen
To: CSD-Faculty: ;
cc: icl.white at SU-SCORE, csd.broder at SU-SCORE
@subheading(Schedule for J. Schwatz and F. Allen Friday, October 23)
@tabclear
@begin(Verbatim)
@u(Schwartz) @u(Allen)
9:00 Ullman Ullman
9:30 X X
10:00 Cheriton
10:30 Cheriton
11:00 White/EE White/EE
11:30 Rosse (Dean) Rosse (Dean)
12:00-1:30 Lunch CSD Faculty
1:30 Roth(Robotics) Roth(Robotics)
2:00 CS Students CS Students
2:30 X X
3:15-4:15 Seminar/Lecture X
4:15-5:30 T G I F
@end(Verbatim)
-------
∂21-Oct-81 2016 RINDFLEISCH@SUMEX-AIM (SuNet) [T. C. Rindfleisch <RINDFLEISCH>: System Industries Periphs for VAX]
Date: 21 Oct 1981 1739-PDT
From: T. C. Rindfleisch <RINDFLEISCH at SUMEX-AIM>
Subject: [T. C. Rindfleisch <RINDFLEISCH>: System Industries Periphs for VAX]
To: Equip at SU-HPP-VAX
1) 20-Oct To: SAD at SRI-KL, H System Industries Periphs for
2) 21-Oct Tony Holland <tony a Re: System Industries Periphs
Msg 1 -- ************************
Date: 20 Oct 1981 2215-PDT
From: T. C. Rindfleisch <RINDFLEISCH>
Subject: System Industries Periphs for VAX
To: SAD at SRI-KL, Holland at SRI-KL
cc: TUCKER
The other day you mentioned someone at SRI had (bad) experience with
System Industries disks and tapes for your VAX's. Can you give me a
pointer to the cognizant person so we can get more details?
Thanks, Tom R.
-------
Msg 2 -- ************************
Mail-from: ARPANET host SRI-KL rcvd at 21-Oct-81 1706-PDT
Date: 21 Oct 1981 1658-PDT
From: Tony Holland <tony at SRI-KL>
Subject: Re: System Industries Periphs for VAX
To: RINDFLEISCH at SUMEX-AIM, SAD at SRI-KL, Holland at SRI-KL
cc: TUCKER at SUMEX-AIM, tony at SRI-KL
In-Reply-To: Your message of 20-Oct-81 2215-PDT
Tom I can help get you the information. We are currently in the process
of returning the tape drive we purchased from them. It took a lot of effort on
our part to make them finally operate without crashing the system. Most
oof the debugging had to be done by our systems programmer. They seemed
unable to give us very many answers. As for the disks. We are right back
into trouble again with the disk. Seems like alot of problems with both
the SBI interface and the disk interface. To sum up we are very unhappy
with the reliability of their equipment.
tony
-------
-------
∂21-Oct-81 2136 ullman@Diablo (SuNet) meeting
Date: 21 Oct 1981 14:10:17-PDT
From: ullman at Diablo
To: equip
Subject: meeting
As we are awaiting the results of the ``summit meeting'' to negotiate
priorities, there is probably little business to come before our
regular Friday meeting. I am therefore going to cancel it just
this once. I hope that by a week from Friday we can resume meeting
and begin the process of putting the proposal together.
In the meanwhile, everybody who has a piece to write, please work
on it, at least for the 2 hours we would have spent arguing over
how many LISP cycles would fit on the head of a pin.
If anyone has urgent business to bring up, I'll reconvene the meeting.
∂21-Oct-81 2212 Tom McWilliams <TM at S1-A> Schwartz talk and dinner at Louie's
Date: 21 Oct 1981 2204-PDT
From: Tom McWilliams <TM at S1-A>
Subject: Schwartz talk and dinner at Louie's
To: jmc at SU-AI
∂21-Oct-81 2150 John McCarthy <JMC at SU-AI> Schwartz talk and dinner at Louie's
Date: 21 Oct 1981 2153-PDT
From: John McCarthy <JMC at SU-AI>
Subject: Schwartz talk and dinner at Louie's
To: pmf at S1-A, jbr at S1-A, tm at S1-A
Prof. Jack Schwartz of NYU will talk at 3:15 on Friday in Wiederhold's
database seminar on "ultra-computers". There will be a dinner for him
and Fran Allen Friday at Louie's at 7pm for whoever would like to come
but let me know asap.
[John,
Thanks for the invitation, but I won't be able to make it.
Tom]
∂22-Oct-81 0841 SIS Computer Science Colloquium Notice of October 26 - 30, 1981
To: "@COLLOQ.DIS[INF,CSD]" at SU-AI
Date Place Person
Day Event From
Time Title
-------------------------------------------------------------------------------
10/26/81 Math380C Liviu Lustman
Monday Numerical Analysis Old Dominion University
4:15 p.m. Seminar ``Generalized Dufort-Frankel Spectral Methods''.
10/27/81 Bldg. 200-30 Carl Hewitt
Tuesday Concurrency and AI MIT
2:30 p.m. ``The Scientific Community Metaphor''.
10/27/81 Jordan 041 Vint Cerf
Tuesday Computer Science Defense Advanced Research Projects Agency
4:15 p.m. Colloquium ``Current Research Topics in Computer
Networking''.
10/28/81 Terman 153 Len Bosack & George Schnurle
Wednesday Computer Systems Computer Science Dept., Stanford
4:15 p.m. Laboratory Seminar ``A High-Performance Ethernet Interface''.
10/30/81 MJ301 Arthur Keller
Friday Database Research Stanford University
3:15 p.m. Seminar ``Efficacy of Replication in Distributed
Databases''.
∂22-Oct-81 1143 JMC
To: "@PROVE.LIS[F81,JMC]"
Meeting 4pm today on proving facts about LISP programs.
∂22-Oct-81 1529 Jrobinson at SRI-AI Tinlunch ALERT!
Date: 22 Oct 1981 0959-PDT
From: Jrobinson at SRI-AI
Subject: Tinlunch ALERT!
To: tlgrp:
We will meet in EK242 today, instead of in the park. Please alert
tinlunchers in your vicinity who may not have read their mail today.
Jane
-------
∂22-Oct-81 1335 RWW
∂22-Oct-81 1143 JMC
To: "@PROVE.LIS[F81,JMC]"
Meeting 4pm today on proving facts about LISP programs.
I am working at home today so I won't be there.
Richard
∂22-Oct-81 2009 mogul@Diablo (SuNet)
Date: 22 Oct 1981 17:29:15-PDT
From: mogul at Diablo
To: equip
As of a week ago, I am the elected student representative to whatever
committee plans future equipment purchases for the Department. I've
read all the messages to "equip" since late August, and I have some
comments; of course, since I've not been present at any meetings, these
might be naive, but I'm not sure student's interests have been
adequately considered.
(I.) Who is the equipment for? -- Brian, at one point, stated that he
thought "this committee should think about buying departmental
computing resources and not HPP computing resources." Since I believe
I should represent all students, rather than participants in sponsored
research, I want to push this point of view.
I think we need to identify the needs of the students outside those
which are met by research sponsorship. For example, many (if not all)
students do things that do not really fall into either sponsored
research or LOTS-supported areas. I know from personal experience that
the PRIMARY inconvience I suffer as a user of the "common" machines is
lack of disk space; lack of CPU cycles is also a problem, of course,
but only a certain times of the day. Just because a student is not
writing 20000 line LISP programs does not mean that s/he is not doing
useful work; at the very least, I think the primary responsibility of
the department (if not this committee) is to provide a good environment
for education. Also, I'd like to stress my feeling that master's
students are often treated rather shabbily when it comes to resources.
(II.) Given this, I think what the DEPARTMENT needs (as opposed to any
given research group) is (1) File space, probably on a server [instead
of merely adding disks to existing systems], (2) Terminals, and (3)
more general computing cycles.
(1) File server: I've noticed that, except for the brief discussion of
11/750's, that the machine proposed for a file server is the Foonly.
I'm not too familiar with these, but do we really want to invest in
36-bit machine when the world is moving more and more to 8 bit bytes?
{maybe we do, but let's examine this.} Also, if the Foonly will only
take a small set of peripherals, what if we don't like them/want to buy
cheaper/etc? [If it uses Unibus peripherals, that's another thing.]
Finally, what about maintenance 10 years from now? Are we going to be
holding it together with rubber bands, a la the Data Disk?
I think a small Vax (750) with non-Dec disks/tapes sounds better,
except possibly from the point of view of peripherals maintenance.
(2) Terminals: Faculty members with terminals in their offices and at
home, or researchers with their own dolphins, may not realize that most
students do not have the same access to the computer systems. I
suggest that, unless it is done for the purpose of restricting the use
of CPU cycles, the current shortage of terminals is a waste. I think
one solution would be to purchase a number of simple terminals (h-19s,
e.g.). Preferably, we should look into using Sun's (in inexpensive
configurations) to offload some of the editing load from the larger
machines; my impression is that screen editors (Emacs, TVedit, etc.)
are using up far more Score cycles than they should.
I often walk into a terminal room to use an Alto as an Alto (e.g., to
use Icarus, Sil, etc.) and find that the ones that aren't broken are
all being used as simple network terminals to Shasta/Diablo/Sail/Score/etc.
(3) General computing cycles: I think there are several ways to meet
this. The machines commonly named have been: Foonly 2080 S1 11/780
11/750 Sun. I think it would be a mistake to go for a machine for
which there was a lack of language support NOW, or for which there was
a poor operating system. (Look what happened to the 4331; why should
we waste programmer time on such things when it could be directed at
research and not wheel-reinvention.) I think that this argument
restricts the choice to the 2080 or the 11/780 (the 750 isn't worth it
for computes.) Suns might be useful in the future, but let's not count
on them for computes until a working operating system is here.
As for chosing between the 2080 and 11/780, I'd rather have both, but
given the current lack of a generally available Unix system, and a
system with a large address space available to ALL languages (Pascal,
Fortran, C, Pascal*, etc., besides Lisp), I think the 11/780 might be a
better choice. We've got TOPS-20 for them that wants it, but only some
students have Unix access now. Buying Diablo and perhaps adding some
more disk space might be just the thing.
Just remember: "use of computers" is NOT just use of computer cycles;
Lisp users may feel this way, but I think most students are more concerned
about basic access.
-Jeff
1. I agree about the main problem being lack of file space, and it was
the unanimous consensus that it should be in the first year's DARPA budget.
The Foonly is the prime candidate (no need to make a final decision
until the money is in our hands) mainly for the reasons that caused you
to advocate the VAX. D.E.C. has no current plan to offer large disks
on the VAX (we wrote to Gordon Bell about the file server problem and
got no sympathy), and the Foonly now offers inexpensive CDC 670 megabyte
disks, and we plan to get them to promise to interface the cheap
imitations of the IBM 3380 thin head disks when they appear. Foonly
also offers the Ethernet interface at no extra cost.
2. I also agree that terminals for everyone is the second major problem
for the Department. For this reason I have been pushing the inexpensive
terminal based on the SUN boards. However, DARPA cannot accept supplying
the whole department with terminals as one of its goals, so that can't
be included in the proposal. On the other hand, there is $150,000 in
a grant from the Stauffer Foundation that has been earmarked for this
purpose for some years. We would have them by now if the SUN project
hadn't stalled of an outside purchase by claiming that the SUN board
would be directly useful for the purpose and also misstating when it
would be available.
3. DARPA has (by law) to take the position that the equipment it supplies
are for the benefit of the research it sponsors, but it is inclined
to take an expansive view of the extent to which its interests are
advanced by making the equipment it provides available to potential
DARPA sponsored researchers. What this amounts to is that we can
get a big head start on equipment with DARPA aid but must find at least
some other money.
4. Getting more time-shared cycles is being postponed, because there
is no available machine a lot better than what we have. Anyway something
has to be left out of the first year's budget. We will be able to
look at the 2080 and the S-1 before having to decide. Foonly
doesn't offer a machine much faster than the 2060, and 18 bit addresses
aren't acceptable anyway.
5. In general I am not an enthusiast for personal machines in terminal
rooms. I will want one when it will fit on my desk and not require a
fan. However, personal machines seem to be the idea whose time has come.
6.There is a very serious problem of financing the current computer
complex quite apart from that of supporting new machines.
7. Did you receive my previous message. There seems to have been some
problem with messages.
∂23-Oct-81 0017 Robert Elton Maas <REM at MIT-MC>
Date: 23 October 1981 03:15-EDT
From: Robert Elton Maas <REM at MIT-MC>
To: JMC at SU-AI
Indeed, the critical thing is to get out to space an avoid nuclear
war, not to make the final decision what energy source people will
use in 1000 years. Regardless of whether we tap solar energy radiated
from stars, or pump unburned (unfused) hydrogen out of stars to fuel
our fusion reactors, I forsee physically inhabiting our galaxy
and neighboring galaxies long before the "heat death". (Unless
of course somebody already is there and won't share with us.)
∂23-Oct-81 1114 Grosz at SRI-AI Talk on Thursday Oct. 29 instead of tinlunch reading
Date: 23 Oct 1981 1046-PDT
From: Grosz at SRI-AI
Subject: Talk on Thursday Oct. 29 instead of tinlunch reading
To: tlgrp:
cc: AIC-Associates:
Gordon Novak will talk about his work on including reference
in LISP at 10:45 next Thursday in EK242. Tinlunch discussion
will follow (after a short break to get food). An abstract for
Gordon's talk is appended to this message:
GLISP: An English-Like Programming Language
Gordon S. Novak Jr.
Heuristic Programming Project
Stanford
GLISP is a LISP-based language which is compiled into LISP.
The compiler maintains a computational "context of
conversation", which flows through the program along the
control flow paths. As in English, GLISP allows definite
reference to objects which are in context or to their
features; newly referenced objects are added to the
context, permitting further definite references relative to
them. The ability to use definite reference eliminates the
need to specify complete access paths, allowing radical
changes to object structures with no changes to the code
which references them. GLISP objects may be LISP objects or
Units in a representation language. Compilation is
performed relative to a knowledge base of structure
descriptions and other information (e.g., definitions of
virtual fields and ways of referencing the objects in
English). This permits operations which are often done
interpretively in existing systems (e.g., invoking an
inherited if-needed procedure) to become compile-time
computations. In this way, GLISP provides the benefits of
object-centered programming within the LISP environment with
high runtime efficiency.
-------
∂23-Oct-81 1240 mogul@Diablo (SuNet) equipment
Date: 23 Oct 1981 12:37:36-PDT
From: mogul at Diablo
To: JMC@Sail, mogul@DIABLO
Subject: equipment
I'm free this afternoon after about 2:15; I'll try to drop by your office
at some point, or you can reply to me here.
-Jeff
2:15 will be fine.
∂23-Oct-81 1514 Danny Hillis <DANNY at MIT-AI> Old NASA docs.
Date: 23 October 1981 17:33-EDT
From: Danny Hillis <DANNY at MIT-AI>
Subject: Old NASA docs.
To: RAH at SU-AI, LLW at SU-AI
cc: MINSKY at MIT-AI, DANNY at MIT-AI, JMC at SU-AI
NASA claims not to have them in their library. Here is where they are:
N62-13348 : One way missions
This is an Istitute for Areospace Studies (I.A.S.) report #62-131.
Do you know who or where they are?
N64-23712 : Expandible lunar shelters. Ths is pp255-284 in a collection
of papers starting with N64-23701.
This contained in an Air Force Systems Command Report #AD-432006, or can be
ordered from DDC as N6423717. You LLL should be able to get DDC stuff.
N65-16672 : Centaur RL10 specs
N70-17551 : Umanned lunar cargo vehicle using RL10
These are available from NTIS. I have ordered them. If you feel rich and in a hurry
call (703) 557-4700 and ask for "Rush Handling Service". For about $20 per item, they
will ship in 48 hours. The NTIS numbers are the same as the NASA numbers.
-Danny
∂23-Oct-81 1545 JK proposal
A rough sketch can be found in propos.new[doc,jk].
∂23-Oct-81 1940 Mike Genesereth <CSD.GENESERETH at SU-SCORE> Chandra's proposal
Date: 23 Oct 1981 1937-PDT
From: Mike Genesereth <CSD.GENESERETH at SU-SCORE>
Subject: Chandra's proposal
To: jmc at SU-AI, rpg at SU-AI
I've read Chandra's proposal and placed it in your mailbox in Jacks.
The problem he is working on seems to be an important one for the
nuclear industry. I always assumed that computers played a larger
role in the control room, and his account was eye-opening. I'd
certainly like to see more automation and feel that someone ought to
fund it, perhaps industry, perhaps the NRC.
Unfortunately, I am not too keen on his approach.
(1) I believe the problem to be more difficult than his simple approach
would suggest. I find it implausible that simple cause-effect links
would in all cases be adequate to do prediction. He also ignores the issue
of the time course of a problem except for instantaneous derivatives.
(2) His control scheme is currently in vogue; but, personally, I lean to
explicit meta-rules for control rather than "communicating specialists".
I guess this is mostly a matter of style, though.
(3) He makes a case for "knowledge-based" analysis rather than just
"rule-based" but the only difference I can detect in his account has
to do with control knowledge. I could certainly do all of his examples
using a "rule-based" approach, where by "do" I mean get the same answers.
(4) He suggests in several places that there is some AI content in his
proposal. I can find none except for a demonstration that the
blackboard approach can be applied to yet another domain. If he were
trying to capture design knowledge, figure out how to reason with it,
and/or diagnose from it I'd be more interested. If he were concerned
with generating good explanations for the operator or studying the issues
of interface, I'd be more interested.
His resources are probably adequate. I'm particularly impressed with the
nuclear expertise he has on tap at Ohio.
Nevertheless, I'd rate the proposal pretty negatively. As rpg
remarked it's probably unwise to get AI involved with nuclear at this
point unless we're sure of a victory, and that proposal doesn't make
me particularly confident.
mrg
-------
Thanks for your comments on the Chandrasekaran proposal.
∂24-Oct-81 2000 JMC*
Chris Parkyn
Western Aviation, 258-9378
∂24-Oct-81 2128 Lowell Wood <LLW at S1-A> `If Wishes Were Horses. . .`
Date: 24 Oct 1981 2123-PDT
From: Lowell Wood <LLW at S1-A>
Subject: `If Wishes Were Horses. . .`
To: jmc at SU-AI
CC: LLW at S1-A
∂20-Oct-81 2148 John McCarthy <JMC at SU-AI>
Date: 20 Oct 1981 2147-PDT
From: John McCarthy <JMC at SU-AI>
To: llw at S1-A
Stanford received $79.1 million in gifts last year - an ordinary year.
[John: If our operation had alumni of the quantity and financial
quality as does Stanford, my budgetary proposals would be a lot more
ambitious. "Champagne taste, beer money," they say. Lowell]
∂25-Oct-81 0813 energy at MIT-MC
Date: 25 October 1981 09:53-EDT
From: energy at MIT-MC
Sender: OAF at MIT-MC
To: ENERGY-DSN at MIT-MC
Date: 22 Oct 1981 15:29:59-PDT
From: decvax!duke!unc!smb at Berkeley
In-real-life: Steven M. Bellovin
Location: University of North Carolina at Chapel Hill
Please correct me if I'm wrong, but..... I'm under the impression that
the only half-way practical fusion reactions now being considered involve
both deuterium and tritium. The former can be extracted fairly easily
from water, but the latter is unstable and not found in nature; I believe
it's manufactured by bombarding lithium with neutrons. A working reactor
will supply plenty of neutrons to do the bombardment, but that would make
lithium a limiting factor. So: what are the naturally ocurring ores of
lithium? Given its fairly high chemical activity, how hard is it to ex-
tract, and how much energy does that take? What mad dictator controls
the major known deposits? Will the bombardment have other undesirable
side-effects, such as producing radioactive wastes (which are at the very
least politically unacceptable)?
∂25-Oct-81 0906 Konolige at SRI-AI reply
Date: 25 Oct 1981 0905-PST
From: Konolige at SRI-AI
Subject: reply
To: jmc at SAIL
John, consis.kgk[1,kgk] has my thoughts on your file knowle[f81,jmc].
-------
∂25-Oct-81 1651 ME ftp
There are various alternatives. You can login via FTP to the remote
host; this is done with the USER or LOGIN command (either one), e.g.,
USER ESS,JMC. This will require a password, which you'll be asked
for immediately thereafter.
Or you can tell FTP explicit remote filenames, like
GET *.*←*.FOO[ESS,JMC]
(instead of just GET *.FOO[ESS,JMC]).
The problem is that FTP doesn't know which host you mean the PPN to
apply to. To solve that problem, you can give FTP the mode-setting
command RPPN (for remote PPN) or LPPN (for local PPN). I forget
what the default is, probably LPPN, judging from your problem.
∂25-Oct-81 1656 ME ps
If the SAIL FTP server isn't given a PPN, it has no idea what to use,
so it uses [NET,GUE], which is why you got those error messages.
∂26-Oct-81 0021 RAH fiche
To: JMC at SU-AI, LLW at SU-AI
CC: RAH at SU-AI
Second round fiche
To Harry or Richard:
Can you advise me as to how to obtain the following NASA reports?
I would like both hard copy and fiche if both are available but will
take whichever is available. I can pay costs. I assume the best local
source is Ames. Do you know whom to talk to there? I could go there
on Tuesday or Wednesday.
This fiche is listed in order of importance ( best is first ), and
by NASA acession number, which is how Ames will access it.
N70-17551 : Cargo vehicle using the RL10 engine.
fiche from Ames
N79-32225 thru N79-32229 : These are part of a large report, SP-428, on
space habitats. It was done as a summer study at Ames, so
they may have a hard copy also. If so try to get both, with
priority on fiche.
Stanford gov't documents
N65-16672 : Baseline RL10 specs.
Engineering has fiche
N76-26830,26831,26832,26836,26838,26839 : These are life support papers
which are part of a larger report SP-374 Vol 3. The entire
report starts at fiche N76-26829. It is combo of Russian &
US stuff.
Engineering library
N62-13348 : One way lunar mission.
N78-26740 : Mathematical treatment of closed life support systems.
N80-12735 thru N80-12738 : Guidelines for life support design.
∂26-Oct-81 0850 CG
To: "@LOGIC.DIS[1,CLT]" at SU-AI
Theory of Computation Seminar
Place: Margaret Jacks 252
Time: 1:15 pm, Wednesday, October 28
Bengt Nordstrom of the University of Gothenberg, Sweden, will speak
on his implementation (in LISP) of Per Martin-Lof's theory of types.
Martin-Lof's theory provides a formalization of constructive reasoning
in which proofs and programs are built up using a common collection
of primitives.
∂26-Oct-81 0923 C.S./Math Library <ADMIN.LIBRARY at SU-SCORE> NASA Reports
Date: 26 Oct 1981 0921-PST
From: C.S./Math Library <ADMIN.LIBRARY at SU-SCORE>
Subject: NASA Reports
To: jmc at SU-AI
The Science Department Libraries have an agreement with Ames which allows us
to borrow NASA reports from them in exchange for a coupon which allows them
to borrow articles or books from us. Richard does the ordering and we
usually have good turn around time from Ames. We will check first to make
sure we do not already have the materials. The SP publications I believe
we receive through the Engineering Library. Generally we do get the
microfiche which we of course keep. I'll ask Richard to look into the
hard copy possibilities. Harry
-------
∂26-Oct-81 1005 FFL mail jmc,ffl
To: JMC, FFL
Professor Moses from MIT is anxious to speak with you re letter about
Patrick Winston. 617 253 4601.
∂26-Oct-81 1150 TOB
John:
I agree.
Tom
∂16-Oct-81 2308 JMC Russian interest in American image processing
The news story CODE.NS[F81,JMC] treats, among other matters, Russian
efforts to steal image processing technology in this area. I think
we should make sure that our reports on vision and robotics are not
sent to East bloc addresses. If there were no better reason, it
is likely to cause us trouble. One mechanism would be to annunce
the reports as CS reports, but include in the announcement that
the reports are available only directly from your group. Certain
requests could then be ignored. Requests for published papers
should almost certainly be honored, however, since the Russians
can get them anyway, and also their information filter occurs
when something is proposed for publication. During the whole
of our acquaintance, Ershov never sent me anything unpublished,
although in good years, he did some rather local, though still
censored, publication.
∂26-Oct-81 1456 Irmgild Schack <CSD.IRMGILD at SU-SCORE> Advisory Committee (Stanford)
Date: 26 Oct 1981 1451-PST
From: Irmgild Schack <CSD.IRMGILD at SU-SCORE>
Subject: Advisory Committee (Stanford)
To: CSD-Faculty: ;
The President's Advisory Committee on Computer Science and Computing will
be meeting in the new year. It is necessary to make several new appointments
and I would appreciate your recommendations. Last year's Advisory Committee
was made up of the following persons:
W. Zuendt -- Wells Fargo
W. Arms -- Dartmouth
J. Bell -- H-P
R. Cypser -- IBM
P. Ely -- H-P
B. Kernighan - Bell
J. Lederberg - Rockefeller
J. Licklider - M.I.T.
S. Lukasik -- FCC
J. McCredie -- EDUCOM
W. Miller -- SRI
R. Scott -- Harvard
R. Spinrad -- Xerox
R. Taylor -- Xerox
GENE
-------
∂26-Oct-81 1528 FFL
To: JMC, FFL
Karl Cohen asks that you call him. He will be at one of these numbers:
858-0565 or 856-6930. The prefix for both is 9.
∂27-Oct-81 0841 JK proposal
The file propos.new[doc,jk] contains more or less all that I want to say
for the proposal. Let me know if you have any additions or comments.
∂27-Oct-81 1007 FFL
To: JMC, FFL
mail jmc,ffl
mail jmc,ffl
mail jmc,ffl
mail jmc,ffl
mail jmc,ffl
mail jmc,ffl
Call from Ann Mayo in Graduate Program Office, asking if you would be
willing to chair an oral examination. Please call her at 7-3056.
∂27-Oct-81 1010 FFL
To: JMC, FFL
Marlene Dann of a Technology magazine called. Is doing research on profile
of leaders in computer field and would like to talk to you. Please call
at 212 943 9020.
∂27-Oct-81 1159 Irmgild Schack <CSD.IRMGILD at SU-SCORE> Faculty Lunch
Date: 27 Oct 1981 1153-PST
From: Irmgild Schack <CSD.IRMGILD at SU-SCORE>
Subject: Faculty Lunch
To: CSD-Faculty: ;
This is just a reminder that the faculty lunch is, as usual, today
at the Boys Town conference room.
IRMGILD
-------
∂27-Oct-81 1406 RPG Re: Prolog
∂27-Oct-81 0831 David Warren <WARREN at SRI-AI> Re: Prolog
Date: 27 Oct 1981 0826-PST
From: David Warren <WARREN at SRI-AI>
Subject: Re: Prolog
To: RPG at SU-AI
cc: WARREN at SRI-AI
In-Reply-To: Your message of 27-Oct-81 0112-PST
Fine. All the UUO calls are in modules written in Macro.
I'll send you the sources of these plus RELs of all the rest.
It should be relatively easy for you to modify the Macro sources,
re-assemble them, and then re-link the whole system. I was afraid
you were needing to touch the parts of the system which are written
in Prolog, which involves knowing all sorts of intricate details
about how the system is constructed.
Is there an area I can FTP the files to? --David.
-------
John, where would you like the PROLOG sources etc to go? David Warrenhas the
go ahead for us to FTP and hack.
-rpg-
The sources should go in 1,3 or whereever other language sources are kept.
∂27-Oct-81 1631 RPG
∂27-Oct-81 1609 JMC
The sources should go in 1,3 or whereever other language sources are kept.
Well, it depends. There is LSP for Lisp, AIL for SAIL. 1,3 is for
DMP files. Should a new files-only area be created? Who will pay for it?
-rpg-
Well then, create a new file area and propose that it be treated like
LISP and SAIL, i.e. as a system overhead. If someone balks at this,
then I'll guarantee from DARPA money that it is paid for and will argue
later.
∂27-Oct-81 2028 Gene Golub <CSD.GOLUB at SU-SCORE> Gripes
Date: 27 Oct 1981 1737-PST
From: Gene Golub <CSD.GOLUB at SU-SCORE>
Subject: Gripes
To: admin.gorin at SU-SCORE
cc: jmc at SU-AI, CSD.GOLUB at SU-SCORE
Ralph!
Here are some of the (edited) comments of Tom Dietterich. GENE
27-Oct-81 00:38:08-PST,00000002819;000000000001
Date: 27 Oct 1981 0038-PST
From: Tom Dietterich <CSD.DIETTERICH>
Subject: My private gripes
To: csd.golub
cc: CSD.DIETTERICH
Gene,
As I was putting together the proposed agenda for Wednesday, it
occurred to me that I have a few private gripes of my own. I raise
them with you because I don't think they are appropriate for the "town
meeting."
Among the policies that I personally question are:
1. The MRC: disk pack.
As an undergraduate, I held a position very similar to that which Mark
holds here. I maintained and modified the operating system for a
large computer (the Xerox Sigma 9). I would NEVER have dreamed of
consuming an entire disk pack with the system sources and my own
files!! Where I worked, we kept all of the system sources on tape,
with hard copy and micro-fiche for reference. Given the rather severe
disk shortage at SCORE, it seems inappropriate for the system support
staff to be consuming an entire spindle. In a bboard "debate" last
year, this policy was justified by claiming that it was important to
keep one spindle in reserve in case one of the other drives broke. I
do not agree with that justification, and I suspect many other people
share my views. I am willing to trade extended down-time (when a disk
drive breaks) for extra disk space.
2. Lack of an ARCHIVE system.
At SUMEX and at SAIL, archive systems have been installed, and they
work quite well. The disks at SCORE, however, are filled with
seldom-used files that could cheaply be archived. Furthermore, it is
my impression that archive software already exists for TOPS-20.
3. No user access to SEND *, and similar inflexible policies.
Many system commands remain privileged, despite the fact that they
pose no danger to the system or to other users.
-------
-------
∂27-Oct-81 2233 Mike Farmwald <PMF at S1-A>
Date: 27 Oct 1981 2233-PST
From: Mike Farmwald <PMF at S1-A>
To: jmc at SU-AI
∂27-Oct-81 2158 John McCarthy <JMC at SU-AI>
Date: 27 Oct 1981 2157-PST
From: John McCarthy <JMC at SU-AI>
To: pmf at S1-A
What did you think of Schwartz's talk and their scheme?
I like the new idea, i.e. to have a network that can handle multiple
FetchAndAdd operations.
However, I still think that they have the same problems that most
of the "large" (i.e. >1000) processors) have.
1. Economics are not clear.
A high performance micro can do an arithmetic operation in
about 4 us, while a Cray (or an S-1) can do on in about 6.25 ns (12.5 ns).
Thus one needs roughly 1000 such microprocessors to equal the peak performance
of a Cray (or about 500 for a uniprocessor Mark IIA).
The Mark IIA has 25000 chips. Assuming a constant cost per chip, this means
that the network and processors have to fit in less that 25000 chips, or
50 chips per processor. This seems difficult. As technology improves, this
will become easy, but the number of chips in a Mark IIA level machine
will also decrease.
2. The ``ultracomputer'' (and other large multimicroprocessor systems) can't
handle recurrences and programs with limited parallelism at all well. There
is no way around a single fast machine for these.
3. I don't really understand how to use so many machines for many applications.
(but maybe that's because I haven't had one to use.)
I admit that none of these are completely clearcut, and in fact in 5 or 10 years
I may change my mind. Not soon, however. In the meantime, I like to see good
people working on the problems.
∂28-Oct-81 0859 ullman@Diablo (SuNet) There follows a version of the budget that reflects the discussions
Date: 28 Oct 1981 08:52:43-PST
From: ullman at Diablo
To: csl.jlh@score, engelmore@sumex, jmc@sail
Subject: There follows a version of the budget that reflects the discussions
of Monday. As Betty informs me that there is no tax on ARPA purchased
equipment, i added 125K for the Foonly, which just brings us to 3M.
OK to distribute to the committee?
∂28-Oct-81 0901 ullman@Diablo (SuNet)
Date: 28 Oct 1981 08:53:23-PST
From: ullman at Diablo
To: csl.jlh@score, engelmore@sumex, jmc@sail
What follows is the tentative proposal formulated by
representatives of the three major areas of ARPA activity,
Bob Engelmore, John Hennessy, and John McCarthy, meeting
with me. The budget is under 3 million, and includes a
small amount for maintenence. The assumption is that once
our ARPA proposals have a chance to adjust to the costs of
the new equipment, maintenence will be covered by the con-
tracts of those who use the equipment. The major changes
are:
1. The cost of the ethernet was raised by 200K, to reflect
our discovery that such development was seriously
affecting the cost of CSD-CF service and that we cannot
continue to operate this way. Most of this extra money
will go for development of the facility, rather than
equipment per se.
2. The amount devoted to purchase of SUN terminals was
reduced to cover only the 25 requested by the systems
group and 10 more for the general use of the other con-
tractors. We felt that there was no way we could jus-
tify more as being ARPA-related.
3. We have budgeted only half the estimated cost of an S-
1. It is hoped that this amount will cover the cost of
that portion of the beast earmarked for LISP cycles,
with the balance supported in one of a number of ways
and intended for general, probably non-ARPA, work.
4. We included the cost of all 8 Gbytes of storage in the
file server.
1982 1983 1984 HPP SAIL SYST GENL
VAX 250 250
S1 325 325
"Dolphins" 480 420 720 180
SUN 200 70 200 70
McSUN 100 100
Foonly 225 125 350
Ethernet 100 100 100 300
ERL Local Net 30 30 60
Printers 30 60 30 30 60 30
Local file stores 50 50 100
SUVAXen 180 180
Maintenence 45 15 5 10 15
TOTALS 1090 960 950 765 510 760 965
GRAND TOTAL = $3,000,000
October 27, 1981
I think the budget is ready for distribution.
∂28-Oct-81 1149 FFL
To: JMC, FFL
Please call Murry Day at Lawrence Livermore. 415 422 0523.
We have to make some promises about deliverables.
E.g. EKL manual and associated report, report on CG's vision work perhaps
with a program.
Interim report on Advice Taker.
Let me know your ideas, preferably in the form of sentences.
∂28-Oct-81 1341 Bob Engelmore <CSD.ENGELMORE at SU-SCORE> Deliverables on ARPA contract
Date: 28 Oct 1981 1338-PST
From: Bob Engelmore <CSD.ENGELMORE at SU-SCORE>
Subject: Deliverables on ARPA contract
To: JMC at SU-AI, DCL at SU-AI, TOB at SU-AI, ZM at SU-AI,
Wiederhold at SUMEX-AIM
cc: csd.bscott at SU-SCORE, Waldinger at SRI-KL
John Machado (the contract monitor) called me today and asked that I
coordinate the following with you. It seems that the statement of work
for your new ARPA contract contains no deliverables, and without any
he can't proceed further. A deliverable would be an interim technical
report on a specific topic, or a documented program that could be
distributed to and used by others. There should be from 1 to 4 deliver-
able items per task (i.e. per principal investigator in this case).
Each item should have a delivery date associated with it -- be very
conservative with these dates, as I suspect Machado will hold you to
them pretty strictly.
It should be relatively easy to generate a list of deliverables from
the text of your proposal, but each of you should do it as soon as
possible. It might be best if you sent the list to me and I'll
package them into one note to Machado.
Bob
-------
∂28-Oct-81 1522 SIS Colloquium Notice - Week of November 2 - 6, 1981
To: "@COLLOQ.DIS[INF,CSD]" at SU-AI
Date Place Person
Day Event From
Time Title
-------------------------------------------------------------------------------
11/2/81 Math 380C Keith Miller
Monday Numerical Analysis U.C. Berkeley
4:15 p.m. Seminar ``Alternate Modes to Control The Nodes in Moving
Finite Element Methods''.
11/3/81 Bldg. 200-30 Carl Hewitt
Tuesday Concurrency and AI MIT
2:30 p.m. ``Further Development of Scientific Community
Metaphor''.
11/3/81 Jordan 041 Susan Graham
Tuesday Computer Science UC Berkeley
4:15 p.m. Colloquium ``A Progress Report On Table-Driven Code
Generation''.
11/4/81 Terman 153 Bob Kridle
Wednesday Computer Systems UC Berkeley
4:15 p.m. Laboratory Seminar To be announced
11/6/81 MJ301 Joaquin Miller
Friday Database Research
3:15 p.m. Seminar ``Stealing Ideas by Query Title''.
∂28-Oct-81 2215 Marvin Minsky <MINSKY at MIT-AI> AAAI Corporate Fund-Raising Letter
Date: 28 October 1981 23:27-EST
From: Marvin Minsky <MINSKY at MIT-AI>
Subject: AAAI Corporate Fund-Raising Letter
To: MINSKY at MIT-AI, GJS at MIT-AI, PHW at MIT-AI, sacerdoti at SRI-KL,
rick at RAND-AI, jmc at SU-AI, tw at SU-AI, reddy at CMU-10B,
simon at CMU-10A, newell at CMU-10A, athompson at USC-ECL,
bobrow at PARC-MAXC, winograd at PARC-MAXC, dwaltz at BBND,
webber at BBND, woods at BBND, erman at USC-ISIF,
balzer at USC-ISIF, buchanan at SUMEX-AIM, engelmore at SUMEX-AIM,
feigenbaum at SUMEX-AIM, feldman at SUMEX-AIM,
aaai-office at SUMEX-AIM, bledsoe at UTEXAS-20, grosz at SRI-AI,
nilsson at SRI-AI, walker at SRI-AI
I plan to send AAAI letter to companies, asking for contributions
to support the AAAI. It will suggest "Sustaining", "Associate"
and "Contributing" corporate menberships, for $5000, 2000, and 1000.
I sent a draft to some of you. If you want to see final version,
just ask and I'll net it over -- or look in AI:MINSKY: BEG LETTER.
The only way this can work is by "personalizing" it. So, before
you forget:
PLEASE SEND ME A MESSAGE WITH ONE OR MORE OF
<Name of Company><Name of Vice-president-type><PERMISSION>
whom you know to be a plausible target for the letter. Then I will
personalize the letter. <permission> is a Yes or NO on
whether I can use YOUR name is some sentence like
"<name name> suggested that you might be in a
position to persuade your company to contribute to this
important professional organization"
Then I'll follow the letters with friendly, plaintive phone calls.
-- marvin
∂29-Oct-81 0106 MINSKY@AI An idea.
Date: 29 October 1981 04:04-EST
From: MINSKY@AI
Sender: MINSKY at MIT-MC
Subject: An idea.
To: minsky at MIT-AI, gjs at MIT-AI, phw at MIT-AI, jmc at SU-AI,
reddy at CMU-10B, simon at CMU-10A, newell at CMU-10A,
winograd at PARC-MAXC, dwaltz at BBND, webber at BBND,
balzer at USC-ISIF, feigenbaum at SUMEX-AIM, feldman at SUMEX-AIM,
aaai-office at SUMEX-AIM, bledsoe at UTEXAS-20, nilsson at SRI-AI,
walker at SRI-AI
If you think this letter would do any good, please suggest any mods.
If not, say so.
The purpose is twofold:
1. Might do AI and Computer Science some good.
2. Might have slight educational effect about basic
research and the "refurbishment" crisis.
* * * * * * * * * * * * * * * * * * * * * * * * * * * * *
Mr. David Stockman
Budget Director
Washington D.C.
Dear Mr. Stockman,
You should know about a small problem that could have a large
effect on the U.S. economy. Today, the U.S. is a world leader in
Computer Technology because we the most advanced in Computer
Science in general, and in Computer Design. But we are losing our
advantage in producing Hardware Components and already behind in
Computer Memory components and (probably) already lagging Japan
in Robotics.
Our most important advantage, still, is in Software. This letter
is to suggest that we must take measures lest we lose that
all-important primacy.
Japan has just announced a vigorous program to take the lead in
advanced computer systems over the next few years -- in a
national program called the "Fifth Generation Computer Project".
I have met some of the people involved and believe that they may
succeed unless we meet the challenge. If this happens, the U.S.
may lose its lead both in Hardware and in Software.
WHY WE MAY LOSE: We still have more and better trained
researchers. But our best workers are now wasting much of their
irreplaceable talent because they cannot afford to buy the most
modern equipment. So either they use obsolete equipment, or
waste time building better equipment themselves. I know of many
of our best people setting three or four years aside to build
such tools. The result, as I see it: we are progressing at about
half the rate that we should be.
EXAMPLE: RESEARCH COMPUTERS. The best tools for advanced
research on robotics and intelligent computers are certain new
computers developed since around 1976. Developed in three or
four Artificial Intelligence Laboratories, these are now in
commercial production. Each costs around $100,000 and, efficient
research toward the "fifth generation computer" would use one
such machine for each two workers. But no US laboratory can
afford anything like that. Without these machines, it takes at
least twice as long to develop the competitive new systems.
Rumor: the Japanese project plans to distribute such machine to
all the participants.
ROBOTIC EQUIPMENT: Though "robotics" is very popular this year,
I bet you didn't know that there are only three or four research
laboratories in the US doing serious work in that field -- fewer
than there were in 1970! Only one of those laboratories has been
able to purchase modern industrial robotic equipment. The others
either use unreliable early prototypes -- or have abandoned the
aggressive pursuit of the subject because they cannot purchase
adequate capital equipment. The apparent "boom" in robotics is
based on research between '65 and '75. I see much too little
such work now to provide a good base for competition beyond 1985.
ARTIFICIAL INTELLIGENCE: This is the field that produced much of
our current power in robotics, advanced software, office
automation and "expert" computer systems. But the lack of modern
equipment is drawing most new people away from research into new
companies, where few of them can do basic research. Unless we
can correct this, the US will risk a rapid slump in the high
technology software and harware industries. The balance of trade
could reverse in just a few years.
TRAINING: Because we are now training new students on old
equipment, this means that our industrial computer establishment
may sooner become obsolescent, too.
COST OF REMEDY: I estimate that US research in Computer Science,
Artificial Intelligence, and Robotics could be revitalized if some
500 workers were supplied with with modern equipment. This would
cost around 50 million dollars -- not including overhead.
No there is no Agency now able to consider such a request. I see
this as a silent but extremely serious national emergency. I
certainly would not raise this issue it at your level but, as
spokesman for the US professionals in this field, I see nowhere
else to turn.
If you have any suggestions or questions, the resources of the
AAAI are at your service.
Marvin Minsky,
President, AAAI
Donner Professor of Science, M.I.T.
Member, National Academy of Science
∂29-Oct-81 0855 David R. Cheriton <CSL.DRC at SU-SCORE> Cuthbert Hurd
Date: 29 Oct 1981 0823-PST
From: David R. Cheriton <CSL.DRC at SU-SCORE>
Subject: Cuthbert Hurd
To: jmc at SU-AI
I have phoned Cuthbert Hurd and he has agreed to talk on Jan. 26th, 1982.
Thanks very much for the suggestion.
-------
∂29-Oct-81 0858 Don Walker <WALKER at SRI-AI> Re: An idea.
Date: 29 Oct 1981 0838-PST
From: Don Walker <WALKER at SRI-AI>
Subject: Re: An idea.
To: MINSKY at MIT-AI, gjs at MIT-AI, phw at MIT-AI, jmc at SU-AI, reddy at CMU-10B,
simon at CMU-10A, newell at CMU-10A, winograd at PARC-MAXC, dwaltz at BBND,
webber at BBND, balzer at USC-ISIF, feigenbaum at SUMEX-AIM, feldman at SUMEX-AIM,
aaai-office at SUMEX-AIM, bledsoe at UTEXAS-20, nilsson at SRI-AI
cc: WALKER at SRI-AI
In-Reply-To: Your message of 29-Oct-81 0104-PST
Marvin,
I like the idea and the message; apart from a missing word here and there
which you certainly can pick up and the phrase "I'll bet" which could be
replaced, I have no recommendations for modifying it. It does seem to me,
though, that copies should be sent to the President of the Academy, to
Reagan's science advisor, to the chairmen of the House and Senate "science"
committees (and perhaps to their senior staffers), and, perhaps, to the
heads of NSF and ARPA as well as to other people in what might be considered
"key" positions. The function of Council members in responding to this
message might well be that of identifying other potential recipients of
copies.
Don
-------
∂29-Oct-81 0925 Aaai-Office at SUMEX-AIM Letter
Date: 29 Oct 1981 0858-PST
From: Aaai-Office at SUMEX-AIM
Subject: Letter
To: Minsky at MIT-AI, GHS at MIT-AI, PHW at MIT-AI, JMC at SU-AI,
REDDY at CMU-10B, SIMON at CMU-10A, NEWELL at CMU-10A,
WINOGRAD at PARC-MAXC, DWALTS at BBND, WEBBER at BBND, BALZER at USC-ISIF,
FEIGENBAUM at SUMEX-AIM, FELDMAN at SUMEX-AIM, BLEDSOE at UTEXAS-20,
NILSSON at SRI-AI, WALKER at SRI-AI, AAAI-OFFICE at SUMEX-AIM
cc: AAAI-OFFICE at SUMEX-AIM
Marvin:
Your Stockman letter deserves wider distribution. I would suggest
sending the letter to the various House and Senate committees and
sub-committees involved with science and technology. I would also
distribute the letter to the various national science associations,
to the White House and to the leading U.S. science publications.
Approximately a week later, I would send copies to the New York Times,
the Washington Post, the Los Angeles Times and the wire services
(I can provide names of individuals here). If you'd like, I can
help you with the draft of a press release identiofying your
concern and the concern of the rest of the AAAI leadership.
I'm not sure if you intended to go this far, Marvin, but it seems to
me that if you plan to send the letter -- and, of course, you
should -- you should be able to get more mileage out of it than
merely dropping it into Stockman's sizeable "IN" basket.
Lou Robinson
-------
∂29-Oct-81 0927 Nilsson at SRI-AI Re: An idea.
Date: 29 Oct 1981 0926-PST
From: Nilsson at SRI-AI
Subject: Re: An idea.
To: MINSKY at MIT-MC, minsky at MIT-AI, gjs at MIT-AI,
To: phw at MIT-AI, jmc at SU-AI, reddy at CMU-10B,
To: simon at CMU-10A, newell at CMU-10A, winograd at PARC-MAXC,
To: dwaltz at BBND, webber at BBND, balzer at USC-ISIF,
To: feigenbaum at SUMEX-AIM, feldman at SUMEX-AIM,
To: aaai-office at SUMEX-AIM, bledsoe at UTEXAS-20,
To: nilsson at SRI-AI, walker at SRI-AI
cc: NILSSON
Marvin, I like your letter very much. I also talked with many
Japanese about their 5th generation project on my recent trip to
Japan. I think they mean business-- just like they did on their
PIPS projects of a few years ago.
I don't have substantive comments about the letter. It needs a bit
of proof reading to correct typos.
You ought to send the letter also to Keyworth, the new science advisor.
Also, how about using it to lobby appropriate congressional types?
If you decide that the letter would carry more weight by having several
people sign it, I would be glad to join. --Nils
-------
∂29-Oct-81 1150 Wiederhold at SUMEX-AIM Re: note from susan
Date: 29 Oct 1981 1027-PST
From: Wiederhold at SUMEX-AIM
Subject: Re: note from susan
To: PN at SU-AI
cc: jmc at SU-AI, rpg at SU-AI, csd.hill at SU-SCORE
In response to your message sent 29 Oct 1981 1009-PST
At this point I have to ask Susan to remove the account unilaterally from s-1.
I have not heard anything from anybody to indicate the contrary. Gio
-------
∂29-Oct-81 1211 Marvin Minsky <MINSKY at MIT-MC> PROGRAM COMMITTEE CHAIRMAN
Date: 29 October 1981 14:49-EST
From: Marvin Minsky <MINSKY at MIT-MC>
Subject: PROGRAM COMMITTEE CHAIRMAN
To: minsky at MIT-AI, gjs at MIT-AI, phw at MIT-AI, jmc at SU-AI,
reddy at CMU-10B, simon at CMU-10A, newell at CMU-10A,
winograd at PARC-MAXC, dwaltz at BBND, webber at BBND,
balzer at USC-ISIF, feigenbaum at SUMEX-AIM, feldman at SUMEX-AIM,
aaai-office at SUMEX-AIM, bledsoe at UTEXAS-20, nilsson at SRI-AI,
walker at SRI-AI
DAVID WALTZ HAS AGREED TO BE PROGRAM COMMITTEE CHAIRMAN
PROBLEM: ALTHOUGH HE IS DWALTZ@BBND, HIS ARPANET LINK IS
TENUOUS, SO HE DOES NOT EXPECT TO READ HIS NET-MAIL EVERY DAY.
∂29-Oct-81 1212 RPG
To: JMC at SU-AI
CC: GIO at SU-AI
∂29-Oct-81 1154 JMC your usage
To: RPG at SU-AI
CC: GIO at SU-AI
Please estimate the fractions of your usage that are:
a. general maintenance of MACLSIP - an overhead of SAIL
b. chargable to the work you are doing for Livermore. Does
all present Livermore work go through Gio's contract?
c. chargable to the formal reasoning group
All your usage should be charged to one of these three, and it is
just a question of dividing up the aliquots.
If you mean disk usage:
a: 75%
b: 10%
c: 15%
If you mean CPU usage:
a: 50%
b: 25%
c: 25%
I anticipate that next quarter the c: percentage will jump quite a bit.
-rpg-
∂29-Oct-81 1220 RPG Prolog contd.
∂29-Oct-81 0927 David Warren <WARREN at SRI-AI> Prolog contd.
Date: 29 Oct 1981 0856-PST
From: David Warren <WARREN at SRI-AI>
Subject: Prolog contd.
To: rpg at SU-AI
Finally got our FTP to access [pro,log].
Have so far transferred core image and documentation
for DEC-20 Prolog. PROLOG.EXE[pro,log] almost certainly
won't run in situ, but maybe you can transfer it to the -20
John McCarthy was using. It should be an improvement on the
version he already has. Soon, when I have time, I'll dig out
and transfer the MACs and RELs for DEC-10 Prolog. --David.
-------
Here's our progress on Prolog: Slow. We had some trouble with FTP
and the cache problems didn't help much.
-rpg-
∂29-Oct-81 1336 FFL
To: JMC, FFL
John Baker from Kent State called. He reported the following information
on schedules from SFO to Clevelan.d:
Lv SFO 1:50 p.m. and Ar Cleveland at 9:10 p.m.
or
Lv SFO 7 a.m. and Ar. Cleveland at 4:31 p.m.
The latter flight has a layover in Chicago. Both flights will be met to take
you the one hour drive to Kent State.
Mr. Baker will call you again. His number is 216 672 2417, or you may leave
a message at 216 672 2430.
Akron Kent airport to Pittsburgh
I decided to lecture at Kent State. They say there is a 1:50 pm flight
on Sunday that gets me to Cleveland at 9:10 pm. That would be suitable.
I would go from Akron-Kent airport to Pittsburgh at 6:10 pm. Please
make reservations accordingly.
∂29-Oct-81 1620 Grosz at SRI-AI tinlunch reading
Date: 29 Oct 1981 1621-PST
From: Grosz at SRI-AI
Subject: tinlunch reading
To: tlgrp:
The reading for next week is Brian Smith's "A Commentary on Allen
Newell's 'Symbols and Knowledge in Cognitive Science'". Copies are on
my file cabinet. The paper comments on two of Newell's papers: the
paper from his Cognitive Science talk at the San Diego conference
which appeared in the CS Journal (1980) and his AAAI talk which was in
the latest issue of AI Magazine. Of these, the CSJ paper is more
important. I assumed that almost everyone had copies of the AI Mag so
did not make any copies of that. A small number of copies of the CSJ
paper are on the file cabinet, since I thought most people would have
copies of that too. Brian will be here for the discussion.
-------
∂29-Oct-81 1911 RPG Meeting
To: csd.ullman at SU-SCORE, JMC at SU-AI
Is there a meeting friday? I will assume so.
-rpg-
∂30-Oct-81 0652 JK
Here's another proof of Wilson's theorem:
(1) By the binomial theorem Exp(x+y,p)=Exp(x)+Exp(y)
(2) By induction from (1), Exp(x,p)=x
(3) Hence Exp(x,p-1)=1 for x≠1
(4) So the poly Exp(x,p-1)-1 has distinct p-1 roots 1,2,...,p-1
Factoring, get (since the leading coefficient is 1)
Exp(x,p-1)-1 = (x-1)(x-2)....(x-p-1).
Set x=0.
This proof requires knowledge of behaviour of polynomials and factorization
in a field - an interesting challenge.
∂30-Oct-81 1103 FFL
To: JMC, FFL
Walter Blum from the SF Examiner would like you to call him, please.
415 777 7916. I think he is writing an article on AI for the California
Living section of the paper.
∂30-Oct-81 1534 JMM ekl at lots
To: JMC, JK
CC: JMM
EKL has been transported to Lots. The relevant .fasl files etc are in the
directory ps:<ekl> with password magic . The suspended core image is the file
ekl.exe in that directory.
∂30-Oct-81 1538 FFL
To: JMC, FFL
Your schedules for November l5-l8 are in your calendar file. Pls. note
that your flight on Nov. l5 leavesat 2:15 p.m., not at 1:50 as Baker of
Kent State suggested.
∂30-Oct-81 1657 Marvin Minsky <MINSKY at MIT-MC>
Date: 30 October 1981 19:55-EST
From: Marvin Minsky <MINSKY at MIT-MC>
To: sacerdoti at SRI-KL, minsky at MIT-AI, gjs at MIT-AI, phw at MIT-AI,
rick at RAND-AI, jmc at SU-AI, tw at SU-AI, reddy at CMU-10B,
simon at CMU-10A, newell at CMU-10A, athompson at USC-ECL,
bobrow at PARC-MAXC, winograd at PARC-MAXC, dwaltz at BBND,
webber at BBND, woods at BBND, erman at USC-ISIF,
balzer at USC-ISIF, buchanan at SUMEX-AIM, engelmore at SUMEX-AIM,
feigenbaum at SUMEX-AIM, feldman at SUMEX-AIM,
aaai-office at SUMEX-AIM, bledsoe at UTEXAS-20, grosz at SRI-AI,
nilsson at SRI-AI, walker at SRI-AI
Regret that David Barstow declined to be Tutorial Chairman.
Quick. Suggest someone. Or Robinson will kill me.
∂30-Oct-81 2013 Keith A. Lantz <CSL.LANTZ at SU-SCORE> New Systems References for the Comprehensive
Date: 30 Oct 1981 1726-PST
From: Keith A. Lantz <CSL.LANTZ at SU-SCORE>
Subject: New Systems References for the Comprehensive
To: bboard at SU-SCORE
Remailed-date: 30 Oct 1981 1731-PST
Remailed-from: Keith A. Lantz <CSL.LANTZ at SU-SCORE>
Remailed-to: CSD-Faculty: ;
The systems faculty has revised the hardware and software portion of the
comprehensive reading list. These revisions consolidate the previous
references into 6 widely used textbooks. That is, they cover the same
material in a more orderly fashion. These revisions, if acceptable,
will be in effect for at least one year (i.e., the upcoming winter and
spring comprehensives).
However, because the complete reading list has already been "released",
we cannot force the new list on you students. We can promise that care
will be taken to compose questions which can be answered based either on
the old reading list or on these revisions. (Obviously, if you haven't
started studying (!) you should use the new references.) Nevertheless,
if you disagree with this approach please voice your disagreement soon.
Herewith the new reading list:
Hardware Systems
Courses: 111 (see Software Systems), 112, 311
1. J.P. Hayes. Computer Architecture and Organization. McGraw-Hill,
1978.
Software Systems
Courses: 107, 111, 142, 143, 246A
1. A.V. Aho and J.D. Ullman. Principles of Compiler Design. Addison-
Wesley, 1977.
2. O.-J. Dahl, E.W. Dijkstra, and C.A.R. Hoare. Structured
Programming. Academic Press, 1972.
3. T.W. Pratt. Programming Languages: Design and Implementation.
Prentice-Hall, 1975.
4. A.C. Shaw. The Logical Design of Operating Systems. Prentice-Hall,
1974.
5. J.F. Wakerly. Microcomputer Architecture and Programming. John
Wiley & Sons, 1981.
-------
∂31-Oct-81 0848 Engelmore at SUMEX-AIM Reminder: Deliverables
Date: 31 Oct 1981 0839-PST
From: Engelmore at SUMEX-AIM
Subject: Reminder: Deliverables
To: jmc at SU-AI, tob at SU-AI, waldinger at SRI-AI
I know it's an annoyance, but send me a list of deliverables no later than
Tuesday noon. --Bob
-------
∂31-Oct-81 1117 CLT Guarneri
We have tickets for the following concerts this coming week.
*Wed. 4 Nov. 20:00 Guarneri (Dink)
*Fri. 6 Nov. 20:00 Guarneri (Dink)
*Sun. 8 Nov. 14:30 Guarneri (Dink)
∂31-Oct-81 1632 Jeffrey D. Ullman <CSD.ULLMAN at SU-SCORE> ARPA proposal
Date: 31 Oct 1981 1627-PST
From: Jeffrey D. Ullman <CSD.ULLMAN at SU-SCORE>
Subject: ARPA proposal
To: CSD-Faculty: ;
We are in the last stages of formulating a proposal to ARPA for
$3M over 3 years for hardware support of arpa contracts. As part of
that proposal, we shall need the resumes of all arpa contractors.
If you are one, would you please send me a (hard) copy of your resume.
If you have an up-to-date one on file with Gene, then you may ignore
this request.
-------
∂01-Nov-81 1737 REM
Did you send me a message to my console a few days ago when I was running
a binary downloader? The downloader hung and upon inspection I saw part of
a message about restoring CRU1 and CRU3/SPINDL, but the sender's ppn was lost
and much of the message was also lost.
It wasn't me who sent the message.
∂01-Nov-81 2342 TOB computing
John
I would like to resume our discussions about computing
for AI and how to integrate with others in the department.
Tom
∂01-Nov-81 2347 TOB deliverables
To: JMC, DCL
I have finished mine.
∂02-Nov-81 0941 NAN p message
Mr. Ron Palocci's secretary called on Nov. 2 at approx 9:45 a.m. and left
this message: He would like to know if you'll be able to attend the III
planning committee meeting in Culver city on Nov 17 starting late in the morning
and running the rest of the day. He would like to know if you can attend this
meeting with himself and Joe Walsh. the phone no. is: 213 391-7211 ext. 2880.
nan
∂02-Nov-81 1000 JMC*
Call John Gill about Jacobo Bulaevsky.
∂02-Nov-81 1118 NAN p message
please call Dennis from Oak Tree Mazda pertaining to your car. the phone no.
is 247-2212. This call came in on Nov. 2 at 11:00.
nan
∂02-Nov-81 1122 DCL
To: TOB, JMC
∂01-Nov-81 2347 TOB deliverables
To: JMC, DCL
I have finished mine.
REPLY:
Well, I finished mine on Oct 30th.; I sent them to Bob Engelmore.
I assume they will be collated and forwarded by Bob.
- David
∂02-Nov-81 1327 FFL
To: JMC, FFL
Please call Mr. Paolucci at 213 39l 7211, X2880.
∂02-Nov-81 1552 CG
To: JMC, CG
Here are three sentences on "deliverables".
Chris Goad has recently completed tests of the practical effectiveness of
his methods for the automatic generation of special purpose algorithms.
The tests involved the generation of algorithms for hidden surface elimination.
The results of the work will be presented in a computer science department
technical report to be completed by February, 1982.
∂02-Nov-81 1559 Denny Brown <CSD.DBROWN at SU-SCORE> Re: grader for cs206
Date: 2 Nov 1981 1554-PST
From: Denny Brown <CSD.DBROWN at SU-SCORE>
Stanford-Phone: (415) 497-2274
Subject: Re: grader for cs206
To: JMC at SU-AI, DPB at SU-AI
In-Reply-To: Your message of 2-Nov-81 1527-PST
Yes, we can pay. Finding the body is not necessarily easy. I assume that
your current TA is 1/2 time. We can go to 1/4 time grader or (equiv.) 10
hrs per week. I suggest your TA announce it on the Bbds and do some word
of mouth. I have nobody who is looking for money and capable of 206.
-Denny
-------
∂02-Nov-81 1625 Irmgild Schack <CSD.IRMGILD at SU-SCORE> Meeting on Thursday, November 5
Date: 2 Nov 1981 1624-PST
From: Irmgild Schack <CSD.IRMGILD at SU-SCORE>
Subject: Meeting on Thursday, November 5
To: tob at SU-AI, baskett at PARC-MAXC, csd.buchanan at SU-SCORE,
or.dantzig at SU-SCORE, csd.feigenbaum at SU-SCORE, rwf at SU-AI,
csd.golub at SU-SCORE, csd.herriot at SU-SCORE, dek at SU-AI, zm at SU-AI,
jmc at SU-AI, csl.crc.ejm at SU-SCORE, csd.wmiller at SU-SCORE,
csd.oliger at SU-SCORE, csd.ullman at SU-SCORE, tw at SU-AI
This is just a reminder that a Tenured Faculty Meeting will take place
at 2:30 p.m. on Thursday, November 5 in MJH220.
There are a number of important issues to discuss. I am particularily
concerned about the teaching load of the faculty. In particular, we do not
have a uniform policy. The past rule was that each faculty member was to
teach six courses unless s/he had outside rsearch support. This rule
is certainly anticipated and I would like to discuss other proposals.
I'm also concerned about our recruiting program. I don't think it has
been sufficiently well organized. Any suggestions you can make will be
greatly appreciated.
See you on Thursday.
GENE
-------
∂03-Nov-81 0958 SIS Correction to CS Colloq. Notice
To: "@COLLOQ.DIS[INF,CSD]" at SU-AI
Correction For Computer Science Colloquium Notice of Nov. 2 - 6, 1981
DATABASE RESEARCH SEMINAR -- Nov. 6th at 3:15p.m. in MJ301 -
The title should have been listed as:
``Stealing Ideas From Query by Example''.
∂03-Nov-81 1021 RPG Courses
To: DPB at SU-AI
CC: JMC at SU-AI, GHG at SU-AI
Since the emphasis in 206 has shifted away from Lisp as a
programming language, perhaps there should be a course about
this created? Note that Mike Genesereth taught a short course
on Lisp programming this quarter. Cordell Green and I once
plannned out a `Lisp Wizardry' course for advanced Lisp programming.
With all the Lisp around here in AI perhaps we should correct the
deficiency?
-rpg-
∂03-Nov-81 1307 FFL
To: JMC, FFL
Please call Eric Chenoweth, YEI, 212 989 0909.
∂03-Nov-81 1414 FFL
To: JMC, FFL
Please call Mimi Scrandis, SE2, in the morning. 212 840 6595.
∂03-Nov-81 1315 FFL
To: JMC, FFL
Marlene Dann of TECHNOLOGY NEW called again. She would like to speak
with you. 212 943 9020.
∂03-Nov-81 1547 Irmgild Schack <CSD.IRMGILD at SU-SCORE> Annual Faculty Reports
Date: 3 Nov 1981 1456-PST
From: Irmgild Schack <CSD.IRMGILD at SU-SCORE>
Subject: Annual Faculty Reports
To: CSD-Faculty: ;
Please send me your Annual Faculty Report and/or a recent copy of your
c.v. by Friday, November 6 if you have not done so. I plan to forward
those I have to the Dean's Office and to the Provost.
GENE
-------
∂03-Nov-81 1605 JK ekl
To: JJW, JMC, JMM, YOM
A new ekl is up with some incompatible changes - most notably
the new re-writer.
∂03-Nov-81 1621 Jay Stewart <CSD.STEWART at SU-SCORE> Yes - I am in your course..
Date: 3 Nov 1981 1616-PST
From: Jay Stewart <CSD.STEWART at SU-SCORE>
Subject: Yes - I am in your course..
To: csd.mccarthy at SU-SCORE
cc: csd.malik at SU-SCORE
The Registrar lost my studylist. I'm clearing up the snafu, but I wanted to
confirm that I am in your course for purposes of LOTS, etc..
Jay Stewart
<J.JAYSTE> @ LOTS
-------
∂03-Nov-81 1629 FFL
To: JMC, FFL
Please call Jack Cates, 321-1225.
∂03-Nov-81 2041 Denny Brown <CSD.DBROWN at SU-SCORE> cs200 lecture
Date: 3 Nov 1981 2039-PST
From: Denny Brown <CSD.DBROWN at SU-SCORE>
Stanford-Phone: (415) 497-2274
Subject: cs200 lecture
To: jmc at SU-AI
Reminder: You are on for the CS200 lecture for this Thursday.
It is at 2:45 in room 320 (Quad). This is Geology corner;
the corner of the quad closest to Earth Sciences. -Denny
-------
Thanks Denny. Had you not reminded me, I would have gone to the
tenured faculty meeting.
∂04-Nov-81 0157 JMC
PROJ.F80[F80,JMC] Possible 206 projects
∂04-Nov-81 1311 FFL
To: JMC, FFL
Mr. Tang called out hs passage home. He would like to know if Stanford could
pay about 680 worth f excess luggage for him?
And he asks me to express his regret that he will not hear your talk at CMU
on the l7th. He will be giving a talk at Yale that day.
∂04-Nov-81 1356 RPG
∂04-Nov-81 1348 JMC sail charges for maclisp maintenance
To: RPG at SU-AI, csd.hill at SU-SCORE, FFL at SU-AI
CC: REG at SU-AI
It was agreed between jmc and reg that a separate account for this
purpose should be created, and rpg's activities in maintaining
maclisp should be charged to it.
Does this mean I have to do something differently when I work?
-rpg-
∂04-Nov-81 1418 Bob Engelmore <CSD.ENGELMORE at SU-SCORE> Deliverables
Date: 4 Nov 1981 1411-PST
From: Bob Engelmore <CSD.ENGELMORE at SU-SCORE>
Subject: Deliverables
To: Mach at USC-ISI
cc: JMC at SU-AI, ZM at SU-AI, DCL at SU-AI, TOB at SU-AI,
Wiederhold at SUMEX-AIM
John,
The following is a list of deliverable items for the five
proposed tasks. I hope this will suffice.
Regards,
Bob
Deliverables for Research in Formal Reasoning
-----------------------------------------------
March 1982
(1) A technical report will be delivered which describes Chris Goad's
new method of hidden line and hidden surface elimination, based on
formal methods for algorithm specialization.
May 1982
(2) A manual for the EKL interactive theorem prover by Jussi Ketonen and
Joe Weening will be delivered.
October 1982
(3) A technical report by John McCarthy on formalisms for representing
knowledge about knowledge will be delivered.
December 1982
(4) A summary technical report will be delivered which compares
approximately twenty Lisp systems in current use. This report will
summarize the results of running a series of common benchmarks on each
of the systems.
Deliverables for Research on Program Synthesis:
-----------------------------------------------
January 1982
(1) A technical report will be delivered which describes a method for
the synthesis of the synchronization component of concurrent programs
from specifications in an extended temporal logic.
July 1982
(2) A technical report will be delivered which describes an operational
interactive system for the synthesis of sequential programs from their
specifications.
Deliverables for Research in Techniques for Program Specification and
Verification
---------------------------------------------------------------------
June 1982
(1) Technical report on Techniques for Specification and Verification of
Encapsulation Constructs in programming languages.
This will be a report giving techniques for specifying Ada packages and
axiomatics proof rules for automated verification of packages.
Application of the techniques to similar constructs in other modern
programming languages will be discussed. The report is a milestone
towards the design of verifiers for secure subsets of Ada.
December 1982
(2) Technical report on Design of ANNA, an annotation language for Ada.
This will be a report detailing the design of an annotation language for
the Dod programming language Ada. It represents the confluence of a
major portion of the research goals in the proposal. The report will
detail extensions to the official Ada reference manual of July 1980
necessary to define Anna.
January 1983
(3) Technical report on a strongly typed language for specifying programs
and concepts.
This report will present a language in which user-defined programming
concepts intended for mechanical reasoning are expressed. This language
is directed towards development of high level description languages for
sophisticated programming environments.
September 1983
(4) Report on the design of an interactive theorem prover.
This will be a preliminary report describing the structure of the redesigned
theorem prover, the language for expressing proof strategies and its
implementation. This project in automated deduction is a milestone goal
directed towards development of automated program analysis tools.
Deliverables for research on Knowledge-based Data Management Systems
--------------------------------------------------------------------
July 1982
(1) A technical report will be delivered on the optimal design of
physical databases using the Relational database model. This report will
present a formal analysis of the algorithms, and demonstrate their
application to practical database design problems.
(2) A technical report will be delivered on performing updates to
databases via Natural Language interaction. This report will present
methods for appropriately interpreting requests expressed in Natural
Language, and will show the practical feasibility of the techniques by a
demonstration program.
July 1983
(3) A documented program, which will aid a database designer in
integrating user data models developed at diverse distributed sites,
will be available for use. A report detailing procedures for performing
this integration will accompany the program.
(4) A technical report will be delivered on the creation, maintenance,
and use of a statistical "database abstract". The database abstract will
contain statistical information on various fields and values in the
underlying database, and will be used by an inferencing program to
efficiently compute answers to queries whose cost would otherwise be
prohibitive. The techniques will be illustrated by a demonstration
program.
(5) A procedure will be implemented and documented for responding to
queries that require access to remote information at sites in a network
of distributed databases, without having complete knowledge about the
information available.
Deliverables for Research in Image Understanding
------------------------------------------------
December 1982
(1) ACRONYM will be presented for delivery to the ARPA-DMA Image
Understanding Testbed. ACRONYM will have been tested on a VAX at
Stanford under the UNIX operating system, ready to be transported to the
IU Testbed which runs on a VAX under the UNIX-compatible system EUNICE.
(2) A manual for ACRONYM will be delivered at the same time. The manual
will contain sections for two types of use: the first is applications
use, assuming some sophistication of the user, but not expert status;
the second use is system development by experts.
July 1983
(3) An interim technical report will be delivered, reporting on
mechanisms for geometric reasoning and their relevance to systems for
defense applications. Evaluations will be made concerning the
performance of systems using these reasoning mechanisms. Analysis of
defense applications will be made to determine which reasoning
mechanisms should be extended or which should be added, to provide the
basis for recommendations for possible follow-on work.
September 1983
(4) An advanced ACRONYM will be presented for delivery. It will have
incorporated improved boundary finding, curve segmentation, and ribbon
finding. It will have integrated som simple stereo capabilities. It
will have some improved geometric reasoning mechanisms. It will include
a rule base for a second class of objects, either vehicles or buildings.
It will have been demonstrated on a few scenes containing objects from
that class. A revised and expanded ACRONYM manual will be delivered at
that time.
-------
∂04-Nov-81 2108 SIS Colloquium Notice of November 9 - 13, 1981
To: "@COLLOQ.DIS[INF,CSD]" at SU-AI
Date Place Person
Day Event From
Time Title
-------------------------------------------------------------------------------
11/9/81 Math 380C Gary Rodrigue
Monday Numerical Analysis Lawrence Livermore Laboratory
4:15 p.m. Seminar ``Incomplete Block Cyclic Reduction For Iterative
Methods''.
11/10/81 Bldg. 200-30 Carl Hewitt
Tuesday Concurrency and AI MIT
2:35 p.m. ``The Discription System Omega''.
11/10/81 Jordan 041 Ron Graham
Tuesday Computer Science Bell Labs
4:15 p.m. Colloquium ``Bin Packing: A Paradigm In The Analysis of
Algorithms''.
11/11/81 Terman 153 To be announced
Wednesday Computer Systems
4:15 p.m. Laboratory Seminar
11/13/81 MJ301 To be announced
Friday Database Research
4:15 p.m. Seminar
∂04-Nov-81 2201 JMM Grader for CS206
To: DPB at SU-AI
CC: JMC at SU-AI
I have got no response to my BBOARD message about getting a grader for CS206.
However Sanjai Kasturia(SSK) and Yoram Moses(YOM) have agreed to do some limited
part-time grading each .Both of them are already supported
by the department as RA's and so there must be a maximum limit to the time they can
work extra- getting paid on an hourly basis. What is it? Also both of them are
taking the course too- but as there are several precedents of the grader also
doing the course it should be OK . I would like to have the official go-ahead
as soon as possible so that they can get started.
Jitendra Malik ( TA for CS206)
∂05-Nov-81 0742 Grosz at SRI-AI tinlunch INSIDE today
Date: 5 Nov 1981 0655-PST
From: Grosz at SRI-AI
Subject: tinlunch INSIDE today
To: tlgrp:
Even if it's warm and sunny and you think it ought to be outside,
TINLUNCH (Brian Smith will be here to discuss his commentary on
Newell's "Physical Symbol Systems") will be in EK242. NonSRI folks
who come after 12 should just tell the receptionist they are here
for tinlunch and know the way up (you can even tell her that if
you come before 12).
-------
∂05-Nov-81 0823 Shortliffe at SUMEX-AIM Re: Chandrasekaran at Ohio State
Date: 5 Nov 1981 0807-PST
From: Shortliffe at SUMEX-AIM
Subject: Re: Chandrasekaran at Ohio State
To: JMC at SU-AI
In response to your message sent 16 Oct 1981 2208-PDT
Sorry for the delay in responding to you; I was out of town until
yesterday.
Yes, I know Chandrasekaran and MDX. It is some of the newest medical
AI work -- the group there is about 2-3 years old. Has suffered from a lack
of adequate medical input, but the computer science seems interesting. Most
of my reservations about the work have been medical rather than technical.
It is based on a concept of "specialists" that cooperate on a problem solving
task. Would doubt it is a sufficiently mature set of ideas to be used in
nuclear power plants very soon -- guess it depends on whether they're looking
for a basic or applied research project.
Ted
-------
∂05-Nov-81 1411 FFL
To: JMC, FFL
Dr. Blum has called a meeting of SE2 on Wednesday evening, 8 p.m., Nov. ll,
at Tom Conally's home. Would like to know if you can attend. SLAC 2469.
∂05-Nov-81 1423 Denny Brown <CSD.DBROWN at SU-SCORE> Re: Grader for CS206
Date: 5 Nov 1981 1405-PST
From: Denny Brown <CSD.DBROWN at SU-SCORE>
Stanford-Phone: (415) 497-2274
Subject: Re: Grader for CS206
To: JMM at SU-AI, DPB at SU-AI
cc: JMC at SU-AI
In-Reply-To: Your message of 4-Nov-81 2201-PST
Go ahead. I believe that 8 hours of hourly grading is legal. Betty
will know for sure. Have them talk to Betty about the details of
how much is legal, and how to get paid. -Denny
-------
∂05-Nov-81 1612 SIS Updates to CS Colloquium Notice of
To: "@COLLOQ.DIS[INF,CSD]" at SU-AI
Updates for Computer Science Colloquium Notice ****
PhD Oral Examination --- Wednesday, November 11, at 9:30 am. in
Margaret Jacks Hall, Room 352. Lawrence Paulson, of Stanford
University, will speak on ``A Compiler Generator for Semantic
Grammars''.
Database Research Seminar --- Friday, November 13, at 3:15 p.m. in
Margaret Jacks Hall, Room 301. Neil Rowe, of Stanford University,
will speak on ``Rule-Based Statistical Calculations On A Database
Abstract''.
∂05-Nov-81 1707 CSD.ULLMAN@SU-SCORE (SuNet) equipment meeting
Mail-from: ARPANET host SU-SCORE rcvd at 5-Nov-81 1440-PST
Date: 5 Nov 1981 1427-PST
From: Jeffrey D. Ullman <CSD.ULLMAN at SU-SCORE at SUMEX-AIM>
Subject: equipment meeting
To: equip at SU-HPP-VAX, equip at SU-SHASTA
Don't forget that our committee still exists and will meet as usual on
Friday, 10am.
-------
∂05-Nov-81 1747 JMC@Sail (SuNet) KA-10 as file server
Date: 05 Nov 1981 1712-PST
From: John McCarthy <JMC at SU-AI>
Subject: KA-10 as file server
To: equip at DIABLO
The KA-10 has the right speed for a file server, but is not suitable,
because an interface to modern disks would have to be constructed.
I only respond to a suggestion that might not have been seriously
intended so as to underline the requirement that as little as possible
have to be designed, built and maintained.
∂05-Nov-81 2203 ME slow machine
∂05-Nov-81 2159 JMC
Can it really take almost six seconds to finger a person who is logged in?
ME - Sure, with the cache off everything takes a lot longer.
With the cache off, the KL isn't very fast.
∂06-Nov-81 0733 JK
∂06-Nov-81 0221 JMC obscure error message
Consider
lispax.lsp[f81,jmc]. Page 2 gives the axioms, and ekl liked
all but the last 3 and complained as shown at the end of page 3. It is
obscure to me and will be obscure to others what it didn't like.
Presumably it was something about the use of the infix *. Please consider
both my immediate problem of getting ekl to accept these axioms and the
long term problem of making the syntax error messages more informative.
Apparently the problem of making parsers give informative error messages
isn't easy to solve.
-----------------------
both * and = have the binding power of 800 in your current axioms.
Thus x=y*u gets parsed as (x=y)*u.
Change the binding power of * to something higher.
The parser could stand for some improvements - I have not paid much
attention to its error messages.
∂06-Nov-81 0735 JK
∂06-Nov-81 0138 JMC (list x y ... z)
Is there any convenient way of declaring and axiomatizing the properties
of the lisp list function? Note that it isn't associative. Currently
I am using list1(x) and list2(x,y).
----------
I am not sure what you want . You can declare functions of variable
arity using the * construct; f:ty*→type refers to a function
that takes ≥1 arguments of type ty.
I guess that after I have defined list(x,...,z) my problem will be
to write the axiom
list(x,y,...,z) = cons(x,list(y,...,z)).
Thanks for the answer on binding power.
∂06-Nov-81 0909 JK
∂06-Nov-81 0838 JMC
I guess that after I have defined list(x,...,z) my problem will be
to write the axiom
list(x,y,...,z) = cons(x,list(y,...,z)).
Thanks for the answer on binding power.
----------------
This requires an axiom schema - currently there is no good facilities
for this.
Suppose I declare the function list(x,...,z) as you mention in your
previous message. Will I have any problems with separate axioms
for the first few lengths, i.e.
(axiom |∀x.list(x)= cons(x,nnil)|),
(axiom |∀x.list(x,y) = cons(x,list(y))|)
(axiom |∀x.list(x,y,z) = cons(x,list(y,z))|)?
∂06-Nov-81 1007 FFL
To: JMC, FFL
Dr. Blum's secretary called to remind you of 12:30 appointment at
Faculty Club with Dr. Blum today.
That's Elliott Bloom - 2 Ls, 2Ts and 2 Os.
∂06-Nov-81 1111 FFL Message from Mr. Paolucci
To: JMC, FFL
There is no meeting on Monday, November 9, of III. Mr. Walsh would like
to have the meeting on November 19, which is the day after the Board
meeting. 213 391 7211.
∂06-Nov-81 1130 Bob Amsler <AMSLER at SRI-AI> TinLunch of Thursday November 12th - Automated Dictionaries
Date: 6 Nov 1981 1111-PST
From: Bob Amsler <AMSLER at SRI-AI>
Subject: TinLunch of Thursday November 12th - Automated Dictionaries
To: tlgrp: ;
TINLUNCH, November 12th
TOPIC: Automated Dictionaries
ABSTRACT: George A. Miller, in 1979, chaired a conference
sponsored by the NIE which examined the prospects for
developing an an automated dictionary. This results of
this conference are continuing to surface in subsequent
publications and proposals. Carnegie-Mellon's Mark S.
Fox, Donald J. Bebel, and Alice C. Parker reported on an
analysis of the hardware/software they foresaw for the
automated dictionary in an article in July, 1980 Computer.
"The Automated Dictionary", pp. 35-48, Vol. 13(7).
NSF is currently evaluating a proposal from George Miller
to undertake "Psycholexical" research which would carry out
many of the recommended tasks of the conference. In view
of the continued effort to realize the conference's goals I
have selected the original conference report and the
related CMU paper for the next TINLunch.
I've ordered 20 copies of the Conference report and 15
copies of the "Computer" article (since some may already
have seen it). They will appear when available on
Barbara's venerable file cabinet.
Bob Amsler
-------
∂06-Nov-81 1230 Brian K. Reid <CSL.BKR at SU-SCORE> Finding the Potluck house
Date: 5 Nov 1981 2044-PST
From: Brian K. Reid <CSL.BKR at SU-SCORE>
Subject: Finding the Potluck house
To: RSF at SU-AI
cc: BBoard at SU-SCORE, BBoard at SU-AI, Guarino at PARC-MAXC
In-Reply-To: Your message of 5-Nov-81 1419-PST
Remailed-date: 6 Nov 1981 1204-PST
Remailed-from: Carolyn Tajnai <CSD.TAJNAI at SU-SCORE>
Remailed-to: CSD-Faculty: ;
How to find our house from Stanford for the potluck at 6pm this Saturday:
Palo Alto north/south streets are numbered from 100's starting at the
creek, so that 2960 Waverley is 29 blocks south of Menlo Park. More to
the point, it is 3 blocks south of Oregon expressway. By bicycle, the
best way to get there is to leave campus on the road by the Credit
Union, cross El Camino, then follow that road to California Street. Take
the California Street bicycle underpass, then continue down California
street to Waverley, turn right, and go 4 blocks. The faint of heart can
just go down Waverley the whole way from University Avenue.
Plenty of bicycle parking in the back. People with cars are on their own.
|
-----------Palm Dr.--------------|
| |
El| |
Camino| |
| |Alma
Credit Union* | |
=========+===\ |
| I |
| I | California St.
+---+=== | ==========+-------
| | I
| | I
+---+---------------------+---Oregon Expwy-----> 101
| | I
| I
2960 I
Waverley *|
Our contractor hasn't quite finished the remodeling job in our living
room, which means that (a) there will be lots of space for the potluck
because all the furniture is in the back yard, and (b) we will dine in
the pleasant atmosphere of curing plaster, uninspected electrical
fixtures, and an occasional stepladder for character.
-------
∂06-Nov-81 1234 Brian K. Reid <CSL.BKR at SU-SCORE> Re: Quarterly Potluck dinner!!!
Date: 30 Oct 1981 2002-PST
From: Brian K. Reid <CSL.BKR at SU-SCORE>
Subject: Re: Quarterly Potluck dinner!!!
To: RSF at SU-AI, bboard at SU-SCORE
cc: csd.yellin at SU-SCORE
In-Reply-To: Your message of 30-Oct-81 1445-PST
Remailed-date: 6 Nov 1981 1207-PST
Remailed-from: Carolyn Tajnai <CSD.TAJNAI at SU-SCORE>
Remailed-to: CSD-Faculty: ;
The potluck will be held at Loretta and Brian Reid's house, same address
as previously posted.
-------
∂06-Nov-81 1237 Ross Finlayson <RSF at SU-AI> Quarterly Potluck dinner!!!
Mail-from: ARPANET site SU-AI rcvd at 30-Oct-81 1448-PST
Date: 30 Oct 1981 1445-PST
From: Ross Finlayson <RSF at SU-AI>
Subject: Quarterly Potluck dinner!!!
To: bboard at SU-SCORE
CC: RSF at SU-AI, csd.yellin at SU-SCORE, csl.bkr at SU-SCORE
Remailed-date: 6 Nov 1981 1208-PST
Remailed-from: Carolyn Tajnai <CSD.TAJNAI at SU-SCORE>
Remailed-to: CSD-Faculty: ;
***** POTLUCK *****
***** POTLUCK *****
The quarterly CS departmental potluck dinner will be held on
Saturday, November 7th
at 6pm
at Brian Reid's home -
2960 Waverly St., Palo Alto.
There will be no sign-up sheets (people have tended not to
make use of them in the past); please bring your favorite entree, salad,
appetizer, dessert, beverage, or anything else consumable!
We would appreciate volunteers to help with the cleaning up
afterwards.
∂06-Nov-81 1240 JK
∂06-Nov-81 1207 JMC
Suppose I declare the function list(x,...,z) as you mention in your
previous message. Will I have any problems with separate axioms
for the first few lengths, i.e.
(axiom |∀x.list(x)= cons(x,nnil)|),
(axiom |∀x.list(x,y) = cons(x,list(y))|)
(axiom |∀x.list(x,y,z) = cons(x,list(y,z))|)?
-------
This should be fine.
∂06-Nov-81 1452 Carolyn Tajnai <CSD.TAJNAI at SU-SCORE> 1982 Annual Computer Forum Meeting
Date: 6 Nov 1981 1446-PST
From: Carolyn Tajnai <CSD.TAJNAI at SU-SCORE>
Subject: 1982 Annual Computer Forum Meeting
To: CSD-Faculty: ;
cc: CSD.BScott at SU-SCORE, Admin.Gorin at SU-SCORE, CSD.MWalker at SU-SCORE
The 1982 Annual Computer Forum Meeting will be held Thursday, Friday,
February 4 and 5, 1982. Please mark your calendars.
-------
∂06-Nov-81 1530 Danny Berlin <CSD.BERLIN at SU-SCORE> Autumn 1980 Course Evaluations
Date: 6 Nov 1981 1524-PST
From: Danny Berlin <CSD.BERLIN at SU-SCORE>
Subject: Autumn 1980 Course Evaluations
To: csd.ydens at SU-SCORE, csd.dbrown at SU-SCORE, csd.yao at SU-SCORE,
csd.knuth at SU-SCORE, csd.schreiber at SU-SCORE, csd.karlb at SU-SCORE,
csd.schnepper at SU-SCORE, csd.buchanan at SU-SCORE,
csd.herriot at SU-SCORE, csd.manna at SU-SCORE, csl.bkr at SU-SCORE,
rwf at SU-AI, csd.siegel at SU-SCORE, csd.dietterich at SU-SCORE,
csd.jock at SU-SCORE, csd.haiman at SU-SCORE, csd.griffiths at SU-SCORE,
csd.oliger at SU-SCORE, csd.ullman at SU-SCORE, or.dantzig at SU-SCORE,
csl.lab.jfw at SU-SCORE, csd.mccarthy at SU-SCORE, csl.crc.ejm at SU-SCORE,
tob at SU-AI
cc: jjw at SU-AI
We've finally managed to get our act together and the Autumn book for 1980
is ready for printing. If you want to review the summaries of your respective
course(s) before it is printed please send me a message. If I don't hear
anything by next Friday I'll assume the summaries are satisfactory.
Incidentally, the Winter evaluation should also be ready for review in a couple
of weeks.
Danny Berlin
Course Evaluation Committee Chairwhatever
-------
∂06-Nov-81 1535 Carolyn Tajnai <CSD.TAJNAI at SU-SCORE> 1982 Annual Computer Forum Meeting
Date: 6 Nov 1981 1446-PST
From: Carolyn Tajnai <CSD.TAJNAI at SU-SCORE>
Subject: 1982 Annual Computer Forum Meeting
To: CSD-Faculty: ;
cc: CSD.BScott at SU-SCORE, Admin.Gorin at SU-SCORE, CSD.MWalker at SU-SCORE
The 1982 Annual Computer Forum Meeting will be held Thursday, Friday,
February 4 and 5, 1982. Please mark your calendars.
-------
∂06-Nov-81 2352 RPG
∂06-Nov-81 2344 JMC slowness of page printer
A preliminary experiment indicates that it takes about half the computer
time for a process to output to E with your hack than it does to print
it on the page printer. If this is so, then the ancient page printer is
one of the main causes of slowness in the system. Do you have or can you
develop any info on this point?
What exactly do you mean by `my hack'? I think that the E/Lisp interface
always uses the page printer code unless you are outputting to the file
or the attach buffer.
-rpg-
I was outputting to the file.
∂07-Nov-81 0000 RPG@sail (suNet) Meeting
Date: 06 Nov 1981 2350-PsT
From: Dick Gabriel <RPG at sU-AI>
subject: Meeting
To: equip at DIABLO, csd.genesereth at sU-sCORE,
masinter at PARC-MAXC
Today I attended a meeting at sRI concerning some technical
aspects of implementing InterLisp on the symbolics L machine (3600).
Present were: Paul Martin, Daniel sagalowisc, Daniel Weinreb, Neil Goldman,
Bob Balzer, steve Gadol, Harry Barrow, Heathman (?), Doug Appelt, and
one or two others. We drew up a list of required features and discussed
some implementation issues, some being trivial, some difficult. We
also speculated about the availability of the Dolphin InterLisp code
and the virtual machine specifications used on that implementation,
but no one from Xerox was there, so nothing was known.
It sounded as though symbolics was ready to make a commitment
on this, including microcode support. IsI, Fairchild, and sRI appeared
ready to move InterLisp code to Lisp Machines as well as use the native
Common Lisp.
If this is really a possibility, we ought to consider it seriously
as an alternative when the ARPA money arrives.
-rpg-
∂07-Nov-81 0823 pratt@Diablo (SuNet) Meeting
Date: 7 Nov 1981 08:21:47-PST
From: pratt at Diablo
To: RPG@Sail, csd.genesereth@SU-SCORE, equip@DIABLO, masinter@PARC-MAXC
Subject: Meeting
Who besides Dan Weinreb was representing Symbolics at the meeting? It would
be nice to hear support for Interlisp from several sources within Symbolics
and at several times.
DLW@AI will reach Weinreb. HIC@MC, TK@AI, JLK@MC will reach other Symbolics
designers, the last (John Kulp) being director of technical operations (I
forget his exact title).
Stanford is moving into a competitive position with Marine World; there are now
5 Dolphins on campus. People interested in finding out at first hand what
Dolphins are like to use should find it easy to get at them. Two have
recently appeared in MJH433 and are presently usable as network terminals;
hopefully they will soon be usable as Lisp workstations.
Vaughan
∂07-Nov-81 1442 Konolige at SRI-AI reading committee
Date: 7 Nov 1981 1441-PST
From: Konolige at SRI-AI
Subject: reading committee
To: jmc at SAIL
John, I should get my reading committee formed. You and Nils are
obvious members. For the second Acedemic Council member, perhaps Zohar
Manna or Terry Winograd would be a good choice. For someone to keep me
honest, Chris Goad, Carolyn Talcott, or one of the philosophy faculty
would be appropriate. I would like your comments on this. --kk
-------
∂07-Nov-81 1452 RPG@Sail (SuNet) Meeting
Date: 07 Nov 1981 1448-PST
From: Dick Gabriel <RPG at SU-AI>
Subject: Meeting
To: equip at DIABLO, VRP at SU-AI, csd.pratt at SU-SCORE,
pratt at DIABLO, csd.genesereth at SU-SCORE,
masinter at PARC-MAXC
As I understood things, this technical meeting was a followup
of one or more business meetings during which the principle of InterLisp
support on Symbolics machines had been discussed, and possibly solidified in
writing. As I pointed out, several people from several companies in California
who have purchased Lisp Machines were there to explain the technical
requirements for the project. Henry Baker was to attend but was not able to
at the last moment. My impression was that I was at the nth meeting, not
at the first, and my role was to assist SRI in judging the technical merits
of the proposed plan. In reality, I assisted Weinreb in making judgements
about the various strategies for implementation of features.
I believe Symbolics is serious, and perhaps interested parties
would like to call the appropriate saleman for details.
-rpg-
∂08-Nov-81 0856 JK
∂08-Nov-81 0047 JMC
Say again what files contains the description of the new rewrite, etc.
------------
ekl.man[ekl,jk]
∂09-Nov-81 0023 RPG
∂09-Nov-81 0021 JMC writing a paper
To: RPG, ME
The two of you should write a short paper about running programs and the
system out of the editor. This should be the way all operating systems
should be controlled, but this fact is scarcely known. If necessary I'll
help with the advocacy part.
We are writing it, and I've started a draft already.
-rpg-
∂09-Nov-81 0900 JMC*
Call Walsh or Paolucci.
∂09-Nov-81 1030 RWW VACATION
To: FFL, JMC
Please consider me on vacation today. Fran would you please determine how
vacation time I have and send me a message. Thanks
Richard
Date: 9 Nov 1981 1245-PST
From: Kanerva at SUMEX-AIM
To: McCarthy at SU-AI
Subject: Effective Interactive Use of Large Character Sets
Thank you for the copy of your paper. I have read it twice and
in light of it feel confident about our work on extending the character
set. It seems to me that our work is very much in the spirit of your
paper and could benefit even other DEC-10 and 20 sites at Stanford.
Let me describe it in some detail.
Files are written in 7-bit characters, and the code is a superset
of ASCII. Files written in the extended code make sense even if printed
with programs for ASCII files, just as Scribe and TEX source files make
sense. Old programs like Spell will work with the new files. The file
code is converted easily to any reasonable code (a possible ANSI
standard code) with a large character set.
In addition to ASCII, three things are represented in the file:
(1) symbols for math, languages with small (phonetic) alphabets, the
common type styles found in manuscripts (italics, bold, small caps);
and (3) half-line spacing. A good way to think of what we are trying
to provide is what is available on a typewriter with exchangeable type
elements. The extensions will be available in TVEDIT and the printing
programs. How they will be carried over to TEX and Scribe is still
open but should not be overly difficult.
Initially the code has space for 500 or 1000 symbols. We have
not yet decided which, but most likely it will be 1000. The case of
a letter--upper or lower--is ignored in counting symbols. So, as an
example, English requires 26 symbols and the Cyrillic alphabet between
30 and 40 (deBray, R. G. A., 1969. GUIDE TO THE SLAVONIC LANGUAGES).
The American Mathematical Society lists 180 symbols in MATHEMATICS INTO
TYPE (1979). The U. S. GOVERNMENT PRINTING OFFICE STYLE MANUAL (1976)
is a good source of symbols, and there are others.
Assuming a file code with 1000 symbols, we will select and fix
500-700 of them and leave the remaining few hundred free for special
applications, to be specified by the user. Our plan is to select an
initial set of 300-400 symbols, distribute it to potential users at
Stanford, and ask for their suggestions for additions. This "survey"
may be under way by the end of the year. I hope that you will
participate--even by reviewing the call for suggestions before it
goes out.
All characters will be of one point size and one width (monospace).
Anything beyond that is too complicated right now and may be too
complicated forever. To make it simple would require a world-wide
agreement on the relative widths of characters, as without such an
agreement the on-line exchange of text is difficult.
(I have a nagging feeling that the exchange of information will
be hindered by attempts to simulate typeset print. This is not yet
abvious because information is still mostly carried on paper, but what
happens when more of it is exchanged on line? People do not want to be
reading TEX or Scribe source files, and it will be a long time before
everybody has a display capable of much more than that. A similar
thing happend with typesetting. The early attempts were to produce
texts indistinguishable from the work of scribes.)
Our "stock" terminals, Datamedias and Heaths, will do for editing.
Highlighting will be used on these terminals to show special characters
(bright "a" for lowercase alpha). The Concept 108 by Human Designed
Systems has room for 512 characters. We have the facilities to design
the characters and program them in PROMS. You saw a sample in the demo
last Friday.
We are working on the quite difficult problem of mapping the master
character set to the set available on different display terminals. We
are trying to find a general mechanism that allows a user to describe
his terminal's character set to the display-dependent part of the
editor. The person working on the problem is Brian Tolliver, the author
of the first TVEDIT for the PDP-1. He wrote the part of the demo
program you saw that displays the special characters and the attributes.
No special keyboard is necessary to type special characters.
Some (Greek) charactes will be entered by typing a prefix and the
"corresponding" ASCII (Latin) character and others will be entered as
mnemonic sequences ("not equls" could involve the typing of / and =).
We have not yet worked out all details of typing.
The Printronix printer will print the entire character set in plot
mode, and the Canon laser printer will have the entire character set
available in Elite type. The Diablo printer will do half-line spacing
and will indicate italics, bold, and small caps with different
underlines (single for italics, wavy for bold, and double for small
caps--according to publishers' standard for manuscripts) but will not
have extra symbols.
I hope that this work interests you and that we can talk about it
more. If you want more details, please let me know. If you feel that
some or all of our work is misguided, that is even more reason to let
me know. Thanks. - Pentti
-------
∂09-Nov-81 1848 Bob Amsler <AMSLER at SRI-AI> "The Automated Dictionary" by Fox, Bebel and Parker
Date: 9 Nov 1981 1549-PST
From: Bob Amsler <AMSLER at SRI-AI>
Subject: "The Automated Dictionary" by Fox, Bebel and Parker
To: tlgrp: ;
Copies have been made of the 2nd article for this Thursday's TINLunch
and are available on Barbara's filing cabinet.
-------
∂10-Nov-81 0912 JK
∂08-Nov-81 1740 JMC
The proof is samele.prf[f81,jmc]. The proof name is samelength.
A decsimp resulted as follows:
(decsimp samelength#3#1#1 nil (10 9 lispax#24 lispax#20 6) nil
(lispax#12 lispax#23 lispax#19))
error in type calculation - tell jk
-----------------
This has been fixed.
∂10-Nov-81 1021 JJW Samelength proof
I worked on the proof some more last night, and reduced the number of lines,
but when I tried to do a decsimp like the one that failed on Sunday, I got
the same error message. Jussi says he has now fixed the bug, but the load
is currently 25 so I won't attempt to get any further until tonight.
∂10-Nov-81 1134 Irmgild Schack <CSD.IRMGILD at SU-SCORE> Faculty Lunch !Reminder!
Date: 10 Nov 1981 0954-PST
From: Irmgild Schack <CSD.IRMGILD at SU-SCORE>
Subject: Faculty Lunch !Reminder!
To: CSD-Faculty: ;
There will be a faculty luncheon today in the Boys Town Conference Room.
Jim Bell from HP will be here to meet the faculty; he is interested in
improving the relationship between Stanford's CSD and HP.
Next Tuesady, November 17, Dean Rosse (Humanities & Sciences) will be
at the lunch.
-------
∂11-Nov-81 0114 LGC Advice Taker
Will you have some time this afternoon to discuss the Advice Taker?
RPG and I have been discussing deliverables for ARPA, and I can pass on our
thoughts on that subject, unless you've already arranged for the
deliverables. Anyway, I've been making good progress developing the basic
internal logical formalism, together with programs to translate LISP-form
input expressions into it, and to display internal formulas in a concise
and easily readable format. I thought you might like a progress report;
would 3:00 or 4:00pm today be ok?
-- Lew
-------------------------
Appendix -- Example of current program performance:
Being careful not to move anything else on his desk,
John picks up hammer1 with his right hand in order to drive a nail with it.
((some ↑x (conceptof ↑x (!desk john)) ;; input formula
↑y (conceptof ↑y hammer1)
z (nail z)
↑w (conceptof ↑w z))
(pickup john hammer1
(with (!right-hand john))
(from (!desk john))
(inorderthat (↑(drive i (↓ ↑w) (with (↓ ↑y)))))
(withcarethat (↑((all v (and (thing v)
(on v (↓ ↑x))
(not (= v (↓ ↑y))) ))
(not (move i v)) ))) ) )
; This gets translated by the function ENCODE-LINFORMULA into a fairly
; complicated (but logically clean) internal data structure in the language
; of thought. This internal language has many of the features of semantic
; network formalisms, but is nevertheless a formal logic in a strict sense,
; for which a fully rigorous model-theoretic semantics can be provided.
; The DISPLAY function produces from the internal data structure (which is
; easily readable only by the machine) the following external representation
; for reading by humans (the correspondence of particular variables used in
; the input ond output formulas is only coincidental, since the language of
; thought is variable-free):
{∃↑X.CONCEPTOF(↑X, !DESK JOHN)
↑Y.CONCEPTOF(↑Y, HAMMER1)
Z.NAIL
↑V.CONCEPTOF(↑V, Z)}.PICKUP(JOHN, HAMMER1, WITH:!RIGHT-HAND JOHN,
FROM:!DESK JOHN, INORDERTHAT:↑[DRIVE(I, ↓[↑V], WITH:↓[↑Y])],
WITHCARETHAT:↑[{∀W.THING W ∧ ON(W, ↓[↑X]) ∧ ¬=(W, ↓[↑Y])}.¬MOVE(I, W)])
∂11-Nov-81 1000 FFL
To: JMC, FFL
Richard Wortman of the Independent Network News, New York, is in Bay
Area. Wants to do story on AI. He will call you again this afternoon
about 1:30, but if you have a chance, he would like a call at
415 540 0827.
∂11-Nov-81 1114 Danny Berlin <CSD.BERLIN at SU-SCORE> Course Evaluations for Aut 1980
Date: 11 Nov 1981 1059-PST
From: Danny Berlin <CSD.BERLIN at SU-SCORE>
Subject: Course Evaluations for Aut 1980
To: csd.ydens at SU-SCORE, csd.dbrown at SU-SCORE, csd.yao at SU-SCORE,
csd.knuth at SU-SCORE, csd.schreiber at SU-SCORE, csd.karlb at SU-SCORE,
csd.schnepper at SU-SCORE, csd.buchanan at SU-SCORE,
csd.herriot at SU-SCORE, csd.manna at SU-SCORE, csl.bkr at SU-SCORE,
rwf at SU-AI, csd.siegel at SU-SCORE, csd.dietterich at SU-SCORE,
csd.jock at SU-SCORE, csd.haiman at SU-SCORE, csd.griffiths at SU-SCORE,
csd.oliger at SU-SCORE, csd.ullman at SU-SCORE, or.dantzig at SU-SCORE,
csl.lab.jfw at SU-SCORE, csd.mccarthy at SU-SCORE, csl.crc.ejm at SU-SCORE,
tob at SU-AI
cc: csd.trickey at SU-SCORE, jjw at SU-AI
Several suggestions have been made with regard to course evaluations.
The original forms and summaries will be sent to each teacher. Any teacher
disputing the accuracy of the summary should send me a message by Friday
Nov. 20. Also, any teacher that wishes to place a policy statement to reply
to comments in the summary should also submit it to me by that date. All
such statements will be published without editting (assuming abusive terms
are not used).
Finally, to save money only those professors specifically requesting a copy
of the entire book will receive one. Our default assumption is that each
teacher is primarilly concerned with the evaluation of their own course(s).
Danny Berlin
Course Evaluation Committee Etc.
-------
I don't think the adversary attitude expressed in your November 11 message is
appropriate. Therefore, I will not look at the "course evaluations" you
send me, and I will not make time this quarter for "course evaluation".
∂11-Nov-81 1248 Gene Golub <CSD.GOLUB at SU-SCORE> [John McCarthy <JMC at SU-AI>:]
Date: 11 Nov 1981 1158-PST
From: Gene Golub <CSD.GOLUB at SU-SCORE>
Subject: [John McCarthy <JMC at SU-AI>:]
To: JMC at SU-AI
WHAT WAS IN HIS MESSAGE? GENE
---------------
Mail-from: ARPANET site SU-AI rcvd at 11-Nov-81 1127-PST
Date: 11 Nov 1981 1126-PST
From: John McCarthy <JMC at SU-AI>
To: csd.berlin at SU-SCORE
CC: csd.golub at SU-SCORE
I don't think the adversary attitude expressed in your November 11 message is
appropriate. Therefore, I will not look at the "course evaluations" you
send me, and I will not make time this quarter for "course evaluation".
-------
∂11-Nov-81 1249 Gene Golub <CSD.GOLUB at SU-SCORE> BELL RESPONSE
Date: 11 Nov 1981 1159-PST
From: Gene Golub <CSD.GOLUB at SU-SCORE>
Subject: BELL RESPONSE
To: JMC at SU-AI
cc: CSD.GOLUB at SU-SCORE
I MENTIONED YOUR PROPOSAL AT THE LUNCH ON TUESDAY AND BELL AGREED
TO SUPPLY SPEAKERS. GENE
-------
∂11-Nov-81 1446 Jeffrey D. Ullman <CSD.ULLMAN at SU-SCORE> equip.
Date: 11 Nov 1981 1442-PST
From: Jeffrey D. Ullman <CSD.ULLMAN at SU-SCORE>
Subject: equip.
To: csl.jlh at SU-SCORE, csl.jhc at SU-SCORE, csl.bkr at SU-SCORE,
csd.cheriton at SU-SCORE, csl.lantz at SU-SCORE, engelmore at SUMEX-AIM,
rpg at SU-AI, jmc at SU-AI
cc: csd.golub at SU-SCORE, csd.feigenbaum at SU-SCORE
I am assuming that the people addressed are in the process of writing a
section for the ARPA proposal. In some cases, my assumption is based on
nothing more than a sincere wish on my part, but I would appreciate
eceiving drafts from all parties by Weds. Nov. 25. The pieces I am expecting
are:
Engelmore--HPP needs.
McCarthy, Gabriel--other needs of the LISP community and issues
related to comparison of LISP machines.
Clark, Hennessy--VLSI related needs.
Cheriton, Lantz--Distributed systems needs.
Reid--System overview.
-------
∂11-Nov-81 1629 Irmgild Schack <CSD.IRMGILD at SU-SCORE> Vaughn Pratt Appointment
Date: 11 Nov 1981 1605-PST
From: Irmgild Schack <CSD.IRMGILD at SU-SCORE>
Subject: Vaughn Pratt Appointment
To: Tenured: ;
I think it is time to conclude, if possible, the business of the
Vaughn Pratt appointment. There will be a meeting at 2:30 on
Thursday, November 19 in the conference room next to my office
GENE
-------
∂11-Nov-81 1650 Danny Berlin <CSD.BERLIN at SU-SCORE>
Date: 11 Nov 1981 1648-PST
From: Danny Berlin <CSD.BERLIN at SU-SCORE>
To: JMC at SU-AI
cc: csd.berlin at SU-SCORE
In-Reply-To: Your message of 11-Nov-81 1126-PST
The purpose of this is not to serve as an adversary attitude but to allow
professors to clarify their positions on what the course is intended to
accomplish. Frequently, it is misunderstandings of this sort that cause
many of the students' complaints.
The opportunity to check the summaries first is meant as a method to guarrantee
that the summaries accurately express the written comments and do not exagerate
the case (eg. Most students felt the lectures were boring when only 4 people
made such a comment).
Summaries will only be changed when they are shown to be inaccurate and comments
by professors only printed when they help clarify why some problem occurred.
If you still feel strongly about this your courses will not be included in the
evaluation book. However I will feel obligated to state that you refused to
have the course included and would prefer to have a statement from you as to
your reasons to use in the book.
Danny Berlin
Computer Science Course Evaluation Committee Chairman
-------
∂11-Nov-81 2151 ENGELMORE at SUMEX-AIM Re: equip.
Date: 11 Nov 1981 2148-PST
From: ENGELMORE at SUMEX-AIM
Subject: Re: equip.
To: CSD.ULLMAN at SU-SCORE, csl.jlh at SU-SCORE, csl.jhc at SU-SCORE,
csl.bkr at SU-SCORE, csd.cheriton at SU-SCORE, csl.lantz at SU-SCORE,
rpg at SU-AI, jmc at SU-AI
cc: csd.golub at SU-SCORE, csd.feigenbaum at SU-SCORE
In-Reply-To: Your message of 11-Nov-81 1442-PST
I have an outline for the whole proposal, and have begun filling in portions
relevant to both HPP and SAIL needs (Dick Gabriel has supplied a good deal of
material for this). Since I have some experience with ARPA proposals, I
will volunteer to be the integrator for this effort.
The main material that each group should provide is:
1. Background and Technical Need
a. What's the problem? That is, in this context, what are your current
computing facilities, and why are they inadequate?
b. What are your requirements?
c. What alternatives are there for satisfying those requirements?
2. Proposed Acquisition Plan
After considering the alternatives, what equipment are yu actually proposing
to get?
That's basically it. The global material, like a Summary section, a budget,
CV's and other supporting material we can add at the end.
Bob
-------
∂12-Nov-81 0929 Konolige at SRI-AI meeting
Date: 12 Nov 1981 0928-PST
From: Konolige at SRI-AI
Subject: meeting
To: jmc at SAIL
John, I'd like to meet with you somtime today to discuss
the formation of my committee. Will you be available at Stanford
this afternoon? --kk
-------
∂12-Nov-81 0953 OPERATOR at SRI-AI TinLunch Today on "Automated Dictionaries" in EK242 at noon
Date: 12 Nov 1981 0930-PST
From: OPERATOR at SRI-AI
Subject: TinLunch Today on "Automated Dictionaries" in EK242 at noon
To: tlgrp: ;
How will natural language processing attain the capability of processing
unrestricted input text?
What is the role of machine-readable dictionaries in computational
linguistics and information science and other fields.
Is specialized hardware required before dictionaries can be usefully
applied in artificial intelligence?
What would be interesting ways to access dictionary data
-------
∂12-Nov-81 1421 Jrobinson at SRI-AI Tinlunch reading for next week.
Date: 12 Nov 1981 1421-PST
From: Jrobinson at SRI-AI
Subject: Tinlunch reading for next week.
To: tlgrp:
I am listing the abstract for next week's reading in the next message.
It will be 8 pages long. If you know of possible attendees who are not
on the net, will you offer them a copy? Thanks.
Jane
-------
∂12-Nov-81 1511 Konolige at SRI-AI (Kurt Konolige)
Date: 12 Nov 1981 1506-PST
From: Konolige at SRI-AI (Kurt Konolige)
To: JMC at SU-AI
In-Reply-To: Your message of 12-Nov-81 1503-PST
Ok, what time tomorrow? --kk
-------
∂12-Nov-81 1519 Jrobinson at SRI-AI Tinlunch abstract
Date: 12 Nov 1981 1514-PST
From: Jrobinson at SRI-AI
Subject: Tinlunch abstract
To: tlgrp:
Morphological Analysis of Finnish by Computer
←←←←←←←←←←←←← ←←←←←←←← ←← ←←←←←←← ←← ←←←←←←←←
Lauri Karttunen, Rebecca Root, and Hans Uszkoreit
University of Texas at Austin
1. Introduction
Finnish is a higly inflected, agglutinative language in which nouns,
adjectives, and verbs take on numerous suffixes to express number, case,
possession, tense, mood, person, and other morphological categories.
The form of the word stem varies depending on what suffixes attach to
it, and the form of the suffixes themselves also varies depending on
their environment. This is illustrated in the chart below, which shows
a small part of the inflectional paradigm for ten different types of
nouns. (The glosses are: law, top, pig, blind, wolf, muscle, thief,
intention, guarantee, and moon.)
Nom.Sg. Gen.Sg. Part.Sg. Ill.Sg. Gen.Pl. Part.Pl. Ill.Pl.
laki lai-n laki-a laki-in laki-en lake-j-a lake-i-hin
laki lae-n lake-a lake-en lak-i-en lak-i-a lak-i-in
sika sia-n sika-a sika-an siko-j-en siko-j-a siko-i-hin
sokea sokea-n sokea-a sokea-an soke-i-den soke-i-ta soke-i-siin
susi sude-n sut-ta sute-en sut-ten sus-i-a sus-i-in
lihas lihakse-n lihas-ta lihakse-en lihas-ten lihaks-i-a lihaks-i-in
varas varkaa-n varas-ta varkaa-seen varas-ten varka-i-ta varka-i-siin
aie aikee-n aiet-ta aikee-seen aike-i-den aike-i-ta aike-i-siin
takuu takuu-n takuu-ta takuu-seen taku-i-den taku-i-ta taku-i-siin
kuu kuu-n kuu-ta kuu-hun ku-i-den ku-i-ta ku-i-hin
Table 1: Sample case forms for ten types of nouns
A noun stem, such as susi 'wolf', can have as many as five different
←←←←
allomorphs: susi, sude-, sut-, sute-, and sus-. Some case endings also
←←←← ←←←← ←←← ←←←← ←←←
come in many different shapes; for example, the above chart contains
seven of the 36 variants for the illative case suffix. The standard
dictionary contains paradigm tables for 82 different types of nouns and
45 types of verbs.
Another complicating factor is that in some cases there are many
alternatives to choose from. For example, a word like peruna 'potato'
←←←←←←
has two possible forms in the illative plural, three plural partitives,
and five plural genitives:
! Page 2
Ill.Pl. peruno-i-hin, perun-i-in
Part.Pl. peruno-i-ta, peruno-j-a, perun-i-a
Gen.Pl. peruno-i-den, peruno-i-tten, peruno-j-en,
perun-i-en, peruna-in
Table 2: Alternative forms for peruna 'potato'
Finnish also has a rich system of derivational morphology. Even if we
only take into account the most common productive suffixes, the set of
derivational expansions of a typical noun or verb stem contains a large
number of possible words. This is illustrated in Table 3.
oppi-a to learn
ope-tta-a to teach ("to cause to learn")
ope-tta-ja teacher
ope-tta-ja-tar female teacher
ope-tta-ja-ttare-ton lacking a female teacher
ope-tta-ja-ttare-ttoma-sti without having a female teacher
Table 3: Iteration of derivational suffixes
As the above table shows, derivational suffixes also typically have
several allomorphs depending on their environment.
Fluent speakers of Finnish obviously must have the ability to decompose
complex words and to assemble them from stems and suffixes. Because
Finnish is spoken just about as fast as any other language, they also
must have very efficient means for dealing with the problem
We have designed a computer program, called TEXFIN, to model the ability
of word recognition/formation in Finnish. This program is intended to
serve as the first stage of a more ambitious program for automatic
syntactic analysis. In its present form, TEXFIN can recognize, in a
fraction of a second, any inflected form of a word which is represented
in its "root" lexicon. The current root lexicon contains a few hundred
entries with at least a couple of words for each inflectional type in
the language. It could easily be enlargened. In addition to the root
lexicon, there is also a large number of auxiliary lexicons that contain
alternation patterns, inflectional, and derivational suffixes.
The program can also produce the complete inflectional paradigm for any
word it can recognize. TEXFIN accepts and generates only forms that are
morphologically and phonologically correct; ungrammatical forms are
rejected. It can also handle the most common derivational suffixes.
The program is written in Interlisp for a DEC-20 running under TOPS-20
operating system. TEXFIN can analyze short unambiguous words in less
than 20 milliseconds. Long words and words that require a lot of
disambiguation can take ten times longer. Generating a complete
paradigm for a typical noun or verb takes less than a second.
In the following sections we first give a overview of TEXFIN and then
! Page 3
describe in some detail the methods it uses in decomposing words. This
is followed by a discussion of the linguistic implications of the
analysis. We conclude with some comments about of TEXFIN as a model of
human word recognition.
2. Overview of TEXFIN
In its present form TEXFIN is an interactive program that expects the
user to type in one of more Finnish words on a terminal. It analyzes
each input word and responds by displaying its translation and
morphological properties. A conversation with TEXFIN might go as
follows:
User: aikeisiimme
TEXFIN: INTENTION Noun Pl Illat Pl2nd
indicating that aikeisiimme is the plural illative form of the noun
meaning 'intention' with the 2nd person plural possessive suffix. The
program is designed to recognize only correctly inflected forms. If it
receives an incorrect form, TEXFIN responds with "?".
User: aieisiimme
TEXFIN: ?
If a form can be analyzed in more than one way, all the possible
analyses are listed:
User: lakiin
TEXFIN: LAW Noun Illat Sg
TOP Noun Illat Pl
User: halua
TEXFIN: WANT Verb Active Imperat Sg 2nd
WANT Verb Active Present Neg
DESIRE Noun Sg Partit
In the first case two analysis are printed because lakiin is
←←←←←←
simultaneously a form of two different nouns that happen to be
homonymous in some of their case forms. The latter case illustrates a
three-way ambiguity between two verb analyses and one noun analysis.
TEXFIN can also recognize compound forms. It distinguishes the
modifying noun from the head by indentation:
User: sikavarkaita
TEXFIN: PIG Noun Sg Nominat
THIEF Noun Pl Partit
If a compound can be segmented in more than one way, TEXFIN finds all
the possible analyses:
! Page 4
User: koululaiskuri
TEXFIN: STUDENT Noun Sg Nominat
DISCIPLINE Noun Sg Nominat
SCHOOL Noun Sg Nominat
GOOF-OFF Noun Sg Nominat
This form can be analyzed either as being composed of koululais-
←←←←←←←←←
'student' and kuri 'discipline' or of koulu 'school' and laiskuri
←←←← ←←←←← ←←←←←←←←
'goof-off'. b. TEXFIN can also recognize some of the most common
derivational suffixes and it knows where they can occur. As the
following examples show, derivational processes are iterative.
User: vahvempia
TEXFIN: STRONG Adj Comparat Pl Part
In this case, only the root vahv 'strong' exists as an entry in TEXFIN's
←←←←
root lexicon with e as one possibility for the final vowel. The suffix
←
mp is one of the allomorphs of the comparative suffix. The same
←←
comparative suffix can also occur at the end of a string of other
derivational suffixes.
User: vahventajattomampia
TEXFIN: STRONG Adj NTA JA TON Comparat Pl Part
Here the addition of the suffix -nta to an adjective stem yields a
←←←
causative verb (vahventaa 'strengthen'), from which -ja derives an agent
←←←←←←←←← ←←
noun (vahventaja 'something that strengthens'). The last suffix -toma
←←←←←←←←←← ←←←←
before the comparatitive mpis one that turns nouns into adjectives
←←
(vahventajaton 'lacking something that strengthens'). Any other order
←←←←←←←←←←←←←
of these particular four derivational suffixes is correctly rejected by
TEXFIN as ungrammatical.
3. Analysis method
TEXFIN recognizes a word in a single left-to-right pass. It pursues all
possibilities in parallel in the manner of a breadth-first, all-paths
parser. Although most words at the end turn out to have only one
possible interpretation, in the early part of the analysis TEXFIN
typically has to consider many alternative hypotheses. For example, in
processing the string
m a t o n (Genit.Sg. of matto 'rug')
←←←←←
the program cannot be sure that it has encountered a form of matto 'rug'
←←←←←
and not a form of mato 'worm' until it has scanned through the entire
←←←←
word and recognized the the final n as the genitive singular suffix.
←
Every available analysis path is pursued until it either succeeds or
fails. This strategy requires a great deal of bookkeeping in the early
stages of an analysis but it enables TEXFIN to do all its work in a
single pass with no backtracking.
TEXFIN processes an input word by trying to match consequtive parts of
! Page 5
the string against entries in a number of separate lexicons. It accepts
a form just in case it can find all of the subparts in the the right
places. In the case at hand, TEXFIN begins by looking up all the
entries in the root lexicon that match an initial substring of the word.
Let us suppose that there are just two such entries:
ma Transl WORM Cat Noun Cont (t-d o)
mat Transl RUG Cat Noun Cont (t-0 o)
Table 4: Entries for mato 'worm' and matto 'rug' in the
←←←← ←←←←←
root lexicon
The root entry for each word contains the initial substring that all of
its forms have in common. In addition to the root itself, each entry
contains additional information about the word in the form of property-
value pairs. For example, the entries ma and mat in Table 4 list their
←← ←←←
translation (Transl) and syntactic category (Cat). They also indicate
what comes after the root string. This information on the required
"continuation" (Cont) is given by listing a sequence of alternation
patterns. The terms "t-d", "t-0", and "o" in Table 4 are names of
independent suffix lexicons that contain the letters or letter sequences
that make up a particular alternation pattern. Their role is very
similar to that of morphophonemes in structuralist analysis of Finnish
morphology. Entries in these suffix lexicons have the following format:
[suffix] [conditions] [features]
where [suffix] is some string of letters (possibly the empty string),
[conditions] is a description of the environment in which the suffix may
occur, and [features] is a list of whatever syntactic or morphological
properties are associated with the suffix. The last two parts are
optional; if there are no conditions associated with the particular
sufix, the word NIL appears as the second entry. This is the case for
both entries in the t-d lexicon:
d NIL (Grade Weak)
t NIL (Grade Strong)
Table 5: Example of an alternation pattern: t-d lexicon.
The two suffixes in this lexicon each bring in an additional feature to
the stem. When the root ma is continued with d to form mad, for
←← ← ←←←
example, it aquires the feature Grade Weak. This in turn has an effect
on what inflectional endings can be attached to the stem later on. The
genitive singular suffix n, for example, has (Grade Weak) as a
←
condition. The way in which these alternation patterns are used by
TEXFIN is very similar to the role morphophonemes play in structuralist
descriptions of Finnish phonology; it is a way of encoding the fact that
the word mato 'worm' has a morphologically conditioned internal
←←←←
! Page 6
alternation between t and d. In a structuralist description this would
← ←
be captured by postulating an underlying representation such as maTo and
←←←←
by giving rules that determine the spelling of T in various
←
environments.
The empty string as a suffix is represented by "-". For example, the
t-0 alternation pattern consists of the following two entries:
- ((BEF V)) (Grade Weak)
t NIL (Grade Strong)
Table 6: Example of an alternation pattern: t-0 lexicon
The condition (BEF V) on the empty string in Table 6 indicates that it
is an acceptable match in this lexicon only before a vowel.
The first step in the recognition of maton is to find all the items in
←←←←←
the root lexicon that match the beginning of the word. Let us assume
that the two roots in Table 4 are the only ones that meet the condition.
TEXFIN takes note of both possibilities and creates a separate record
for each. These records are updated and modified as the analysis
proceeds. A record has the following form:
[remaining characters] [accumulated features]
In this case, TEXFIN creates the following two records
(t o n) (Transl WORM Cat Noun Cont (t-d o))
(o n) (Transl RUG Cat Noun Cont (t-0 o))
Because both records specify a continuation, TEXFIN attempts to match
the remaining characters in the record against the entries in the
alternation lexicons. The initial t in the first record matches one of
←
the two entries in Table 5 and the second record fulfills the conditions
for the empty string in Table 6 because (o n) begins with a vowel.
TEXFIN now updates each record by removing the characters used in the
match and adds the new features obtained from the suffix lexicon. The
resulting records are:
(o n) (Grade Strong Transl WORM Cat Noun Cont (t-d o))
(o n) (Grade Weak Transl RUG Cat Noun Cont (t-0 o))
The next step in both cases consists of finding a continuation in the
o-lexicon. This leaves (n) as the remaining string. At this point
TEXFIN takes note of the syntactic category of the two records. Since
both are nouns, TEXFIN tries to find a plural marker in the beginning of
the remaining string. Not finding any, it marks them as singular. At
this point the two records look as follows:
(n) (Number Sg Grade Strong Transl WORM Cat Noun Cont (t-d o))
(n) (Number Sg Grade Weak Transl RUG Cat Noun Cont (t-0 o))
! Page 7
The last step involves looking for a match for (n) in the case lexicon.
Both records match the entry for genitive singular:
n ((Grade Weak)(Number Sg)) (Case Genit)
but only the latter record satisfies the condition for weak grade. Now
only one record remains:
() (Case Genit Number Sg Grade Weak Transl RUG Cat Subst
Cont (t-d o))
Because there remain no characters yet to be analyzed and the record has
all the features required of a noun, TEXFIN accepts it and prints the
values of some of the features as a final message:
RUG Noun Sg Genit
4. Comments on the method
One notable feature of TEXFIN is that it recognizes the allomorphic
variants of morphemes as such without attempting to transform them to
underlying canonical representations. It makes no use of operations
that delete, insert, or transpose segments. All conditions that
determine whether a stem can be followed by a particular suffix refer
only to phonological or morphological features of the items in question.
It would be possible to rewrite TEXFIN's linguistic analysis as a
standard phrase structure grammar, although we suspect that it would be
more difficult to implement efficiently in that format.
Since TEXFIN does not use transformational rules, morpheme structure
conditions or the like, it clearly does not express the kind of
linguistic generalizations about Finnish morphology that can only be
expressed by using such devices. There are, for example, 36 alternative
endings in the case lexicon for the illative, each of them with its own
environmental specification. Although these allomorphs are obviously
related to one another, TEXFIN does not exploit that relationship in its
recognition routine.
If it turned out to be possible to improve TEXFIN's performance by
making it linguistically more sophisticated, we would certainly do so.
There are many features that we could have designed differently and
could easily change if there was some compelling reason to do so.
However, the current design has a couple of important advantages that we
would not want to lose. First of all, TEXFIN uses the same data
structures both for recognition and generation and the two parts of the
program are very closely related. The set of strings it generates is
exactly the same that it recognizes. Secondly, the processing times are
fast enough for the program to serve as a front end for a more ambitious
program for syntactic analysis. And finally, because the root morphemes
are identified before anything else is done, the semantic information
associated with root morphemes become available at a very early stage.
In a system where lexical recognition is a part of a more comprehensive
! Page 8
interpretation process, it is thus possible to exclude some of the
alternatives very quickly because of syntactic or semantic
incompatibility with the preceding environment. It can be expected that
in such an integrated system the recognition of individual words will
take less time than it does in isolation. We regard this as a desirable
feature in TEXFIN because it is also a characteristic of human word
recognition.
Although we have no hard evidence to support our position, it does seem
to us, that word recognition as a psychological process is such a fast,
automatic routine that it probably does not involve in any significant
way the kind of higher-order knowledge about the structure of the
language that linguistic descriptions of morphology usually try to
capture. We think the current version of TEXFIN as a model of such a
process. In order to really represent the linguistic competence of
speakers of the language, the program would have to be augmented with
more sophisticated knowledge about the language. There are many
activities for which such general information about the language is
essential; for example, incorporating new words in the lexicon,
recovering from errors, etc. We are currently working on some ideas on
how to expand TEXFIN in this direction.
-------
∂12-Nov-81 1616 FFL
To: JMC, FFL
A researcher from Paris VI came in to see you. He will be here until Sunday
and would appreciate talking with you. He is calling me at 11:30 a.m.
Friday to see if you will be in in the afternoon and if he can catch you for
a few minutes. Would you please let me know if you will not be in. Thanks.
∂12-Nov-81 1614 SIS Computer Science Colloquium for November 16 - 20, 1981
To: "@COLLOQ.DIS[INF,CSD]" at SU-AI
Date Place Person
Day Event From
Time Title
-------------------------------------------------------------------------------
11/16/81 Math 380C William Gropp
Monday Numerical Analysis Stanford University
4:15 p.m. Seminar ``Computer Science Issues For Scientific
Computing''.
11/17/81 MJ301 William J. Clancey
Tuesday Knowledge Stanford University
2:30 p.m. Representation ``Recent Studies in Expert/Novice Problem
Group Seminar Solving''
11/17/81 Jordan 041 Robert Wilensky
Tuesday Computer Science UC Berkeley
4:15 p.m. Colloquium ``A Model of Planning In Complex Situations''.
11/18/81 Terman 153 B.R. Rau
Wednesday Computer Systems ESL Inc., San Jose
4:15 p.m. Laboratory Seminar ``The Polycyclic Architecture''.
11/20/81 Gazebo Professor Edward Feigenbaum
Friday Siglunch Stanford University
12:05 p.m ``The Japanese 5th Generaton Computer System''.
11/20/81 MJ301 Francisco Corella and Jerry Kaplan
Friday Database Research
3:15 p.m. Seminar ``Cooperative Rsponses to Ad-Hoc, Boolean Queries
In A Large Bibliographic Retrieval System''.
∂12-Nov-81 2141 Brian K. Reid <CSL.BKR at SU-SCORE> Re: equip.
Date: 12 Nov 1981 2141-PST
From: Brian K. Reid <CSL.BKR at SU-SCORE>
Subject: Re: equip.
To: CSD.ULLMAN at SU-SCORE, csl.jlh at SU-SCORE, csl.jhc at SU-SCORE,
csd.cheriton at SU-SCORE, csl.lantz at SU-SCORE, engelmore at SUMEX-AIM,
rpg at SU-AI, jmc at SU-AI
cc: csd.golub at SU-SCORE, csd.feigenbaum at SU-SCORE
In-Reply-To: Your message of 11-Nov-81 1442-PST
Can I assume that since Scribe is the formatting language in use by the
majority of people on this committee, and also the one in which it is
easiest to combine the work of multiple authors, that we will be doing
the proposal in Scribe?
-------
∂13-Nov-81 0849 Kanerva at SUMEX-AIM Cash problems
Date: 13 Nov 1981 0849-PST
From: Kanerva at SUMEX-AIM
Subject: Cash problems
To: McCarthy at SU-AI
I understand that the KL10 has a serious problem. Would it make sense
for Stanford to buy Foonlies for running things that no longer change
much? Context fits that bill.
Building on Foonlies may require local hadware talent, but we have
exceptional people like George Blanck and Tom Dienstbier who want
to stay with the university. - Pentti
-------
My impression is that the users of Foonlies are reasonably satisfied.
If their F4 (said to be equivalent to a KL) satisfies its first
customers, then I would consider them very price competitive. I wouldn't
presume that their maintenance (whatever they agree to contract for)
would be inadequate so close to their home base.
∂13-Nov-81 1056 Sharon.Burks at CMU-10A Your visit next week
Date: 13 November 1981 1355-EST (Friday)
From: Sharon.Burks at CMU-10A
To: JMC at SU-AI
Subject: Your visit next week
Message-Id: <13Nov81 135541 SB40@CMU-10A>
We are looking forward to your visit and to your lecture next week.
I would like to set up a schedule for you while you are here, and it would
be helpful if I knew exactly when you plan to arrive and depart. Could you
let me know by return message??
If there is anyone you are particularly interested in talking with while you
are here please let me know that as well.
Thanks,
sharon
I will arrive Monday evening on Us AIR 841 from Akron at 6:50, and
I will leave on Tuesday evening on UA 823 to Los Angeles at 10:30pm.
I won't need hotel reservations as I will stay Monday night with
Ed Fredkin. I would like to talk to Allen Newell and Raj Reddy and
anyone else who wants to talk with me. I will be available all day
Tuesday from (say) 10am.
∂13-Nov-81 1054 CLT driveway
could you please do something about the thing sitting with the
garbage cans? I am not pleased about having scraped the side of
my car this morning on garbage cans that weren't put where they
belong.
Thanks
∂13-Nov-81 1326 Konolige at SRI-AI meeting
Date: 13 Nov 1981 1326-PST
From: Konolige at SRI-AI
Subject: meeting
To: jmc at SAIL
John, burritos were late today, so I should arrive at Stanford
about 2pm. --kk
-------
∂13-Nov-81 2248 Danny Hillis <DANNY at MIT-AI>
Date: 14 November 1981 01:45-EsT
From: Danny Hillis <DANNY at MIT-AI>
To: MINsKY at MIT-AI, DANNY at MIT-AI, rah at sU-AI, llw at sU-AI,
jmc at sU-AI
Fellow Conspirators,
Lunar pickings are slim indeed. The place makes death valley
look like an oasis. There are only a few things of any use lying
around: oxygen, aluminum, silicon. Notable missing are hydrogen and
carbon. Also nothing highly reactive, like florine or chlorine. Makes
for boring chemistry.
A settlement must expand to prosper. The bulk materials noted
above would provide sheer mass for such expansion, but we will have to
keep feeding in the fertilizer. This is a costly proposition. The
settlement can help by reducing the cost of resupply or by producing
something of value.
Here are some possible way for the lunar pilgrims to earn their keep:
INFORMATION
1. Pop Press coverage. (TV specials, interviews, etc.)
2. Data for NAsA (medical data, contracted experiments)
Comment: These are readily salable items, but with limited markets.
Easiest to do at first, but not a long term solution.
MATTER
3. Bulk materials (sand) delivered to earth orbit. Being in a
relatively shallow gravity hole, we can throw things up at less energy
cost. A kevlar sling is enough to get it out, although getting it to
where you want is harder.
Comment: Relatively low tech, but who is the customer? This seems to
depend on something like sPs.
4. Oxygen delivered to earth orbit.
Comment: Easier to sell, but harder to make. Has the additional
advantage of lowering shipping costs to the moon.
ENERGY
5. We have cheap, but intermittent power. A polar station could have full-time
power.
Comment: I can't think of easy customers.
LABOR
6. We could offer to build a base for someone else, on contract.
Or an antenna. Or a telescope.
Comment: Once we are established, it will be cheaper and less risky
for a government to hire us than it to do it themselves. That assumes
that they want the result, not just the status of having done it.
All told, a pretty meager list. suggestions?
-Danny
∂13-Nov-81 2313 CLT SEMINAR IN LOGIC AND FOUNDATIONS OF MATHEMATICS
To: "@LOGIC.DIS[1,CLT]" at SU-AI
SPEAKER: Prof. Stan Wainer, University of Leeds
TITLE: "The slow-growing hierarchy"
TIME: Tues. November 17, 4:15-5:30
PLACE: Room 383-N, 3rd. floor Math. Bldg.
∂14-Nov-81 0031 Jerry E. Pournelle <POURNE at MIT-MC> report
Date: 14 November 1981 03:30-EST
From: Jerry E. Pournelle <POURNE at MIT-MC>
Subject: report
To: JMC at SU-AI
cc: POURNE at MIT-MC, llw at S1-A, rah at S1-A
No disagreement.
Date: 13 Nov 1981 1751-PST
From: John McCarthy <JMC at SU-AI>
I think the phrase on page 2 of the summary "obviate the need for neutron
weapons in Europe" should be replaced by "give the President alternatives
to the use of nuclear weapons in Europe". The history of defense debates
is full of the use of hypothetical weapons systems to clobber existing
weapons systems or systems under development. The original phrase may
unnecessarily generate opposition from someone who would favor the kinetic
weapons but might fear that their advocacy would be used to prevent what
he believes to be necessary in the short term. Someone who wants to use
your proposal as a means of clobbering other proposals is a weak ally,
because he'll desert you as soon as his purpose is accomplished or shown
to be infeasible.
∂14-Nov-81 0032 Jerry E. Pournelle <POURNE at MIT-MC>
Date: 14 November 1981 03:32-EST
From: Jerry E. Pournelle <POURNE at MIT-MC>
To: jmc at SU-AI
Who is RAH @s1-a? Rod Hyde? I get no answer when I go WHOIS
RAH@S1-A but I don't know why.
RAH is indeed Rod Hyde. I don't know that whois works for non its sites.
What works from here is FINGER.
∂14-Nov-81 1929 Lowell Wood <LLW at S1-A> Grubbing For Sustenance, cont'd.
Date: 14 Nov 1981 1920-PST
From: Lowell Wood <LLW at S1-A>
Subject: Grubbing For Sustenance, cont'd.
To: danny at MIT-AI
CC: LLW at S1-A, RAH at S1-A, minsky at MIT-AI, jmc at SU-AI
∂13-Nov-81 2241 Danny Hillis <DANNY at MIT-AI>
Date: 14 November 1981 01:45-EST
From: Danny Hillis <DANNY at MIT-AI>
To: MINSKY at MIT-AI, DANNY at MIT-AI, rah at SU-AI, llw at SU-AI,
jmc at SU-AI
Fellow Conspirators,
Lunar pickings are slim indeed. The place makes death valley
look like an oasis. There are only a few things of any use lying
around: oxygen, aluminum, silicon. Notable missing are hydrogen and
carbon. Also nothing highly reactive, like florine or chlorine. Makes
for boring chemistry.
A settlement must expand to prosper. The bulk materials noted
above would provide sheer mass for such expansion, but we will have to
keep feeding in the fertilizer. This is a costly proposition. The
settlement can help by reducing the cost of resupply or by producing
something of value.
Here are some possible way for the lunar pilgrims to earn their keep:
INFORMATION
1. Pop Press coverage. (TV specials, interviews, etc.)
2. Data for NASA (medical data, contracted experiments)
Comment: These are readily salable items, but with limited markets.
Easiest to do at first, but not a long term solution.
MATTER
3. Bulk materials (sand) delivered to earth orbit. Being in a
relatively shallow gravity hole, we can throw things up at less energy
cost. A kevlar sling is enough to get it out, although getting it to
where you want is harder.
Comment: Relatively low tech, but who is the customer? This seems to
depend on something like SPS.
4. Oxygen delivered to earth orbit.
Comment: Easier to sell, but harder to make. Has the additional
advantage of lowering shipping costs to the moon.
ENERGY
5. We have cheap, but intermittent power. A polar station could have full-time
power.
Comment: I can't think of easy customers.
LABOR
6. We could offer to build a base for someone else, on contract.
Or an antenna. Or a telescope.
Comment: Once we are established, it will be cheaper and less risky
for a government to hire us than it to do it themselves. That assumes
that they want the result, not just the status of having done it.
All told, a pretty meager list. Suggestions?
-Danny
[With respect to mass, the most crucial item would seem to be hydrogen, as
we need it for the water which humankind loves to wallow around it *and*
for input to the RL-10s which will serve as our long-range point-to-point
transport on the surface as well as to *all* other points in Earth orbit
and on out to the Martian moons and the Belt. We've *got* to have it, and
the only question is how hard it is to get it. The solar wind has been
pounding the lunar surface with something of the order of a billion
protons per square centimeter-second for 5 eons, or a hundred moles per
square centimeter since the game began. Only the highest energy portion
of this could bury itself sufficiently deeply in the surface dust to
stick, but 1) I'm sure that at least a thousandth of it has (as that's
about the fraction that has the 10 kev energy necessary to dig into the
dust suface, if the solar wind over geologic time has been basically the
same as it is today) and 2) I'm certain that I've read that the surface
dust is supercharged with hydrogen (as rational expectation suggests).
Dig deeper into the literature, Sir!
The other two items that we'll have a real hankering for are carbon and
nitrogen. There are some low Z carbides and nitrides that are stable, so
there's got to be a non-negligible amount of it present in some minerals;
are you quite certain that none of the Apollo-retrieved rocks had such
inclusions in them? If worse comes to worse, we can import them (perhaps
as HCN, or {CN}2) and conserve them carefully, but not being able to
provide yourself with the materials needed for basic life support in the
long haul is a real political/psychological handicap.
As for what to do, the Manifest Destiny gambit seems to be the best one
for the near term (survive, expand your lebensraum, `reproduce and
multiply and subdue the moon'), supplemented by lots of scientific reseach
and knowledge-gathering--since it's a virtually brand-new environment,
everything you do is research, and everything you notice, measure and
document is new knowledge; you can't go wrong. What we need are rational
priorities and then details of what it takes (people of what flavors and
numbers, materials, etc.) to do what, and when to do it (and maybe a bit
of why).]
∂15-Nov-81 0008 Lowell Wood <LLW at S1-A> Base Powering
Date: 14 Nov 1981 2359-PST
From: Lowell Wood <LLW at S1-A>
Subject: Base Powering
To: jmc at SU-AI
CC: LLW at S1-A, RAH at S1-A, danny at MIT-AI,
minsky at MIT-AI
∂14-Nov-81 1256 JMC via SU-AI nuclear power source
To: LLW, RAH
Suppose there were no political or regulatory obstacle to
building a nuclear power source for the moon and getting it
there. Would it be economically competitive with solar
at 10 to 50 kilowatts taking into account development costs?
[John: Studies to date indicate that nuclear begins to compete with solar
at about 100 kWe for most space applications. I would believe that you
might push the indifference point down to 50 kWe if you worked hard on
squeezing mass out of the reactor and its radiator, but this would drive
the R&D costs and time scales way up. Also, these numbers are for
unshielded reactors, which are fine for space, but which would soon
activate their lunar surroundings with leakage neutrons to a regrettable
extent; even light shielding would put the indifference point into the
multi-megawatt regime.
On the other hand, solar is here-and-now, and minor mass economization
(e.g., in structure, which is unimportant on the moon) might push the
indifference point up to 200-300 kWe. However, nuclear delivers all the
time, while solar works only during the lunar day, and at peak rates only
for a few days during the month.
Politics and regulation aside and considering only time and cost, I
strongly suspect that a prime power system consisting of solar panels
(sheets, film, or whatever is lightest) deployed onto the surface manually
and inputting excess power to water electrolytic cells whose output goes
to small cryostat-cooled dewar flasks (the RL-10 fuel tanks?) hooked up
to hydrogen-oxygen fuel cells and inverters will turn out to be *the*
preferred way to go, for our purposes. (Danny: The new--and slightly
unreliable in zero-g--Shuttle fuel cells mass 97 kg in about 4 cubic feet
of volume, deliver 7 kWe continuous--and 12 kWe for 15 minute periods no
more frequently than every two hours--with something of the order of 70%
efficiency continuous and 50% at peak rating; they also produce potable
water.) Lowell]
∂15-Nov-81 0030 Lowell Wood <LLW at S1-A> Division Of Labor
Date: 15 Nov 1981 0022-PST
From: Lowell Wood <LLW at S1-A>
Subject: Division Of Labor
To: RAH at S1-A, danny at MIT-AI
CC: LLW at S1-A, jmc at SU-AI, minsky at MIT-AI
It was agreed this past week to divide up the documentation of the plan
along the following lines:
Rod: How to get there and what to take with you; what steps you
might/have to take along the way (LEO base, etc.). Mass and equipment
budgets, staffing levels and alternatives; enhancement options and
follow-ons to initial beacheads. Clear identification of what definitely
exists, what probably presently or will then exist, and what we'll
absolutely have to develop ourselves
Danny: What to do once you've got there; priorities and schedules for
these activities; support and re-supply needs for various options; base
expansion/replication rates; pay-back mechanisms (including
quantifications as appropriate) of various flavors
Lowell: Political and financial considerations; ways and means; initial
sequencing; alternate routes and fall-back ploys; scheduling of first
steps and constraints thereon
Marvin and John: Constant constructive criticism; flogging of laggards;
oversight elimination and comprehensive plan enhancement
It was agreed to have a first draft by the end of the Thanksgiving weekend
for circulation and comment among ourselves, and to re-work this into a
final draft for commentary by discreet-and-knowledgable friends
immediately after recovery from New Year's.
Lowell
Even before I got this message, I was about to advocate asking Carolyn
Talcott to serve on the MTC-qual committee. She has taken the exam
(as given by Zohar and Jim Morris) and passed outstandingly, and she
knows as much of the literature in MTC (the verification part) as anyone.
∂15-Nov-81 2014 WOL Message from Zohar Manna
John:
I talked to Zohar on the phone today (he is in Israel) and he asked me to
send you the two following messages:
(1) Zohar will probably not be here in December as he had previously planned.
He will thus not be able to be on the MTC-qual committee.
(2) He met Ehud Shapiro and was quite impressed with him. If you are still
considering him his recommendation is favorable.
Pierre Wolper
∂15-Nov-81 2337 Rod Hyde <RAH at S1-A> We got the first four microfiche documents you sent. The next two,
Date: 15 Nov 1981 2316-PST
From: Rod Hyde <RAH at S1-A>
Subject: We got the first four microfiche documents you sent. The next two,
To: danny at MIT-AI, minsky at MIT-AI, jmc at SU-AI, LLW at S1-A
CC: RAH at S1-A
the LLV & the 1965 RL10, haven't come yet, but we got them thru John.
He also was able to get some life support reports.
Of the 3 RL10 papers the 1965 is best. The 1976 & 1980 are nice reading,
but detail high cost modifications that they want to do for NASA. They
pick up 40 seconds of Isp ( about 9 % ) and some low thrust options, at
the cost of about 200 million $ and a bigger nozzle, and heavier engine.
The 1965 report is when they were trying to get the full Apollo mission-
they wanted to use 6 RL10's for the SIVB, and RL10's for the SPS & LEM
also. So for the LEM mission they had to be continuously throttleable.
They built & tested versions which were throttleable. We should use
this version- the payload difference from their 1980 top of the line
proposal, not hardware, is about 15% payload.
The Apollo propulsion report is moot, since the RL10 is clearly best.
The NAR lunar base is a 12 man scientific design. As such it is not
what we want, but their life support system was modular for 4 man groups,
so is relevant. They recycled air & water but not food. They had a
hardware mass of about 1 ton for a 4 man system. This excludes food &
makeup air-they leaked a lot. The report lists masses for all the basis
components & the alternatives, but gives no info on where the numbers
came from.
The LLV report will have some good stuff in it for the ship design, but
was unmanned so of no use on the life support end. They used a RL10
and got around the throttleability by dumping it off just before
touchdown & finishing up with a hypergolic storeable set. I haven't gone
over this enough yet, but their guidance system stuff, landing gear, etc.
should be useful.
One of the life support documents ( SP374 ) was quite good, although it
did not have much in the way of hardware masses.
I've got hardcopy of the 1965 RL10 report & the relevant portions of the
NAR base report made up that we can send you two if you want them.
I don't have any location desires yet, so feel free to choose. See if there
is a peak in C,H, or N at any of the sites.
∂17-Nov-81 1019 FFL
To: JMC, FFL
Gorin and Keller would like you to come to a SAIL Payers meeting on
Nov. 24, 3 p.m. I have put it on your calendar.
∂17-Nov-81 1121 REG SAIL Payers Meeting
To: "@PAYERS.DIS[CSD,REG]" at SU-AI
There will be a meeting to figure out what to do about paying for SAIL
on Tuesday, Nov. 24 at 3 pm in room 252 MJH.
∂17-Nov-81 1352 FFL
To: JMC, FFL
III called to say that Julius Koppelman will not be attending the meeting of
the Board on Wednesday.
∂17-Nov-81 1424 FFL
To: JMC, FFL
Prof. Golub has asked you to attend a meeting on Thurs., Nov. l9, at 2:30 p.m.,
re V. Pratt's appointment.
. Is on your calendar.
∂17-Nov-81 1915 Oded Anoaf Feingold <OAF at MIT-MC> energy digest #3004
Date: 17 November 1981 22:10-EST
From: Oded Anoaf Feingold <OAF at MIT-MC>
Subject: energy digest #3004
To: energy
Unpaid costs of energy
Nuclear power (health hazards of)
----------------------------------------------------------------------
doug humphrey@MIT-AI (Sent by DIGEX@MIT-AI) 11/15/81 23:03:25
Re: Unpaid Costs of energy....
To: energy at MIT-MC
The book you mention, "The unpaid Costs of Electrical Energy " by
William Ramsey is a course text at the University of Maryland is PHYS 318,
"Risks of Nuclear power". This is a very interesting course, looking at the
technical and environmental problems not juswt with nuclear, but also with the
other sources of energy generaly listed as possible replacements for
nuclear. The book seems to have a little pro-nuke bias, but when you weigh
all of the factors considered, nuclear fares pretty well on its own merits.
The environmental impact of coal is fully documented, and much worse than
nuclear power, even assuming the occaisional total meltdown. It will most
likely get knocked by anti-nuke types, but it is hard to ignore.
DH
------------------------------
Date: 16 Nov 1981 2248-PST
From: Jim McGrath <JPM at SU-AI>
Subject: Nuclear Power
To: energy at MIT-MC
I have been reading up on health hazzards of nuclear and coal power
recently, and would like to present some numbers to this community.
COAL: The biggest unknown here is the exact effect of air pollution
on health. However, yearly deaths from all sources has been estimated
from 6,000 to 30,000.
NUCLEAR: the yearly loss of life is small - estimates range from tens
to 400 or so people for the entire power cycle. However, potential
losses from severe accidents are rather high - the highest figure I
have seen is 50,000.
These numbers seen to be generally accepted by everyone who has
produced decent research on the subject. Since precise figures
would be misleading (i.e. local factors can come to dominate
any detailed calculations), it is probably reasonable to state
that the normal operating costs of these power sources measured
in number of lives is about 10,000 for coal and 200 for nuclear.
I would appreciate discussion on these figures. More accurate
ones are sought. But I would really appreciate references and
reasons for radically different figures.
Let P = probability of a severe reactor accident per year.
Assume such an accident costs 30,000 lives. Now solve for
P such that 10,000 = 200 + 30,000*P. ie P = .327 (approximately)
Assume a complete nuclear economy requires 1000 reactors.
Then as long as the major accident probability is less than
.000327 per reactor year (about one out of three thousand),
a nuclear economy wins. No one I know of has seriously proposed
that major accident rates are anywhere near that high.
This discussion has deliberately simplified things. I do not
think this is bad, since local circumstances would otherwise
dominate any serious discussion of the health risks. But
it appears that on a health basis nuclear power is, in
general, at least an order of magnitude safer than coal.
Does anyone disagree with this general conclusion? If so,
why?
Jim
(end)
∂17-Nov-81 2239 Bob Engelmore <CSD.ENGELMORE at SU-SCORE> Partial Proposal Draft
Date: 17 Nov 1981 2237-PST
From: Bob Engelmore <CSD.ENGELMORE at SU-SCORE>
Subject: Partial Proposal Draft
To: jmc at SU-AI, rpg at SU-AI, rindfleisch at SUMEX-AIM
The following is a partial proposal draft, which contains material for the
AI part, and also shows the slots which need to be filled in. Section 1
should be a concise summary of the rest of the proposal, and one of us can
do that last. Section 2 is pretty close to complete, I think, for the AI
related groups (the Feigenbaum - McCarthy axis), but, as I said, is only a
draft and I'd appreciate criticism on it. The CSL people can read the
section for structure and style, if that will help. Section 3 should contain
our shopping list. Would anyone like to volunteer to do that section?
The rest is straightforward and non-creative.
Bob
@make(report)
@style(indent 0)
@heading(Proposal for
Computer Equipment
Modernization
DRAFT
@value[date])
@newpage
@section(Summary)
@section(Background and Technical Need)
@subsection(ARPA-sponsored research related to this proposal)
This proposal addresses the combined equipment needs of ARPA-sponsored
research which is conducted at Stanford University in the Computer
Science Department (CSD) and in the Electrical Engineering Department's
Computer Systems Laboratory (CSL). CSD research that will benefit from
equipment refurbishing proposed here includes basic Artificial Intelligence
(AI) research in Formal Reasoning, Knowledge Representation and
Acquisition, development of Knowledge Engineering tools, and advanced
AI applications in the areas of user interfaces and knowledge-based data
management systems (KBMS). Relevant CSL research includes network-based
graphics for VLSI design and distributed systems architectures.
As discussed in more detail below, these research projects are being
seriously hampered by a lack of adequate computing facilities, including
basic CPU cycles, addressble memory, secondary and tertiary storage, network
hardware and software, and high-bandwidth display terminals.
@subsection(Artificial Intelligence Research)
Basic and applied research in Artificial Intelligence is a key component
of the ARPA sponsored research in the Stanford University Computer
Science department. Much of that research has an empirical thrust which
places heavy demands on our computing systems as we investigate new
methods of computer-based reasoning, new knowledge representation
languages, new tools for building expert systems, etc. These efforts
typically involve the development of programs and systems written in the
Lisp programming language. Several ARPA-sponsored projects use Lisp
nearly exclusively: the Heuristic Programming Project, Image
Understanding, Formal Reasoning, and the Verification Project. The
flexibility of the Lisp language, and the excellent programming
environment that exists with the major Lisp dialects, make it an ideal
vehicle for rapid development of, and experimentation with new AI
concepts. However, as our projects grow in both complexity and in
number, each has run up against the barriers of available computing
cycles and addressable memory with our current facilities.
@paragraph(Current Facilities)
The single most important bottleneck to rapid progress in empirical
research in AI (e.g. Knowledge Engineering) is the lack of adequate
computer power. AI research is now being conducted at Stanford
primarily using three time-shared computer facilities. One is the
Stanford Artificial Intelligence Laboratory (SAIL) DEC 1080 facility
that has been the mainstay of the Formal Reasoning, Image Understanding,
Hand-Eye and Verification projects since 197x. The SAIL machine is
generally fully utilized, and maintenance costs and down time are rising
as the machine components age. Its cost/effectiveness ratio is
sufficiently high compared to modern mainframes that this facility will
need replacement within two years.
The second facility is SUMEX, a dual-processor KI-10. This facility has
been used for ARPA-supported research in the HPP for approximately seven
and a half years. Although SUMEX is a fully funded facility supported
by the NIH, ARPA research is permitted in exchange for SUMEX access to
the ARPANET. HPP projects receive about 40% of the computing power of
the facility, the rest being distributed to other Stanford projects and
to other projects aroung the nation. Half of the "HPP 40%" is devoted
to the NIH-supported DENDRAL project, so that only a small slice of the
facility is available for ARPA-sponsored work. The Sumex facility is
slow, somewhat unreliable by modern standards, and of a now-obsolete
technology. By programming around the address space limitations (see
below), very heavy demands are placed upon the system. Consequently,
the SUMEX facility is so heavily used that little or no Lisp program
development can be accomplished on it during normal working hours.
The third facility is SCORE, a DEC 2060, which is used by the Computer
Science Department, The EE Department's Computer Systems Lab and
Integrated Circuits Lab, and the Operations Research Department. SCORE
is used extensively, by hundreds of researchers, support staff and
students, and generally runs at a load average which is uncomfortably
high for Lisp program development.
All three of the above facilities have 18 bits of address, and since the
word length is 36 bits, there are on the order of 2↑18 cons cells
available to the user (256K words). That address space is insufficient
for the programs that AI researchers are developing and testing in the
1980s.
In 1980 the HPP obtained a VAX 11/780 from ARPA to pursue research in
knowledge-based VLSI design. The facility is now also used for our
ARPA-sponsored research in developing methods for Intelligent Agents.
Although the Vax provides an extended address space, it has not yet been
a satisfatory engine for Lisp-based work, as only one dialect, Franz
Lisp, is running on it, and this dialect provides few of the features
one expects of a high-quality Lisp system (see below).
Therefore, the Lisp computing needs of the ARPA research projects are at
a critical point (and have been for some time). There are several
interesting possibilities on the horizon at the moment in terms of
hardware and/or software that will help solve these problems, and the
proposed plan represents a best attempt to chart a sensible course of
action with respect to these possibilities. A major area of concern in
developing the plan has been the future development (and survival) of
the principal Lisp systems currently in use.
@paragraph(On the matter of Lisp dialects.) There are two main dialects
in contention at the moment for the attention of the AI community:
InterLisp and Common Lisp. The former is a groundbreaking Lisp in that
it introduced the AI community for the first time to a modern
programming environment. With many features to ease the system building
burden, researchers can create and test programs that previously would
have been considered far too ambitious. Many of the existing AI systems
at Stanford are written in InterLisp. In particular, nearly all of the
HPP programs are written in InterLisp, and the continued use of these
programs is vital to the success of several HPP projects.
Common Lisp is the distillation of the MacLisp and Lisp Machine Lisp
experiences, coupled with the level-headedness of the SpiceLisp design.
MacLisp and its descendants chose the route of efficient implementation
to bring the community a Lisp that could compete with less esoteric
programming languages, such as FORTRAN, in such tasks as severe numeric
computation. However, as time progressed, the lessons of InterLisp
asserted themselves, and a powerful computing environment was incorporated.
Many of the novel language features of Lisp trace their roots to
the innovative style of the MacLisp community.
The Formal Reasoning group develops programs in MacLisp, as does the
Image Understanding and Program Verification groups. This includes such
programs as the Advice Taker, ACRONYM, the Pascal verifier and the ADAM
compiler.
Since there is a strong commitment to both branches of the Lisp family
tree, the proposed equipment acquisition has been structured to permit
Lisp-based research and development to proceed in both dialects.
@paragraph(Requirements)
The high demand for "Lisp cycles", coupled with the problems of heavily
loaded, aging computer facilities to support our research, make it
imperative that we significantly refurbish and augment our existing
facilities. Our principal requirements are:
@begin(enumerate)
Single user Lisp workstations. Although the current cost of single user
Lisp workstations is currently high compared to the average
capitalization cost per researcher, these machines are already a
cost-effective way to obtain Lisp cycles and extended address space for
meeting current needs, and all signs point to lower unit costs and/or
higher unit performance within the next one to two years. Because the
price/performance issue is so rapidly changing for this class of
hardware, we believe the prudent course is to defer the choice of
specific machines until we have the resources to purchase them. The
budget assumes a unit cost of $60K, which is the approximate cost of
the Xerox EOS 1100 system and the projected cost of the Symbolics 3600
machine which should be available in early 1982.
Central time-shared facility. At minimum, a replacement for the SAIL
computer will be needed by FY1984. By that time, we expect that a very
powerful machine such as an S-1 or a DEC-2080 will be available, at a
cost of approximately $650K (cpu only). We foresee that at least half
of this machine will be used for AI research, including Lisp programming
(mostly non-HPP research), programming in languages other than Lisp, and
for text editing and formatting for technical publications and program
documantation. Consequently we have planned for acquisition of an S-1,
to be funded equally by ARPA and non-ARPA sources.
Central File Server. The AI research projects have a critical need for
on-line access to very large program and data bases. The need will be
particularly obvious as we move in the direction of personal Lisp
workstations, because the individual workstations do not have adequate
disk space to support group project requirements. Moreover, to maximize
utilization of the workstations, individual researchers should be free
to do their work on any available workstation, which can only be done
through a central file server.
High-quality printers. The high utilization of the Dover indicates that
high-quality printers are playing an increasing role in the conduct of
research in the Computer Science Department, and the AI project groups
are no exception to that rule. A laser printer is needed by the HPP for
Sumex-related work, and an additional laser printer is required in
Margaret Jacks Hall to back up the Dover.
Ethernet augmentation. As locally networked computing systems
proliferate, there is a concomitant need for expanded ethernet hardware
and gateways.
@end(enumerate)
@paragraph(Alternatives)
@b[Time-sharing:] Large, essentially time-shared computers with larger
address spaces is one alternative that may be able to support Lisp
computing. Specific possibilities are: the DEC 2080, the DEC 11/780, the
DEC 11/750, and the LLNL S-1. Of these, only the DEC 11/780 and DEC
11/750 are currently available. The DEC 2080 and the LLNL S-1 may or
may not have a suitable Lisp running on them when they come into
existence.
The DEC 11/780 (Vax) has a 32 bit address space of 8 bit bytes which the
hardware is able to address. This is reduced to 30 bits by operating
system needs. Further, since it takes 2 words per non-cdr-coded cons
cell, there are 27 bits of cons cells available.
A major problem with the the DEC 11/780 is that the page size (512
bytes) is very small and, consequently, paging overhead, especially in a
potentially non-localized language like Lisp, is enormous. Coupled with
the fact that the machine is a 1 MIPS machine (1/3 of a KL-10), the DEC
11/780 would be expected to support 4 or 5 Lisp jobs adequately unless
many difficult optimizations are performed. At the current time there is
only one Lisp that runs on the Vax, namely Franz Lisp from Berkeley. Our
experience with Franz Lisp is that it is poorly engineered as a
language and that the Vax simply cannot handle more than a few Lisp jobs
running at a time. The Masinter report (unpublished) and personal
conversations with Jon L. White at MIT lead one to believe that neither
Interlisp on the Vax nor NIL on the Vax will be much improved over this
situation.
VAX NIL, written at MIT, and VAX Interlisp, written at USC-ISI, are
about to become available for the VAX machines. There has not been
sufficient experience with either dialect yet to judge their
acceptability. It is anticipated that VAX NIL will be modified into a
Common Lisp fairly soon after the language design for the latter is
completed; since Common Lisp and NIL descend from the same Maclisp root,
this modification should require relatively little effort (approx. 6 man
months) to accomplish.
The DEC 11/750 exhibits about 60% of the performance of the DEC 11/780
and is therefore not a serious contender as a time-shared source of Lisp
computing. However, since a large proportion of Lisp code involves
memory references and indirect moves, the DEC 11/750 is actually
expected to perform at 70% of a DEC 11/780, due to the faster
busses on the DEC 11/750. We are proposing to purchase 5 SUVAX
configurations at the special DEC reduced price (under the special
arrangement with ARPA), partly in order to evaluate this machine with
respect to Lisp performance.
The DEC 2080 is projected to be a very large machine in terms of
performance, being a 24 MIPS machine nominally. However, given the size
of the cache memory it is doubtful that it will achieve more than 10
MIPS in practice on non-numeric computing. More interestingly, it will
support the extended addressing mode now available on the DEC 2060, but with 30
bits of address rather than 23 bits. This will yield 29 bits of cons
cells given a non-cdr-coded scheme. The main drawback is that there is
no Lisp that will suit the needs of the Projects that will be able to
run on this machine when it exists. The closest contender is Rutgers
Elisp, which is an extended addressing DEC 2060 version of UCILisp. The
mode of addressing is through the mechanism of indirection, and the
address space is broken up into sectors that may be active at any given
time. Hedrick at Rutgers claims that a compiler and an assembler for
Lisp could be constructed that would automate the sectoring of Lisp
code. Moreover, Hedrick is currently considering an implementation of
Common Lisp for this hardware based on Elisp.
The LLNL S-1 is a large multi-processor system being constructed at
Lawrence Livermore National Laboratories by, largely, former Stanford
people. Much of the software development is being done by contract to
the Stanford Computer Science Department. There is an excellent chance
that the department will be able to obtain an S-1 uniprocessor within 2
years. The S-1 is a 20 MIPS machine, but with the use of very large
cache memories (16K words of instruction and 16K words of data) there is
the likelihood of obtaining 15 MIPS in practice. The address space is 29
bits of words with 28 bits of cons cells. It is a tagged architecture,
which is an advantageous feature for Lisp systems.
Common Lisp (S-1 NIL) will be running on the S-1 when the machine
becomes available. That Lisp will use specialized instructions for
numeric computations, making it a highly attractive alternative for
Robotics and Vision work in Lisp. However, the S-1 will run only
custom software in that, except for the Lisp, there will be no attempt
to run any code that can run on any other machine. Thus, there are both
hardware and software maintenance problems for this machine.
The decision to purchase an S-1 or a DEC 2080 will be made in 1983 when
there is better information available on them. A large, time-shared
machine that can replace SAIL and provide high quality computing to the
general Computer Science Department as well as Lisp cycles for the ARPA
research community is planned, and either the S-1 or the DEC 2080 will
be purchased for that purpose.
@b[Personal Lisp machines] The only proven alternatives to large,
time-shared facilities involve the personal computer route. Personal
machines offer advantages in terms of guaranteed performance (once
available) and disadvantages in terms of group interaction in large
projects. Nevertheless, since they are the only sources of Lisp cycles
at the moment that satisfy Lisp address space needs, they must be
considered.
At this point there are 3 alternatives, 2 live and 1 possible. The live
alternatives are XEOS (Xerox) Dolphin and the Symbolics Lisp Machine.
Unfortunately, these machines run different Lisps, and the decision
about hardware now must move strongly into the software domain. Dolphins
run InterLisp and Symbolics Lisp Machines run ZetaLisp. InterLisp enjoys
a high software investment at Stanford, particularly within the HPP
project. MacLisp has a high software investment in the Formal Reasoning,
Image Understanding and Verifcation projects.
The possible alternative is the Extended PERQ from Three Rivers running
SpiceLisp. Neither SpiceLisp nor the Extended PERQ are available at this
time. Stanford will consider the fate of the PERQs at CMU before any
decisions are made regarding them.
Since MacLisp is the ancestor of ZetaLisp, Symbolics Lisp Machines
suffice for the non-InterLisp computing. ZetaLisp, moreover, is one of
the family of Common Lisps (S-1 NIL, VAX NIL, and SpiceLisp being the
others). Thus if a larger, timeshared machine, which runs a Common Lisp,
becomes available for the needs of the non-InterLisp community, there
will be a small conversion job, if any.
The Dolphins are 1/3 to 1/10 the speed of a KL-10 at the moment,
depending on the task (with good improvements very likely in the near
furture), and have 23 bits of words addressing capability. The word size
is 16 bits, and so 4 words are required per non-cdr-coded cons cell,
meaning that there are 21 bits of cons cells available. However, there
are tag bits available as well, which is an advantage for Lisp.
Dolphins currently cost approximately $60K.
The Symbolics 3600 machines are expected to run 3 times faster than the
current Symbolics and LMI Lisp Machines, which means a speed of 50% to
100% of a KL-10. They have an address space of 24 bits, which, due to
cdr-coding yields 23.5 bits worth of cons cells. The 3600 machine comes
equipped with 256K 36-bit-plus-ECC words of main memory, Winchester disk
of at least 67 Mbytes, 1100x800 or so B&W screen, keyboard, mouse,
Ethernet II or Chaosnet, 68000 front end processor, 8-bit audio output,
and some RS-232 lines. They are expected to cost approximately $60K each.
Currently, discussions between the Common Lisp Group and the InterLisp
group at Xerox PARC are under way to see what grounds there are for a
closer association between these traditionally disparate dialects.
Moreover, Symbolics is actively considering an InterLisp implementation
on their 3600 machine. In the event this latter possibility is
realized, perhaps the choice of Symbolics 3600 machines would be wisest.
Consequently, we plan to defer the choice of specific machines until
April 1982, at which time these issues should become clear.
@newpage
@subsection(VLSI Research)
@paragraph(Current Facilities)
@paragraph(Requirements)
@paragraph(Alternatives)
@newpage
@subsection(Distributed Systems Research)
@paragraph(Current Facilities)
@paragraph(Requirements)
@paragraph(Alternatives)
@newpage
@section(Equipment Acquisition Plan)
@i[This section should include a system overview, a list of the proposed
equipment, along with
how it will meet above requirements, who will use what, when it will be
acquired, where we will put it, etc.]
@newpage
@section(Budget)
@subsection(Cost Estimates)
@i[Cost estimates for all items, as detailed as possible, should be
included.]
@subsection(Competition versus Selected Source)
For those items that have been specified generically, we will make a
detailed competitive evaluation of existing products and suppliers at
the planned time of purchase. Specific equipment will be ordered with
the concurrence of the DARPA program manager.
Equipment items that have been explicitly named in this proposal are
either uniquely suited to the research requirements, or are not
commercially available at this time, and hence are research products for
which there are no alternative sources.
@subsection(lease versus purchase)
@i[We need to say why we are purchasing this equipment rather than leasing
it.]
@section(Curriculum Vitae of Feigenbaum, McCarthy and Hennessey)
-------
∂18-Nov-81 2046 Gene Golub <CSD.GOLUB at SU-SCORE> Pratt
Date: 18 Nov 1981 2045-PST
From: Gene Golub <CSD.GOLUB at SU-SCORE>
Subject: Pratt
To: JMC at SU-AI
John! We'll meet at 2PM tomorrow to discuss the Vaughn Pratt case.
There will be a full professor meeting at 2:30. GENE
-------
∂18-Nov-81 2102 Gene Golub <CSD.GOLUB at SU-SCORE>
Date: 18 Nov 1981 2102-PST
From: Gene Golub <CSD.GOLUB at SU-SCORE>
To: JMC at SU-AI
In-Reply-To: Your message of 18-Nov-81 2052-PST
Thanks for the info. I'm agreeable about Carolyn serving on the committee.
Are YOU still interested in Ehud?
Glad to have you back! GENE
-------
Yes, I'm still interested in him.
∂19-Nov-81 0807 FFL
To: JMC, FFL
The meeting scheduled for 2:30 this afternoon will be preceded by another
at 2 p.m., per Prof. Golub.
∂19-Nov-81 0944 NAN pmessage
Elliot Blooms secretary called yesterday to tell you that Elliot has
called an SE 2 meeting on Tues. Nov. 24 at Tom Connolys house at 8 p.m.
Nan
∂19-Nov-81 0959 FFL
To: JMC, FFL
Dr. Gibbons office called to say there is a CIS Executive Committee meeting
on Wed., Nov. 25, 2 p.m., Conference Rm., ERL. John Osborne will talk about
patterns.Have put it on your calendar.
∂19-Nov-81 1002 FFL
To: JMC, FFL
When Dr. Gibbon's office called, I told her I thought you had a previous
engagement for that time. Checking your calendar, I see that you do chair
an oral at that hour. I have called Dr. Gibbon's office to tell them you
cannot be at the CIS meeting.
∂19-Nov-81 1143 Jrobinson at SRI-AI Tinlunch today in EK242.
Date: 19 Nov 1981 1141-PST
From: Jrobinson at SRI-AI
Subject: Tinlunch today in EK242.
To: tlgrp:
It's a little chilly outside.
Jane
-------
∂19-Nov-81 1304 FFL
To: JMC, FFL
E. Bloom's secretary called to say there is an SE2 meeting Tuesday, the 24th,
at Tom Connally's home. Is on your calendar.
∂19-Nov-81 1422 LGC Reading Kurt Konolige's Thesis
Before deciding whether to serve as a reader of Kurt's thesis, I'd like to
discuss with you who the other readers will be, what division (if any) of
responsibility among the different readers is contemplated, what differences
there are between the role of a non-faculty reader and that of a faculty
reader, and other relevant questions. Perhaps we could discuss this matter
sometime tomorrow afternoon. -- Lew
∂20-Nov-81 1145 Guy.Steele at CMU-10A Time/place of Common LISP Meeting
Date: 20 November 1981 1439-EST (Friday)
From: Guy.Steele at CMU-10A
To: rpg at SU-AI, rms at MIT-AI, rg at MIT-AI, dlw at MIT-AI, moon at MIT-AI,
hic at MIT-AI, alan at MIT-MC, gjc at MIT-MC, rlb at MIT-MC,
jonl at MIT-MC, jmc at SU-AI, vanmelle at PARC-MAXC
Subject: Time/place of Common LISP Meeting
CC: brooks at MIT-AI, masinter at PARC-MAXC, deutsch at PARC-MAXC,
Bill.Scherlis at CMU-10A, Scott.Fahlman at CMU-10A, griss at utah-20,
hedrick at rutgers, eak at s1-a
Message-Id: <20Nov81 143905 GS70@CMU-10A>
The Common LISP Meeting will be begin Monday, November 23, 1981,
at 9:30 AM EST at Symbolics, Inc., 257 Vassar Street, Cambridge, Mass.
This is behind the athletic fields at M.I.T., just off Massachusetts
Avenue. The meeting is tentatively scheduled to run from 9:30-5:30
on Monday, and 9:30-5:00 on Tuesday (with suitable breaks). See you
there!
--Guy
∂20-Nov-81 1200 Gene Golub <CSD.GOLUB at SU-SCORE> Lunch with C. Hurd
Date: 20 Nov 1981 1153-PST
From: Gene Golub <CSD.GOLUB at SU-SCORE>
Subject: Lunch with C. Hurd
To: tob at SU-AI, csd.buchanan at SU-SCORE, or.dantzig at SU-SCORE,
csd.feigenbaum at SU-SCORE, rwf at SU-AI, csd.golub at SU-SCORE,
csd.herriot at SU-SCORE, dek at SU-AI, jmc at SU-AI,
csl.crc.ejm at SU-SCORE, csd.ullman at SU-SCORE
cc: csd.irmgild at SU-SCORE
There will be a lunch on Wednesday, December 9 with C. Hurd at the Faculty
Club. Cuthbert is interested in helping the Department raise its endowment.
Please let me know as soon as possible whether you can come.
-------
∂20-Nov-81 1332 Irmgild Schack <CSD.IRMGILD at SU-SCORE> Faculty Lunch
Date: 20 Nov 1981 1330-PST
From: Irmgild Schack <CSD.IRMGILD at SU-SCORE>
Subject: Faculty Lunch
To: tob at SU-AI, baskett at PARC-MAXC, csd.buchanan at SU-SCORE,
or.dantzig at SU-SCORE, csd.feigenbaum at SU-SCORE, rwf at SU-AI,
csd.golub at SU-SCORE, csd.herriot at SU-SCORE, dek at SU-AI, zm at SU-AI,
jmc at SU-AI, csl.CRC.ejm at SU-SCORE, csd.wmiller at SU-SCORE,
csd.oliger at SU-SCORE, csd.irmgild at SU-SCORE, csd.ullman at SU-SCORE,
tw at SU-AI
By popular demand, we will have the lunch on Tuesday, November 24. It
will be held in MJH 252. I hope you can make it - unfortunately I will be
out of town. Vaughn Pratt will be in the chair.
GENE
-------
∂20-Nov-81 1354 Gene Golub <CSD.GOLUB at SU-SCORE> [Irmgild Schack <CSD.IRMGILD>: Faculty Lunch]
Date: 20 Nov 1981 1348-PST
From: Gene Golub <CSD.GOLUB at SU-SCORE>
Subject: [Irmgild Schack <CSD.IRMGILD>: Faculty Lunch]
To: CSD-Faculty: ;
Date: 20 Nov 1981 1330-PST
From: Irmgild Schack <CSD.IRMGILD>
Subject: Faculty Lunch
To: tob at SU-AI, baskett at PARC-MAXC, csd.buchanan, or.dantzig, csd.feigenbaum,
rwf at SU-AI, csd.golub, csd.herriot, dek at SU-AI, zm at SU-AI,
jmc at SU-AI, csl.CRC.ejm, csd.wmiller, csd.oliger, csd.irmgild,
csd.ullman, tw at SU-AI
By popular demand, we will have the lunch on Tuesday, November 24. It
will be held in MJH 252. I hope you can make it - unfortunately I will be
out of town. Vaughn Pratt will be in the chair.
GENE
-------
-------
∂20-Nov-81 1858 CLT SEMINAR IN LOGIC AND FOUNDATIONS OF MATHEMATICS
To: "@LOGIC.DIS[1,CLT]" at SU-AI
SPEAKER: Prof. Stan Wainer, University of Leeds
TITLE: "Recursion on continuous functionals"
TIME: Tues. November 24, 4:15-5:30
PLACE: Room 383-N, 3rd. floor Math. Bldg.
∂20-Nov-81 2056 JK ekl
To: JMC, JJW, YOM, JMM
The new ekl is up, old version can be found in okl.dmp[ekl,jk].
Should one desire old fashioned decision and re-writing procedures,
fasload fix.fas from [ekl,jk]. Documentation can be found in ekl.man[ekl,jk].
∂21-Nov-81 1259 JMC
To: "@SHACKL.LIS[F81,JMC]" at SU-AI
ONEILL[F81,JMC] here or ONEILL[SHA,JMC] at S1 reports a lunch with him.
∂23-Nov-81 0923 ullman@Diablo (SuNet) ARPA proposal
Date: 23 Nov 1981 09:22:40-PST
From: ullman at Diablo
To: equip
Subject: ARPA proposal
I met with Bob Kahn last Thursday. His reaction to our list
was basically favorable; in fact, he wants us to ADD 15 more
SUN terminals (some may go to the ISL group, however), and
5 more SUVAXes. We are thus free to put in a 3.3M proposal.
The only objection he had was to our listing half an S1.
He believes that the S1 will be a one (or two) of a kind item,
and not worth our trying to kick it into shape. He is also
nervious about funding half of something until we have the
details on where the other half comes from.
He suggests we convert the S1 money to LISP machines temporarily.
When the time comes, order something like a 2080, with the balance
paid for by whatever sources we could find, or advanced by the
university against a fourth year of arpa equipment funding.
∂23-Nov-81 1734 Woody Bledsoe <ATP.Bledsoe at UTEXAS-20> Re: Turing award for Boyer and Moore
Date: 23 Nov 1981 1929-CST
From: Woody Bledsoe <ATP.Bledsoe at UTEXAS-20>
Subject: Re: Turing award for Boyer and Moore
To: JMC at SU-AI
cc: ATP.Bledsoe at UTEXAS-20
In-Reply-To: Your message of 23-Nov-81 0313-CST
Yes, I agree, and am willing to help prepare the necessary material. I will
have to find out what the proceduce is for nominating someone.
However, since I was Boyer's Phd advisor, it might be counterproductive
to have me make the nomination. I would certainly support it.
Incidently, when we were trying to hire them here we obtained some excellent
letters from many leaders in CS. WB
-------
∂23-Nov-81 2023 CLT
ive gone to watch the opera on tv
∂24-Nov-81 1225 CSD.ULLMAN@SU-SCORE (SuNet) meeting
Mail-from: ARPANET host SU-SCORE rcvd at 24-Nov-81 1222-PST
Date: 24 Nov 1981 1220-PST
From: Jeffrey D. Ullman <CSD.ULLMAN at SU-SCORE at SUMEX-AIM>
Subject: meeting
To: equip at SU-SHASTA
AS we cannot meet this friday, I'd like to call a meeting of the
group for Monday Nov. 30 at 10AM, in B252.
The agenda will include a discussion of the response from Bob Kahn
regarding our proposal.
Please don't forget that I need drafts of your sections before thursday.
Kahn emphasized that he wants something on his desk by January 1.
-------
∂24-Nov-81 1248 CLT
I talked to Bonnell (12:45). He wants to look at the frigde again.
He can come late this afternoon (4:30ish) if you can be there.
You should call him. Otherwise he will come in the morning.
∂24-Nov-81 1536 Tom Wadlow <TAW at S1-A> Kantrowitz article
Date: 24 Nov 1981 1526-PST
From: Tom Wadlow <TAW at S1-A>
Subject: Kantrowitz article
To: jmc at SU-AI
Chris Ghinazzi (the secretary here at S-1) is attempting to locate the
magazine that the article ''The Ming Navy and the U.S. Space Program''
by Arthur Kantrowitz was in. I have a poor quality copy of a copy that
I got from you over the summer. The article claims to be in some
publication of the AIAA (Sept. 1981) but the LLNL library seems unable
to locate that publication. If you could shed some light on the origin
of your copy, poor Chris would be forever grateful. --Tom
I'm sorry, but my memory of where I got it must be playing tricks on me,
because I think I got it at S-1, possibly from Lowell or possibly even
from a bulletin board there. The copy I put on the bulletin board here
seems to have been taken down.
∂24-Nov-81 2123 Marvin Minsky <MINSKY at MIT-AI>
Date: 25 November 1981 00:17-EST
From: Marvin Minsky <MINSKY at MIT-AI>
To: jmc at SU-AI
cc: MINSKY at MIT-AI
I will speak to her wednesday morning and fix that. I really didn't
realize things had gone that way.
∂24-Nov-81 2145 Vaughan Pratt <CSD.PRATT at SU-SCORE> eiuolcs
Date: 24 Nov 1981 2143-PST
From: Vaughan Pratt <CSD.PRATT at SU-SCORE>
Subject: eiuolcs
To: jmc at SU-AI
cc: pratt at SU-SCORE, bothner at SU-SCORE
This is in response to your paper on large character sets.
I very much agree with most of the paper. The notion of a character apart from its representation either as an integer or as a bit-map is an important
abstraction for the computing milieu.
On the matter of "configurable keyboards" you might keep in mind that it is
more important that touch typists not be compelled to look at the keyboard
than that non-touch-typists not be compelled to look at the screen. For this
reason I would prefer that your "LCD keytops" be displayed on the screen,
in the manner of avionics "heads-up" displays, to be invoked and dismissed
under keyboard control.
It is interesting to note that Smalltalk users rarely touch the keyboard.
They accomplish most of their work using the mouse to point at menu items.
The three mouse buttons seem to provide about all the keyboard they need.
Obviously the keyboard is needed for text input, but if we spent all our
time at the keyboard inputting text we would all be most prolific authors.
The bulk of our time is spent in non-text-input mode.
I didn't agree with either of the reasons you gave for giving no control over
character size and font. Reason 1 (too hard) is inconsistent with
implementation item 5 (a character must be displayable and printable on a
variety of devices with different resolutions), which subsumes providing
for a variety of character sizes. Reason 2 (unusable) is unconvincing
given that (a) people vary character size all the time; (b) it isn't that
hard; (c) it can be made even easier by abstracting away from direct control
of size and providing for it via such abstractions as footnotes (which could
be in a smaller font), headings (larger font), remarks (smaller font), etc.
Here are some remarks that are neither agreeing nor disagreeing. All have
to do with the implementation section.
For item 1 I think we should move as soon as possible to a document editor like
Bravo, Bravo/X, or Etude. Such an editor should deal with characters as
abstractions distinct from the notion of the 8-bit byte. The question of how
characters are represented is a matter of protocol below the level of editor
design.
For item 2, I see a keyboard as a communication medium whose bit rate is a
hard-to-analyze function of the number of keys, the quality of the engineering
(n-key rollover etc.), the extent to which touch-typing skills can be put to
good use, etc. My personal preference is fewer rather than more function keys
since function keys take longer to locate, reach, and return from, than regular
keys. In that respect the Alto keyboard is ideal. My main objection to that
keyboard is that it is unique. I would prefer to see a standard layout emerge,
so that people can move from place to place, and from job to job, without
having to leave their typing skills behind. The Selectric and VT-100 layouts,
which are very similar, are both widely used, and are both widely available.
A fringe benefit of wide availability is economy: for example we can get a very
nice VT-100 layout for $100. Since such economy can be had without
compromising principles (well, my principles anyway, though after the many
discussions I have seen in the past 15 months on keyboards it is clear that
these are principles shared with much of the non-Sail community) it would seem
a waste of money to buy a $400 keyboard to acquire functionality that is both
of debatable merit and not in wide demand. You might feel that the lack of
demand is due to ignorance of the merits of what you are after, but you cannot
look to me for sympathy for that argument.
The rest of the paper I agreed with strongly, as I said, in particular items
1 through 3 of page 1, and items 4 through 8 of the implementation
considerations.
Vaughan
-------
∂25-Nov-81 0050 Howard D. Trachtman <HDT at MIT-AI>
Date: 25 November 1981 03:48-EST
From: Howard D. Trachtman <HDT at MIT-AI>
To: JMC at SU-AI
I am taking a couse taught by Prof. Fredkin here at MIT and your name
came up during the discussion. I was wondering if you could please tell
me a few things about him, as it would come in handy for the seminar.
Thanks a lot. (I am hdt@MIT-AI and will be logged in for a while
longer.)
I would really appreciate a confidential reply.
Your request seems bizarre.
∂25-Nov-81 1025 Woody Bledsoe <ATP.Bledsoe at UTEXAS-20> [CS.DALE: Turing Award]
Date: 25 Nov 1981 1220-CST
From: Woody Bledsoe <ATP.Bledsoe at UTEXAS-20>
Subject: [CS.DALE: Turing Award]
To: jmc at SU-AI
cc: ATP.Bledsoe at UTEXAS-20
John, as you can see we are getting the information. I will get back with
you right away. Woody
---------------
Date: 25 Nov 1981 0959-CST
From: Cs.DALE
Subject: Turing Award
To: atp.bledsoe
cc: CS.DALE
ACM is sending me information on the Turing Award. They have referred
me to Walter Carlson (IBM-San Jose) who, as I understand it, heads the
nominating committee this year. I'll find out more about the selection
process when he returns from vacation on December 1. Al
-------
-------
∂25-Nov-81 1411 JK disk allocation
I need more - I am currently over by 113 blocks.
∂26-Nov-81 0018 RWW thanksgiving
To: JMC, CLT
"party" starts at 12:00 dinner is served at 2:00 .....
Richard
Thanks. We'll be there.
∂26-Nov-81 0035 Robert Elton Maas <REM at MIT-MC>
Date: 26 November 1981 03:35-EST
From: Robert Elton Maas <REM at MIT-MC>
To: RWG at MIT-MC, RWW at MIT-MC, REM-DIARY-READERS at MIT-MC,
FFM at MIT-MC, GZ at MIT-MC, ELLEN at MIT-MC, GNU at MIT-AI
cc: MINSKY at MIT-AI, JMC at SU-AI
I think I've solved an outstanding problem in permutation groups,
how to represent the group so it's trivial to determine whether a
given permutation is in the group or not, and how to update the
representation when adding new elements to make a larger group.
The algorithm is simple enough that I wrote a set of LISP functions
for all the high-level parts of the algorithm rather than try to
describe it in English. All that's needed now to make it (I claim)
a working program is some routines for permutation multiplication
(which I wrote already but are offline) and some routines to do
circuit analysis (finding loops) in a graph and maintaining a
data stucture corresponding to that graph and what the graph means
in terms of the permutation group, plus of course debugging to locate
errors and to patch the boundary conditions such as the empty group
and its corresponding decision procedure.
Let's see, my major solved problems to date:
Differential algebra (1966),
Data compression (1978),
Go endgame analysis (1980-81),
Permutation groups decision procedure (1981).
∂28-Nov-81 1150 Purger exceeding your disk quota
You are exceeding your disk quota.
Files that occupy space beyond your quota are subject to purging.
If you don't delete some of your files, the purger will.
NOTE: Disk allocations and disk usage are now measured in disk blocks (a
block is currently one track); for details, READ PURGE/3P. If you
would like to buy a bigger allocation, contact Susan Hill (SH at SAIL).
The unit of allocation, one aliquot, is currently 50 disk blocks.
Your disk quota is: 1700 blocks
Your files occupy 3006 blocks
∂28-Nov-81 1418 RWW
To: REG, JMC
I would like to know where all the disk space is going. I have about the
same now as ever and when S1 left we had plenty. It seems to me that IN
SPITE OF ALLOCATIONS the historical balance is reasonable to be preserved.
The new encroachers should be purged first, until you can sell the
difference. In addition, if JMC and DEK do not get purged I don't expect
to be either!!!!!
Richard
∂30-Nov-81 0000 JMC*
kingsley davis lunch
∂30-Nov-81 1010 JJW Comprehensive Committee
You and I seem to be the MTC people on the Comprehensive Committee
this Winter. Terry Winograd is going to schedule a meting soon, and
has asked everyone to prepare some questions. Any ideas?
Joe
∂30-Nov-81 1135 JEF via Ethernet host 50#12 equip proposal
We ought to talk sometime about the matter of the S-1.
Also, the committee met Monday, and one of the things we discussed
was what to spend approx. $750K of fiscal '82 money on.
It was proposed that the purchase of the Foonly itself be
deferred to Oct., 1982, and that we begin by purchasing some disk
to be used with one of the SUVAXes.
I think you might not favor this approach, and we ought to take
the matter up again if you have strong objections.
P.S. How is your draft coming?
∂30-Nov-81 1400 Gene Golub <CSD.GOLUB at SU-SCORE> Lunch on Tuesday
Date: 30 Nov 1981 1357-PST
From: Gene Golub <CSD.GOLUB at SU-SCORE>
Subject: Lunch on Tuesday
To: CSD-Faculty: ;
cc: les at SU-AI
This Tuesday we will have lunch in MJH. Les Earnest will describe some
of the activities of IMOGEN and Bruce Buchanan that of TEKNOWLEDGE.
GENE
-------
∂30-Nov-81 1425 FFL Call from Sarah Lippincott, New Yorker magazine
To: JMC, FFL
Has made some revisions in article on basis of earlier conversation with
you. Would like to check them with you today or tomorrow morning.
If you can call within 45 minutes, please call her collect at 212 840 3800.
Later, please call her after 5 p.m. our time at 212 675 3036.
∂30-Nov-81 1433 Gene Golub <CSD.GOLUB at SU-SCORE> Monthly meeting
Date: 30 Nov 1981 1433-PST
From: Gene Golub <CSD.GOLUB at SU-SCORE>
Subject: Monthly meeting
To: tob at SU-AI, csd.buchanan at SU-SCORE, or.dantzig at SU-SCORE,
or.stein at SU-SCORE, csd.feigenbaum at SU-SCORE, rwf at SU-AI,
csd.golub at SU-SCORE, csd.herriot at SU-SCORE, dek at SU-AI, zm at SU-AI,
jmc at SU-AI, csl.crc.ejm at SU-SCORE, csd.ullman at SU-SCORE
The monthly meeting, as usual, will take place on Thursday, December 3 at
2:30 p.m. in the Conference Room next to my office. We have been authorized
to search for several appointments and I want to discuss the various
possibilities with you.
GENE
-------
∂30-Nov-81 1542 Nancy Dorio <CSD.DORIO at SU-SCORE> pmessage
Date: 30 Nov 1981 1541-PST
From: Nancy Dorio <CSD.DORIO at SU-SCORE>
Subject: pmessage
To: jmc at SU-AI
Stanford-Phone: (415) 497-2273
Steve Greenspan from the University of Buffalo called to find out if you
are interested in being a speaker for thier colloquium. The return phone
no. is 716 831-3074.
Nan
-------
∂30-Nov-81 1616 FFL
To: JMC, FFL
Steve Greenspan fof the graduate group in cognitive science at the U. of
Buffalo called to ask if you were able to accept their invitation to speak
to them in early March. He will call again tomorrow.
716 831 3074. 4226 Ridge Lea Road, U. of Buffalo, Buffalo 14226.
If Greenspan calls while I'm out, the answer is that I haven't time
to speak in Buffalo.
∂01-Dec-81 1104 SIS Colloquium Notice of December 7 - 11, 1981
To: "@COLLOQ.DIS[INF,CSD]" at SU-AI
Date Place Person
Day Event From
Time Title
-------------------------------------------------------------------------------
12/7/81 Math 380C Samuel Karlin
Monday Numerical Analysis Stanford University
4:15 p.m. Seminar ``Some Results On Multivariant Splines''.
12/8/81 MJ301 Bob Blum
Tuesday Medical Computing Stanford University
1:30 p.m. Journal Club ``Database Systems for Medical Literature
Search''.
12/8/81 MJ301 John Kunz
Tuesday Knowledge Stanford University
2:30 p.m. Representation ``Representation of Cause-Effect Relations in AI
Group Systems''.
12/8/81 Jordan 041 Steven Wolfram
Tuesday Computer Science Caltech
4:15 p.m. Colloquium ``SMP: A Symbolic Manipulation Program''.
12/9/81 MJ352 Harlyn Baker
Wednesday Robotics Semminar Stanford University
12:00p.m. ``Depth From Edge & Intensity Based Stereo''.
12/9/81 Terman 153 To be announced
Wednesday Computer Systems
4:15 p.m. Laboratory
Semminar
12/11/81 MJ301 Moshe Vardi
Friday Database Research
3:15 p.m. Seminar ``Global Decision Problems For Relational
Databases''
∂01-Dec-81 1131 FFL
To: JMC, FFL
Are you planning to attend the Math-SciCommittee lunch meeting at
Faculty Club tomorrow at 12:15?
∂01-Dec-81 1132 FFL
To: JMC, FFL
Please call Tom Connally at 7-1796.
∂01-Dec-81 1443 CLT
i won't be around for supper
∂01-Dec-81 1501 Gene Golub <CSD.GOLUB at SU-SCORE> Re: Selling Perseus computer time
Date: 1 Dec 1981 1459-PST
From: Gene Golub <CSD.GOLUB at SU-SCORE>
Subject: Re: Selling Perseus computer time
To: JMC at SU-AI, REG at SU-AI
cc: RWW at SU-AI
In-Reply-To: Your message of 1-Dec-81 1454-PST
WHAT IS PERSEUS?
-------
∂01-Dec-81 1607 Jeffrey D. Ullman <CSD.ULLMAN at SU-SCORE> Re: meeting
Date: 1 Dec 1981 1605-PST
From: Jeffrey D. Ullman <CSD.ULLMAN at SU-SCORE>
Subject: Re: meeting
To: JMC at SU-AI
In-Reply-To: Your message of 1-Dec-81 0316-PST
Bob kahn is happy with the file server idea.
He told me that we could figure on $750K for fiscal '82.
Thus, we have to pare down the list of things that we get
before Oct., 1982, which will be a subset of the 1.1M$ worth
of items we listed as ``first year.''
Ed, of course, pushed for LISP machines, and you may want to
go along with that, since they would presumably benefit
your community as well.
Other items are untouchable, e.g., the SUVAXes, and the development
money, or ar so small it doesn't matter.
Anyway, for my personal interest, I'd favor an environment where
storage was cheap to the one we have now, so I'm happy to reconvene
the committee.
-------
∂01-Dec-81 2130 William G. Dubuque <WGD at MIT-MC> permutation groups breakthrough?
Date: 2 December 1981 00:30-EST
From: William G. Dubuque <WGD at MIT-MC>
Sender: BIL at MIT-MC
Subject: permutation groups breakthrough?
To: REM at MIT-MC
cc: RWG at MIT-MC, RWW at MIT-MC, REM-DIARY-READERS at MIT-MC,
FFM at MIT-MC, GZ at MIT-MC, ELLEN at MIT-MC, GNU at MIT-AI,
MINSKY at MIT-AI, JMC at SU-AI
Althought I would love to see an improvement on Sims' permutation
group algorithms, I doubt that much improvement is possible. I suggest
that you examine his algorithms for producing bases and strong generating
sets for permutation groups before you continue with your research. These
algorithms have been known since at least the late sixties and are
currently the most effective tools available for computation in
permutation groups of large degree. Sims demonstrated the effectiveness of
these algorithms with remarkable computations in the Janko sporadic simple
groups and has even extended his techniques to allow the determination of
conjugacy classes (e.g. applied to the Suzuki simple group of degree 1782
and order 448,345,497,600) as well as many other useful properties and
constructions in permutation groups. For a detailed description of some
of these methods, see:
"On an Algorithm for Finding a Base and Strong Generating Set for a Group Given by
Generating Permutations"
Jeffrey S. Leon
Mathematics of Computation, v. 35, #151, July 1980
pp. 941-74
-Bil
∂01-Dec-81 2351 JPM Meeting tonight
Sorry I could not make it tonight (have a prior meeting I had to attend)
Any action items for consideration? I really am meaning to submit some
editorials to the Daily et al next quarter (will be written over the break).
Jim
∂02-Dec-81 0641 Marvin Minsky <MINSKY at MIT-ML>
Date: 2 December 1981 09:42-EST
From: Marvin Minsky <MINSKY at MIT-ML>
To: sacerdoti at SRI-KL, minsky at MIT-AI, gjs at MIT-AI, phw at MIT-AI,
rick at RAND-AI, jmc at SU-AI, tw at SU-AI, reddy at CMU-10B,
newell at CMU-10A, athompson at USC-ECL, bobrow at PARC-MAXC,
winograd at PARC-MAXC, dwaltz at BBND, webber at BBND,
woods at BBND, erman at USC-ISIF, balzer at USC-ISIF,
buchanan at SUMEX-AIM, engelmore at SUMEX-AIM,
feigenbaum at SUMEX-AIM, feldman at SUMEX-AIM, bledsoe at UTEXAS-20,
grosz at SRI-AI, nilsson at SRI-AI, walker at SRI-AI
Last Draft of letter to be sent on AAAI stationery. Comments?
Dear (Reagan, Stockman, etc. ?)
The U.S. now leads the world in Computers and it would be economically
disastrous to lose that leadership. That could happen soon -- unless
we reinforce our basic research in Computer Science. In this fast
moving field of computers, only 30 years old, we could fall behind in
less than 5 years!
THE PROBLEM:f Our basic research establishment in Computer Science is
burdened with old equipment, while our competitors in are targeting
this with top priorities. [1]. Our government support has declined,
and private industry cannot be expected to capitalize our basic
research laboratories -- →because the individual performer is not the
firm that reaps the benefits_.
Up to now, the U.S. has had the best-trained students in computers --
partly because our students could use better equipment than available
in other countries. But now that our schools cannot afford to buy
new, modern equipment, our scientists and students are either using
obsolete equipment -- or wasting irreplaceable time to build equipment
that could be bought.
I estimate this to be so serious as to slow our potential progress by
half! [2]. I would anticipate that a careful study of the problem
would conclude that, to modernize our research establishment would
need about 20M/year for state-of-the-art capital equipment and about
75M/yr for staff. [3].
DEFENSE: Our defense establishment critically needs research toward
better computer systems. And special problems of training volunteer
military recruits makes us especially dependent on the quality of our
research in automatic systems.
REQUIRED ACTION: In the past, ARPA and NSF managed very well to
maintain US research leadership in computers. But new budget
restrictions face us with a possible national emergency. To forestall
rapid decline, we must first (i) sustain our Basic Research budgets at
least at recent real dollar levels and then (ii) modernize the
equipment for Basic Research. [3].
That investment, of the order of 100M/yr, would be an appropriate use
of the proposed increase in defense spending, and might well be the
most cost-effective investment we could make in security. Within that
larger field of Computer Science, I believe the sub-area that will
make the most crucial differences, is the field of Artificial
Intelligence. [4].
If you wish more information or have further questions, the resources
of the AAAI are at your service.
Marvin Minsky,
President, AAAI
!
NOTES
[1]. Japan's MITI has announced a new national-industrial campaign to
seize world leadership in computers. Their project, called the "Fifth
Generation Computer Project", is ambitious and realistic. Unless we
meet this challenge the U.S. could soon lose its leadership both in
Hardware and in Software. It is rumored that the Fifth Generation
project includes plans to supply its workers with modern equipment.
[2]. Progress in computer research depends critically on access to
modern equipment. →It is extremely difficult to advance modern
software concepts without access to modern computers_. Because of the
scarcity of good equipment, researchers are already flowing from
University to Industry: this yields some short-term benefits, of
course, but is rapidly depleting our Basic Research community.
[3]. COSTS: In lieu of a serious study, here are some rough estimates.
The basic research community in Computer Science numbers perhaps 250
senior investigators. An effective research facility needs about
$200K of modern equipment for each principal, staff and students. To
be very up-to-date, research equipment must be considered to have a
3-year lifetime:
Equipment = 1/3 x 200 x 250K = 17M/yr.
A typical group -- senior investigator, two programmer technicians,
and several students -- costs around 150K/yr, which must be doubled
for benefits and overhead.
Personnel = 2 x 150 x 250 = 75M/yr.
[4]. ARTIFICIAL INTELLIGENCE is the specialized area, within Computer
Science, that tries to make more intelligent computer programs.
Research in this field has been especially significant in the growth
of our present knowledge and skill in Robotics, Advanced Software,
Pattern Recognition, Office Automation, and "Expert" computer systems.
∂02-Dec-81 1214 FFL
To: JMC, FFL
Please call Diane (412 578 2565) at CMU. She has some questions about
your visit recently.
∂02-Dec-81 1532 jjf@Shasta (SuNet) Lisp Proceedings
Date: 2 Dec 1981 15:25:05-PST
From: jjf at Shasta
To: jmc@sail
Subject: Lisp Proceedings
I returned your 1980 Lisp Conference Proceedings to the shelf. Thanks.
--Jeff Finger--
∂02-Dec-81 1535 FFL
To: JMC, FFL
Please call Sol Feferman at 7-2439.
∂02-Dec-81 1649 Robert Schreiber <CSD.SCHREIBER at SU-SCORE> Dinner with speaker
Date: 2 Dec 1981 1649-PST
From: Robert Schreiber <CSD.SCHREIBER at SU-SCORE>
Subject: Dinner with speaker
To: CSD-Faculty: ;
I'm taking Steven Wolfram, next week's colloquium speaker, to
dinner after his talk. All are invited, but this isn't on the department.
Any restaurant suggestions?
Rob
-------
∂02-Dec-81 1800 JMC*
Call Bob Shannon.
∂03-Dec-81 0300 LGC Disk Allocation
My disk quota of 100 blocks is now too small, even though I have culled out
most of the small files, thereby raising my overall usage factor to 77% ,
which is higher than that of most of the others in the formal reasoning group.
My files currently take up 187 blocks, and as the advice taker work proceeds,
they could easily go to 250 in the forseeable future. Will you approve the
purchase of 150 more disk blocks for me at this time? The purger has begun
to make threatening noises. -- Lew
250 is ok. Tell Lynn Gotelli.
∂03-Dec-81 1008 pratt@Shasta (SuNet) benchmarks
Date: 3 Dec 1981 10:08:32-PST
From: pratt at Shasta
To: @, CSVAX.pattrsn@UCB-C70, csvax:Sumex-Aim, pratt@Shasta
Subject: benchmarks
Cc: jmc@sail
From CSVAX.pattrsn@UCB-C70 Wed Dec 2 18:55:36 1981
Mail-from: ARPANET host site 1200200116 rcvd at 2-Dec-81 1853-PST
Date: 2 Dec 1981 18:40:18-PST
From: CSVAX.pattrsn at Berkeley
To: pratt@Shasta, @, CSVAX.Sumex-Aim@Berkeley
Subject: benchmarks
If its not too much trouble, I would appreciate getting copies in the
mail. (I can't FTP). I have been doing work similar to the MIPS project
and I need all the information I can to convince people of the merits
of unusual machines.
I think, by the way, that the SUN workstation is a steal at $6800. Has
anyone figured out what to do about maintenance?
Copies of what? Timings, code, ...?
Yes, $6800 is a steal, but that's because that's Stanford's special price
break from CadLinc (which I think is actually a bit high).
Maintenance: Depends on the level at which the problem is observed.
Net level: Nonfunctioning nets are isolated from other nets and checked
for leakage, traffic jams, malfunctioning transceivers, etc. Each net is kept
small (10-20 nodes), so this task need not be the impossibility it could be
for a 200-node net. (Imagine trying to determine which transceiver was
shorting out the net in a 200-node network.)
Node level: The faulty board or other component is easily identified, due
to the relatively few components in each node and the coincidence of
functional and physical boundaries between boards.
Board level: Faulty boards go into the bad board queue, from which
they are retrieved eventually by a technician who repairs them and puts them
into the spares supply. The present goal is to have 10% spares on hand, a
number that may vary as we pick up experience on MTBF and MTTR. There are
presently 4 board types in use, expanding soon to 6-8, most having around
80 chips. We are working on automating board diagnosis as far as possible.
∂03-Dec-81 1011 Gene Golub <CSD.GOLUB at SU-SCORE> ASSISTANTSHIP RENEWAL
Mail-from: ARPANET site SU-SCORE rcvd at 3-Dec-81 1007-PST
Date: 3 Dec 1981 1006-PST
From: Gene Golub <CSD.GOLUB at SU-SCORE>
Subject: ASSISTANTSHIP RENEWAL
To: FACULTY at SU-SCORE
cc: CSD.MWALKER at SU-SCORE, CSD.HILL at SU-SCORE
IF YOU ARE PLANNING TO CHANGE THE STATUS OF ANY OF YOUR RA'S FOR
THE WINTER QUARTER, PLEASE INFORM SUSAN HILL AS SOON AS POSSIBLE.
THANKS. GENE
-------
∂03-Dec-81 1046 WOL Zohar is coming
John,
Zohar called this morning and said that he finally will be coming next week
Pierre
∂04-Dec-81 1449 Jrobinson at SRI-AI TINLUNCH DEC. 10
Date: 4 Dec 1981 1447-PST
From: Jrobinson at SRI-AI
Subject: TINLUNCH DEC. 10
To: tlgrp:
The next TINLUNCH (Dec. 10) will be preceded by a talk by the authors
at 11 o'clock in EK242. Stanley Peters (UTexas, currently at the
Center for Advanced Reseach in Behavioral Sciences) and Bob Ritchie
(UWashington, currently at XEROX) will talk about their present work
on formal grammars, parsing, and complexity. A paper outlining the
work will be available by Tuesday December 8 for distribution.
Jane
-------
∂04-Dec-81 1656 CG
is the mtc qual in the morning or the afternoon of the 14th?
The custom is to arrive at 9am and schedule the quals for that day.
It takes about an hour and a half per student. How long it takes
depends on the ratio of examinees to examiners.
∂06-Dec-81 0033 pratt@Shasta (SuNet) eyboards
Date: 6 Dec 1981 00:31:03-PST
From: pratt at Shasta
To: AVB@SU-AI, DEK@SU-AI, REG@SU-AI, VRP@SU-AI
Subject: eyboards
Cc: SUN@SU-AI, jmc@sail
From DEK@SU-AI Sat Dec 5 20:05:09 1981
Mail-from: ARPANET host SU-SCORE rcvd at 5-Dec-81 2004-PST
Mail-from: ARPANET site SU-AI rcvd at 5-Dec-81 2001-PST
Date: 05 Dec 1981 1951-PST
From: Don Knuth <DEK at SU-AI>
Subject: eyboards
To: VRP at SU-AI, AVB at SU-AI, REG at SU-AI
CC: SUN at SU-AI
On behalf of the TEX project, I would like to make a strong plea for
SAIL-like keyboards (control meta like) on the SUN stations.
The present plan for the default keyboard is to use the VT-100 layout, a
layout that is very close to Selectric layout, and is in wide use, and
therefore likely to be easily learned by new staff, and easily unlearned when
that staff moves on to new jobs or has to use a real Selectric typewriter.
We do want the standard VT-100 keyboard to be up-down encoded, meaning that it
will transmit every key transition rather than every key stroke. Thus you will
be able to play it like a keyboard and so get any effect you want in software,
including being able to make any key act as a control (meta, top, etc) key.
We also want to see additional keys placed about where the Sail/MIT keyboard
control and meta keys are, which in normal use would probably be used as
control-type keys. This should minimize the need to have separate exact
duplicates of Sail or MIT keyboards.
The keyboard we are presently considering for this purpose, by Keytronics,
costs $100. While the importance of the human interface would warrant
spending more on the keyboard, an additional $200 for an exact replica of the
MIT keyboard seems not to buy $200 worth of functionality.
The MIT alphaglyph keytops (my term for alphabetic keytops with an additional
special symbol) may be already available for this keyboard. However if
the scheme at the very end of this message is used then they aren't
necessary, or even terribly useful.
All this is however not to say that we expect Suns to have only these
keyboards. The Sun is a pretty modular gadget all round, and this extends to
its display and input devices. We don't expect the SAIL or MIT keyboards to
be in universal demand other than by Sail users or Symbolics Lisp machine
users, and accordingly have not designed the system to be irreversibly bound to
either one. On the other hand there will be people who want to use one or the
other, and the Sun hardware makes this not only possible but easy.
We have already tried a SAIL-like keyboard on the Sun, and Tovar
and Bill Nowicki have rigged it to work exactly like a SAIL keyboard when
using Nowicki's multiwindow telnet, so you can log into Sail and pretend you
are on a high resolution Datadisc. The Symbolics people demand that
MIT-layout "Space Cadet" keyboards be usable on Suns so that Suns can be
terminals to the Symbolics Lisp Machines. Alto fans may want Alto
keyboards. Some people may even want a mouse, a bit-pad, a joystick, a light
pen, a chord set, but no keyboard at all. You could go so far as to turn a
Sun into an organ emulator using as input just an organ console with manuals,
clavier, and stops, and using the display for the music (you hit a special
stop to turn the page). Such schemes can all be done on the Sun, and quite
easily provided the input devices all stick to EIA RS-232 standards, which
most do.
I hesitate to predict whether all this heterogeneity will eventually drive
everybody nuts or keep everybody content. However we have made sure that the
hardware does not prevent heterogeneity.
But I've said
that before; right now I want to mention that I have done a lot of thinking
about internal character codes during the last couple weeks, and I am
convinced that the best way for us to go is to adopt MIT's character set.
... TEX has adopted
MIT's code as its standard internal code. Whether or not the SUN terminal
adopts it, TEX will use it; but there will be a lot less confusion if
SUN software goes with it too.
Boy, this is a tough one. Let me deal with the easy points first, then get
to the hard ones.
First off, we should probably agree on definitions. My interpretation of
"the MIT code" is as follows: the ASCII assignments to the (decimal) codes
8..10, 13, and 32..126, and the MIT assignments to the codes 0..6, 11,
14..26, and 28..31. There remain questions as to whether you intend 7 to be
bell or pi, 12 to be form feed or plus-or-minus, 27 to be escape or lozenge,
and 127 to be (another) escape or integral. These questions remain because
the concept of "the MIT code" is not well-defined at MIT! Different software
makes different assumptions about the proper interpretation of the MIT code.
However to within those questions I am sure that a substantial body of
software will emerge on the Sun that supports just that standard. (It may
also turn out that the present Sail assignment will find some and maybe a lot
of support, depending on who implements what. After all there is a lot of
extant Sail software.) Certainly no one is proposing either to prevent the
use of control codes in text or to prevent the use of the MIT assignment of
characters to control codes. Thus TEX will be able to proceed with its
business when it migrates to the Sun.
Furthermore, SCORE should use the MIT
scheme if it ever goes for more than 96 characters.
In a sense your wish is already granted, in that files edited with Emacs
can contain any of the 128 character codes. All can be input to Emacs. How
they appear on the terminal depends on whether they are passed to the terminal
directly or expanded to their ↑@,↑A,↑B,... form (an Emacs option) and if the
former on how the terminal interprets those codes. How they appear on the
Dover depends mainly on who produces the press file. Other devices depend
more on the device than on Score itself.
Now let me get to more problematical aspects of the character code question.
Any convention adopted should be propagated to places that text may be
propagated to, or the text would not be readable. This I am sure motivated
you to have Score adopt the standard. Presumably if you made more use of the
various other computers around Stanford you would want to see it propagate to
them as well.
But now why stop at the boundaries of Stanford? Any site that might receive
text from Stanford should also have the MIT character set, and moreover
should treat the control codes as characters of that set rather than displaying
them as ↑A etc.
Unless you can get sufficiently broad agreement from other sites to adopt the
MIT assignment, or even just the acceptance of control codes as legal
characters, you will need to have text sent from Stanford to some sites be
translated to a 96-character ASCII encoding of the Stanford text, or sent as
binary data. This starts to sound like a pain.
An alternative would be for Stanford to use only the 96 ASCII characters in
text files. But then what about all the other characters you would like?
This is where ANSI comes in. There is an ANSI standard that says how to deal
with font changes, as well as a number of display control functions. If you
were to use this already existing ANSI standard, you would have a far better
chance of eventually finding that everyone else can read your text files, since
outsiders will look more kindly on being asked to adhere to an ANSI standard
than to toe the relatively parochial Stanford/MIT line.
An additional advantage is that the ANSI standard does not cater for a small
jump from 96 to 128 characters, but rather for an indefinite number of
characters, by offering a font-switching convention. It is unfortunate that
the standard does not specify what each font contains (at least I believe
there is no such standard), but I think you are on safer grounds supplying
that detail than filling in the ASCII control codes, a practice that has
little support outside parts of MIT and Stanford. Moreover you will be
sending files that are readable by everyone in the sense that they can at
least see what's in the file to within choice of font. MIT codes sent to some
sites (e.g. SAIL, whose editor discards some control characters such as ESC)
may not be visible at all. Another advantage is that the ANSI standard goes
beyond text altogether and also specifies some graphics primitives. Thus you
may include figures in papers you send out on the Arpanet!
There is some support within Stanford for using the ANSI standard. It is
available on the Sun terminal now; thus if I received a file from Norway using
the ANSI standard I could view it immediately, figures and all, provided we
agreed on the choice of fonts. (At present we have assigned a number of Xerox
fonts, a bit parochial I'm afraid, though less so than the MIT codes since
more sites use Alto fonts than MIT or Sail characters.)
Interest in the ANSI standard is understandably absent from the Sail
community. Hence it may be more trouble than it is worth to try to change
that community.
So as you see, there is no simple answer to the question of whether Sun
should arrange its software to standardize on the MIT codes. Maybe the best
that can be said is that Sun software can at least SUPPORT those codes, but
that it would be a mistake to support it to the exclusion of all other
standards, especially if that entailed locking out a much more universal
standard such as ANSI's.
Going a step beyond the ANSI standard, John McCarthy has been advocating a
character registry that would recognize a set of characters and describe
them. New characters may be added as appropriate. I like this proposal very
much.
John's scheme also calls for the assignment to each character of a distinct
(nonnegative?) integer. Such an assignment would be valuable for what one
might call "dumb channels," which are communication channels that do not know
how to translate the sender's representation of an abstraction such as a
character to the receiver's. My (perhaps overly ambitious) goal is the
abolition of channels that are dumb about characters. (I'd like to go further
and have channels be smart about abstractions like integers, rationals, even
representable reals, functions, and processes.) While such a goal remains
unmet it would be handy to have a universal character code. There is no
reason for this code not to coincide with ASCII on its 96 characters. It
could even coincide with the MIT control code assignments without causing too
much upset, I would think.
This brings us back to keyboards, with a vengeance. How on earth do you type
an arbitrary character? John's proposal also deals with this problem, but I
feel his suggestions here are the least satisfactory part of the proposal.
All of them involve looking at the keyboard. My main argument here is that it
is more important that a touch typist not be compelled to look at the keyboard
than that a non-touch-typist not be compelled to look at the screen. Hence my
proposal is to be able to summon up any keyboard on the screen at the touch of
a function key set aside for that purpose followed by an identifying key or
keys. This would happen apparently instantaneously, and the keyboard would
normally disappear again, just as instantaneously, as soon as a character
from it had been typed.
This idea generalizes to a keyboard treatment of menus. Current thinking is
that these are controlled by a mouse, but I believe it is important to have
the option to control them with a keyboard as well as a mouse. Some
applications involve a nontrivial amount of keyboard use, as when entering
text. In such applications frequent moves between mouse and keyboard can
really slow down an efficient user.
Vaughan
∂07-Dec-81 0421 JPM Meeting on Friday
Sorry I missed it. That meeting at noon did not break up until 1:30 -
some joker brought up some last minute stuff that had to be discussed
right away (end of quarter and all that). Any action items discussed?
More importantly, any actions decided upon?
Jim
The Stanford SE2 chapter, by inclination of its members rather than
principle, has tended to undertake projects of national interest,
such as our conferences on waste disposal and what to do in case of
an oil cut-off. Our probable current project is a California energy
recovery plan" based on the idea that the next governor, whether
Democrat or Republican, is likely to be better on energy than Brown.
A preliminary version of the ideas, not looked at by anyone else yet,
is in CERP[F81,JMC]. Why don't you look at it, and then we can
discuss it. It looks like Miro Todorovich, the national chairman,
will be able to find some money to pay someone, probabaly Alan
Pasternak (former State Energy Commissioner) to work out details.
∂07-Dec-81 0818 CL.MOORE at UTEXAS-20 more on brown thumb
Date: Monday, 7 December 1981 10:13-CST
From: CL.MOORE at UTEXAS-20
To: John McCarthy <JMC at SU-AI>
Cc: cl.boyer at UTEXAS-20, CLT at SU-AI
Subject: more on brown thumb
Instances of the associativity of AP are what we use when
someone asks us "what's the simplest theorem your system
can't prove." It just doesn't do well on instantiated
versions of beautiful theorems. Of course, had you first
had the system prove the associativity of AP in its general
form (as a REWRITE lemma using PROVE.LEMMA) all instances of
it would have been proved immediately.
The fact that the theorem-prover produced an invalid subgoal
is not unusual nor cause for worry. The system sometimes
generalizes the formula to be proved. This sometimes
produces a goal that is not a theorem. Assuming the
theorem-prover is sound this just means the system will
fail to find a proof.
Generalizations can happen in several ways. The two most
common are by the elimination of some hypothesis (in our
"cross-fertilization" equality substitution heuristic -
Chapter XI of our book A Computational Logic) and by the
replacement of a nonvariable term by a new variable (in our
"generalization" heuristic -- Chapter XII).
In your example the theorem-prover generalized after the
first induction. This produced formula 1.1, which is still
a theorem and which the theorem-prover tried to prove by a
second induction. The induction step of that second
induction led to the non-theorem *1.1.3. Here are the last
few steps in the production of *1.1.3. Bracketed comments
explain where we lost the validity of the formula.
(IMPLIES (AND (EQUAL (AP X (AP W (CONS Z X)))
(AP Y (CONS Z X)))
(EQUAL (AP X (CONS D (AP W X)))
(AP Y X)))
(EQUAL (AP X (CONS D (AP W (CONS Z X))))
(AP Y (CONS Z X)))).
[This formula, being the inductions step for a valid
formula, is valid. However, note that if the first
hypothesis is true the second is false and vice versa. In
addition, if the first hypothesis is true, the conclusion is
false. In the next step we substitute, into the conclusion,
the left-hand side of the first hyp for the right hand side.
This makes the conclusion false. If we did nothing more the
resulting formula would still be valid because it has
contradictory hypotheses. But we throw away the first hyp
-- on the grounds that it has been "used".]
We use the first equality hypothesis by substituting
(AP X (AP W (CONS Z X))) for (AP Y (CONS Z X)) and throwing away the
equality. We would thus like to prove:
(IMPLIES (EQUAL (AP X (CONS D (AP W X)))
(AP Y X))
(EQUAL (AP X (CONS D (AP W (CONS Z X))))
(AP X (AP W (CONS Z X))))).
[The above formula is invalid. Of course, the theorem-prover doesn't
know that and so it proceeds to generalize it still further before
going into induction on it.]
We will try to prove the above formula by generalizing it, replacing
(AP W (CONS Z X)) by A. We restrict the new variable by recalling the type
restriction lemma noted when AP was introduced. This produces:
(IMPLIES (AND (LISTP A)
(EQUAL (AP X (CONS D (AP W X)))
(AP Y X)))
(EQUAL (AP X (CONS D A)) (AP X A))).
Give the above formula the name *1.1.3.
Bob and J
∂07-Dec-81 1012 Robert Schreiber <CSD.SCHREIBER at SU-SCORE> Visit by S. Wolfram
Date: 7 Dec 1981 1008-PST
From: Robert Schreiber <CSD.SCHREIBER at SU-SCORE>
Subject: Visit by S. Wolfram
To: CSD-Faculty: ;
Tomorrows colloquium speaker, Steven Wolfram, is a Caltech physicist
who has done some extraordinary work in symbolic computation,
for which he got a MacArthur Foundation award.
He is arriving tomorrow at 11:00. Aside from lunch and the talk,
I don't know what to do with him. Please volunteer, therefore,
to talk with him for 1/2 to 1 hour. RSVP to me and
I'll try to knock together a schedule.
Rob
-------
∂07-Dec-81 1021 FFL
To: JMC, FFL
Angela Chiarappa of Italian TV called, referred, she says, by Mark VanDalin of
Harvard. Does a TV series on AI in Italy. Wants to interview you. Would need
about an hour of your time on December 23. Please call her before Wednesday at
212 975 0200.
et msg
∂07-Dec-81 1041 Gene Golub <CSD.GOLUB at SU-SCORE> Lunch on Tuesday
Date: 7 Dec 1981 1040-PST
From: Gene Golub <CSD.GOLUB at SU-SCORE>
Subject: Lunch on Tuesday
To: faculty at SU-SCORE
The lunch this week will take place in MJH 262. Aside from the colloquium
speaker, Jeff Ullman will descibe the ARPA proposal on equipment.
GENE
-------
∂07-Dec-81 1939 Dave Waltz <DWaltz at BBN-TENEXD> AAAI-82 CONFERENCE PROGRAM COMMITTEE
Date: 7 Dec 1981 2228-EST
Sender: DWALTZ at BBND
Subject: AAAI-82 CONFERENCE PROGRAM COMMITTEE
From: Dave Waltz <DWaltz at BBN-TENEXD>
To: MINSKY at MIT-AI, NILSSON at SRI-AI, BLEDSOE at UTEXAS-20, BOBROW at PARC-MAXC, GROSZ at SRI-AI, FELDMAN at SUMEX-AIM, MCCARTHY at SU-AI, SIMON at CMU-10A, GJS at MIT-AI, WEBBER, WOODS, BALZER at USC-ISIF
Cc: DWaltz
Message-ID: <[BBND] 7-Dec-81 22:28:30.DWALTZ>
THE TIME HAS COME TO SETTLE ON THE COMPOSITION OF THE PROGRAM
COMMITTEE FOR THE UPCOMING AAAI CONFERENCE. MY PLAN IS TO ALLOW
A FEW DAYS FOR EACH OF YOU TO SUGGEST NAMES OF THOSE PERSONS YOU
WOULD LIKE TO SEE ON THE PROGRAM COMMITTEE, THEN TO COLLECT THE
NAMES INTO GROUPS BY AREA OF EXPERTISE, AND FINALLY TO DISTRIBUTE
THE LIST TO YOU, WITH A REQUEST FOR APPROVAL OR VETO OF
INDIVIDUALS, OR FOR ADDITIONAL NAMES IF NOT ENOUGH PEOPLE HAVE
BEEN NOMINATED TO COVER AN AREA. I AM A LITTLE UNEASY ABOUT THE
PROCEDURE, BECAUSE I WOULD LIKE TO AVOID BATTLES BETWEEN
FACTIONS, BUT AT THE SAME TIME I WOULD LIKE TO DRAW ON YOUR
COLLECTIVE WISDOM. ONE ANSWER MIGHT BE TO RESPOND TO ME
PERSONALLY ABOUT ANY DOUBTS YOU HAVE, AND TO TRUST ME TO BE
DISCRETE -- ON THE OTHER HAND, I MAY BE CONCERNED UNNECESSARILY.
ANY SUGGESTIONS ON THE PROCEDURE WOULD ALSO BE WELCOME.
THE MEETING OF THE PROGRAM COMMITTEE IS TENTATIVELY SET FOR THE
WEEKEND OF APRIL 24-5, AT A CONVENIENT LOCATION IN THE BOSTON
AREA. PLEASE TAKE THE FOLLOWING FACTORS INTO ACCOUNT IN MAKING
NOMINATIONS: 1) WE DO NOT HAVE MUCH OF A BUDGET, SO PEOPLE WHO
CAN DRIVE TO BOSTON OR WHO CAN MANAGE TO GET THEIR WAY PAID WILL
BE PREFERRED; 2) YOU ARE WELCOME TO NOMINATE YOURSELF; 3) PLEASE
CHECK WITH OTHERS YOU NOMINATE TO VERIFY THAT THEY COULD AND
WOULD SERVE IF ELECTED; 4) WE NEED TO HAVE AT LEAST TWO PEOPLE
FOR EACH OF THE TOPICS LISTED ON THE CONFERENCE ANNOUNCEMENT
(THOUGH ONE PERSON MIGHT REVIEW PAPERS IN MORE THAN ONE AREA).
I PLAN TO STAY VERY CLOSE TO THE HARD DEADLINE OF APRIL 15 FOR
ACCEPTING PAPERS. PAPERS WILL BE DISTRIBUTED TO COMMITTEE
MEMBERS BEFORE THE COMMITTEE MEETING, IN THE HOPE OF AVOIDING
SOME OF THE PROBLEMS ENCOUNTERED LAST TIME.
PLEASE GET MESSAGES TO ME BY THURSDAY, DEC. 10TH.
THANKS,
DAVE
∂07-Dec-81 1946 YM MTC Quals
To: JMC
CC: YM
Hello,
Is it already known who will be in the committee of the MTC Quals?
When will the schedule be known?
Thanks, Yoni Malachi.
Chris Goad has agreed to make a schedule.
∂07-Dec-81 2247 TOB
To: "@PARTY.DIS[1,TOB]" at SU-AI
I invite you to a dinner party to celebrate my appointment as adjunct
professor. I would be very happy if you can join us. Please bring
your spouse or special friend. The party will
take place from 6pm, Sunday Dec 20. Look for a later announcement
about where the party will be. Remember it is for dinner.
∂07-Dec-81 2256 TOB
To: "@PARTY.DIS[1,TOB]" at SU-AI
I invite you to a dinner party to celebrate my appointment as adjunct
professor. I would be very happy if you can join us. Please bring
your spouse or special friend. The party will
take place from 6pm, Sunday Dec 20. Look for a later announcement
about where the party will be. Remember it is for dinner.
RSVP with small integer (0,1,2,..) indicating how many people
by December 15 for Ione's planning.
∂07-Dec-81 2258 TOB lisp machines
ACRONYM is being adopted by one company for an
ARPA project and they plan to use it for other
projects. It is becoming important to have
quite a bit of access to LISP machines for LISP
machine LISP, not INTERLISP. We have strong
interest in what machines the department buys.
This is in addition to the VAXwhich is still important
for Image Understanding Testbed and for automation.
∂08-Dec-81 0904 FFL
To: JMC, JMM, FFL
TV Network called to say that for some reason Hewlett-Packard stuents in
Colorado did not get a copy of the midterm. She is sending them there today.
Wanted you to know so that you would not grade them down. Of course, they
will be quite late in finishing the work.
∂08-Dec-81 1006 CG
Schedule for MTC qual
Students:
>Yellin, Frank - PhD Student [FNY: MTC] 50
>Malachi, Yonatan - Phd student [YM: MTC] 100 0
>Malik, Jitendra - PhD Student [JMM: ULLMAN] 50
>Weening, Joe - PhD student [JJW: FR] 100
Examiners : McCarthy, Manna, Pratt, Talcott, Goad
Possible partition of examining committee (according to
the principle of distribution of interest areas)
(1){McCarthy, Pratt}
(2){Manna, Talcott, Goad}
Schedule :
Committee (1): Jitendra Malik at 9:30, Joe Weening at 11:00
Committee (2): Frank Yellin at 9:30, Yoni Malachi at 11:00
Another partition (concentration of interest areas)
(1) {Manna, Pratt}
(2) {McCarthy, Talcott, Goad}
Schedule same as above.
∂08-Dec-81 1131 CLT
feferman is having a party on Thursday Dec 17
∂08-Dec-81 1139 FFL mail jmc,ffl
To: JMC, FFL
Have had inquiries about the hour of the Seminars, CS320 and CS370. Have
they been set yet?
I know nothing about cs320 and cs370.
∂08-Dec-81 1533 Jrobinson at SRI-AI Talk and Tinlunch on Thursday
Date: 8 Dec 1981 1533-PST
From: Jrobinson at SRI-AI
Subject: Talk and Tinlunch on Thursday
To: tlgrp:
Stanley Peters and Robert Ritchie will present a talk on
Phrase-linking Grammars at 11 am Thursday, EK242 preceding the
Tinlunch meeting (same place) at noon. Copies of their draft of a
paper on the subject will be available tomorrow (Barbara's cabinet).
For those who will not have time to read the draft before Thursday,
here are the Key words and phrases: context-free grammars,
phrase-structure grammars, natural language syntax, natural language
semantics, finite state tree automata.
Excerpts and paraphrases: They formally define "linked trees" and
indicate how the employment of linked trees as structural descriptions
can shed light on a theory of grammar. These devices "seem to permit
the description of natural language syntax by means of
phrase-structure rules, without forcing one to distort and fragment
natural generalizations about the languages in order to fit the
restricted form of phrase structure rules. We believe that
transformations can be eliminated in favor of more restricted devices,
including links." Linked trees may also be useful for describing the
syntax of programming languages; e.g., they might be used to reference
variables declared earlier in the program. Phrase-linking grammars,
as syntactic theories of natural languages, allow efficient parsing
and assign syntactic structures which admit of semantic interpretation
in a natural and general way.
-------
∂08-Dec-81 1549 CG mtc qual
I've confirmed that the four students in the list I sent you earlier
are in fact the only people taking the qual. So, if you like the
schedule in the earlier note, I can send it out.
OK, ship it.
∂08-Dec-81 1650 JDH chess problem
I've been working pretty hard on it and I still have the feeling that I am
close to getting a program which can beat the database. I think I have all
the necessary ideas in the program but I am still debugging. (This refers
mostly to chess oversights not to logical errors which are easier to catch.)
The major problem is that it is very difficult to define when white is 'in control'.
I look first of all at the king positions and then at ways black can force white
away (or escape himself) so that white loses control. I also consider ways
white can regain 'control' in a few moves. It appears that if I fix the two
routines (i.e. routines that decide whether white can regain control or black can
force white to lose it), I should be successful.
The program now gets black close to the edge with reasonable efficiencey
but doesn't quite work in the intermediate regions. I have started to examine
exception positions where my program makes evaluations which don't correlate
well with the database.
Anyway, the point of this whole message is that I am not done yet but I
am abou to leave for the holidays. I meant to talk to you in person since
I am signed up for 6 units indepent study and it might be benificial to get
your feedback.
John Hobby
Glad you're making progress. I'll give you a continuation grade. When the
program is working will come the task of writing up the heuristics
informatively.
∂08-Dec-81 1728 TOB what do you think
I got a request from the NAM, National Association
of Manufacturers, to speak at a session at the
Annual Washington Policy Conference. Stockman,
Volcker (Federal Reserve), Regan (treasury secretary)
will be speakers. Do you think that it is important?
Tom
∂08-Dec-81 2039 Mike Slocum <MAS at S1-A>
Date: 08 Dec 1981 2033-PST
From: Mike Slocum <MAS at S1-A>
CC: jk at SU-AI, jmm at SU-AI, jmc at SU-AI
To: jk%sail
cc: jmm%sail,jmc%sail
From: mike slocum, mas%s1-a, CS206 student [for credit],
at Lawrence Livermore Lab.
Re: ekl, and my difficulties with it...
-------------------------------------------------------
Below is the symptom, any hints on how to fix it?
∂08-Dec-81 2005 MAS PPSAV'd text
.r ekl
proof?
* (proof copypf)
COPYPF started.
* (get-proofs eklisp)
;#404035 PC AT WHICH ATTEMPT TO WRITE INTO PURE PAGE
*
etc... more of the same, and i cannot "(quit)" to leave ekl.
"eklisp" was created by jmm on sail, and ftp'ed to the s1 foonley.
This version of ekl is circa OCT 1, 1981, and was also ftp'ed to s1.
∂08-Dec-81 2107 Mike Slocum <MAS at S1-A>
Date: 08 Dec 1981 2101-PST
From: Mike Slocum <MAS at S1-A>
CC: jmm at SU-AI, jmc at SU-AI
To: jmm%sail
cc: jmc%sail
From: mike slocum, mas%s1-a, CS206 student [for credit],
at Lawrence Livermore Lab.
Re: ekl, and my difficulties with it...
and what to do about it vv CS206...
-------------------------------------------------------
I have been working on the ekl assignment (steadily) since receiving
the assignment handout (only a week ago, Tuesday 12/1). I have been having NO
luck with ekl. The old version (circa 10/1) dies, and I have been unsuccessful
in getting documentation for the newer version (circa 12/1), (see realted
mail concerning the ekl problem).
The question is, what now? Working at LOTS is out of the question, I have
no transportation to get there (and a long distance phone link is
expensive).
I am also concerned that none of the homework which I have turned in
has been returned to me. Did you get it, (or my mid term for that matter)?
If I am not in danger of failing the class for lack of the 3rd and 4th homework
assignments, perhaps it would be best for me to forget about turning in this
4th problem set. However, I would like to do it very much.
Jitandra: I will try to call, but you might contact me at 422-4923 if
you get the chance.
∂08-Dec-81 2320 Mike Slocum <MAS at S1-A> yes
Date: 08 Dec 1981 2309-PST
From: Mike Slocum <MAS at S1-A>
Subject: yes
To: jmc at SU-AI
--------------------
from: mas%s1-a
Yes, I cannot figure out why the old version of ekl is trying to
write pure pages.
I have gotten 'hold of a manual for the new version of ekl, i'm just
trying somew of it's features...
any suggestions?
∂09-Dec-81 0109 Jerry E. Pournelle <POURNE at MIT-MC> urgent
Date: 9 December 1981 04:07-EST
From: Jerry E. Pournelle <POURNE at MIT-MC>
Subject: urgent
To: JEP at MIT-MC, jmc at SU-AI, llw at SU-AI
I would like to write up the Shackleford Project idea. It might
or might not be made to fly, but it ought to be publicised.
I want also for L-5 Society to announce Project Luna, in which
we look for all the possible problems of a lunar colony and
publish our thoughts, getting our technical members to
contribute; we can do a fairly hefty syustem pre-design within
l-5 society, getting a lot of ground work done, and soon.
I think this could be made to catch the imagination of a LOT of
people. I have ideas for mass participation, and I woul dhope
to enlist some academics who might assign Project Luna as class
projects, as O'Neill did with colonies.
Discussion?
Jerry
PS Just returned from briefings in Hawaii. Will also have
reports on othr stuff soon. JEP
Jerry:
I don't see that announcing Shackleton is urgent. We are
proceeding with planning though perhaps too slowly. We can give
ourselves even another year provided we believe we are spending
it effectively. However, should our private planning relapse into
inactivity, then it is our obligation to put the idea into the
public domain. The planning done so far will help that should we
decide that our interests are too diverse to concentrate on
Shackleton.
I'm all for an L-5 Project Luna, although I'm not one of
those academics in a position to assign such things as class
projects. Incidentally, I had lunch with O'Neill recently. His
ideas are still too grandiose in my opinion, but the two practical
projects he is doing, redesigning his lunar mass launcher and
getting some experiments started on processing lunar materials,
are both worthwhile and will help any future efforts.
- John
Two further points.
1. Lowell's S-1 project at Livermore now has its own ARPAnet
connection, so he is best addressed as LLW@S1. However, mail addressed
to him here is automatically forwarded, so nothing is lost.
2. I have been thinking further about the desirability of
a popular book on surviving nuclear war. I have some notes called
SURVIV[F81,JMC], but I have three books to read that I hope will
make some matters more clear, and therefore it might be better to
wait till I mail you something.
I have been set off by the recent propaganda blasts by
"Physicians for Social Responsibility", and I think one can write
a book that avoids the Civil Defense manual approach and also the
"every man for himself" approach. I hope for your participation
in this project or at least advice.
∂09-Dec-81 0126 Jerry E. Pournelle <POURNE at MIT-MC>
Date: 9 December 1981 04:23-EST
From: Jerry E. Pournelle <POURNE at MIT-MC>
To: JMC at SU-AI
cc: POURNE at MIT-MC, llw at S1-A
You're probably right that Shackleford can wait a while,
provided that SOMETHING on it continues; but I do believe it can
be a really vital thing, with some real attractiveness. It
would probably take five to ten years to raise the enthusiasm
adn thus the money; but I think the money could in fact be
forthcoming and the whole project actually be made to work.
I agree that O'Neill looks farther out that we need to just now,
bnut your observations on what he's doing are on the mark. The
projects are worthwhile..
JEP
∂09-Dec-81 0133 Jerry E. Pournelle <POURNE at MIT-MC>
Date: 9 December 1981 04:31-EST
From: Jerry E. Pournelle <POURNE at MIT-MC>
To: JMC at SU-AI
Re book: I now write a column for SURVIVE Magazine, and the
sales potential of a really good book on the subject are in the
75,000 copies at $12.50 per copy range.
It is possible with a book of that sort to do small
press publishing in which authors get considerably more than 10
- 15% of the cover price per copy.
It would be possible for us to get a LOT of free
advertising among potential readers. As you may know, LUCIFER'S
HAMMER is now a cult book among survivalists, and continues to
sell; (in fact we just t more royalties over and above the
$250,000 advance for the book)
The field of survival literature, done right, is very lucrative
just now. One can do well by doing good...
∂09-Dec-81 0838 FFL Reminder from Gene Golub's office
To: JMC, FFL
Luncheon at 12:15, Faculty Club, C. Hurd.
∂09-Dec-81 0853 Gene Golub <CSD.GOLUB at SU-SCORE> [Gene Golub <CSD.GOLUB>: Lunch with C. Hurd]
Date: 9 Dec 1981 0850-PST
From: Gene Golub <CSD.GOLUB at SU-SCORE>
Subject: [Gene Golub <CSD.GOLUB>: Lunch with C. Hurd]
To: tob at SU-AI, or.dantzig at SU-SCORE, csd.feigenbaum at SU-SCORE,
rwf at SU-AI, csd.golub at SU-SCORE, csd.herriot at SU-SCORE, dek at SU-AI,
jmc at SU-AI, csl.crc.ejm at SU-SCORE, csd.ullman at SU-SCORE
This is a reminder that the luncheon meeting with Cuthbert Hurd at 12:15
at the faculty club.
---------------
Date: 20 Nov 1981 1152-PST
From: Gene Golub <CSD.GOLUB>
Subject: Lunch with C. Hurd
To: tob at SU-AI, csd.buchanan, or.dantzig, csd.feigenbaum, rwf at SU-AI,
csd.golub, csd.herriot, dek at SU-AI, jmc at SU-AI, csl.crc.ejm,
csd.ullman
cc: csd.irmgild
There will be a lunch on Wednesday, December 9 with C. Hurd at the Faculty
Club. Cuthbert is interested in helping the Department raise its endowment.
Please let me know as soon as possible whether you can come.
-------
-------
∂09-Dec-81 0953 FFL
To: JMC, FFL
Please call Dave Gifford at Xerox-Palo Alto. 494-4478. re work you have
been doing on LISP proving programs.
∂09-Dec-81 1000 JMC*
Erdos 1:15 erl320
∂09-Dec-81 1008 FFL
To: JMC, FFL
Please call Keith Stobie, 408 725 6348. He is one of your advisees.
He wishes to talk with you about a leave of absence.
∂09-Dec-81 1230 Oded Anoaf Feingold <OAF at MIT-MC> bad mailing ...
Date: 9 December 1981 13:01-EST
From: Oded Anoaf Feingold <OAF at MIT-MC>
Subject: bad mailing ...
To: ENERGY-DSN at MIT-MC
Y'all are hereby enjoined to pay no attention to the message with a
bunch of nuls in it, sent out 12/5/81 at 0920 or some such.
The actual message should have reached you by now.
The flaky messge was generated by starting to send using MAIL, then
discovering that MC thought I was on a printing terminal when I wasn't.
So I interrupted that, told the system what kind of terminal I was on,
and did <esc>g to get back in, properly formatted. I landed in an EMACS
mailer, and assumed it did the right thing when I sent. I wuz wrong.
I neither know nor care about why the mapping failed the first time around.
Thanx for the notice(s) anyway. Anyone who has not received an energy
mailing after the flaky one should let me know.
Oafishly yours,
Oded
∂09-Dec-81 1320 SIS Colloq. Schedule for Dec. 15th
To: "@COLLOQ.DIS[INF,CSD]" at SU-AI
Colloquium Schedule for December 15th
12/15/81
Tuesday
TYPE Lecture
EVENT Medical Computing Journal Club
PLACE MJ301
TIME 1:30 p.m.
PERSON Dr. Glenn Rennels
FROM Stanford University
TITLE ``Deepthroat & Decision Making''.
12/15/81
Tuesday
TYPE Lecture
EVENT Knowledge Representation Group
PLACE MJ301
TIME 2:30 p.m.
PERSON Professor Mike Genesereth
FROM Stanford University
TITLE ``Meta-Level Reasoning in MRS''.
∂09-Dec-81 1338 Gene Golub <CSD.GOLUB at SU-SCORE> Re: Lunch with C. Hurd
Date: 9 Dec 1981 1338-PST
From: Gene Golub <CSD.GOLUB at SU-SCORE>
Subject: Re: Lunch with C. Hurd
To: tob at SU-AI, csd.buchanan at SU-SCORE, or.dantzig at SU-SCORE,
csd.feigenbaum at SU-SCORE, rwf at SU-AI, csd.herriot at SU-SCORE,
dek at SU-AI, jmc at SU-AI, csl.crc.ejm at SU-SCORE, csd.ullman at SU-SCORE
cc: csd.irmgild at SU-SCORE, CSD.GOLUB at SU-SCORE
In-Reply-To: Your message of 20-Nov-81 1152-PST
I guess there was a mix-up in the dates. Cuthbert regrets he did not join us
today. He will be giving the Department 100 shares of IBM stock for
the Hurd Scholarship funds. I propose we use the dividends initially
for travel for students to scientific conferences. GENE
-------
∂09-Dec-81 1354 Gene Golub <CSD.GOLUB at SU-SCORE> Re: Lunch with C. Hurd
Date: 9 Dec 1981 1338-PST
From: Gene Golub <CSD.GOLUB at SU-SCORE>
Subject: Re: Lunch with C. Hurd
To: tob at SU-AI, csd.buchanan at SU-SCORE, or.dantzig at SU-SCORE,
csd.feigenbaum at SU-SCORE, rwf at SU-AI, csd.herriot at SU-SCORE,
dek at SU-AI, jmc at SU-AI, csl.crc.ejm at SU-SCORE, csd.ullman at SU-SCORE
cc: csd.irmgild at SU-SCORE, CSD.GOLUB at SU-SCORE
In-Reply-To: Your message of 20-Nov-81 1152-PST
I guess there was a mix-up in the dates. Cuthbert regrets he did not join us
today. He will be giving the Department 100 shares of IBM stock for
the Hurd Scholarship funds. I propose we use the dividends initially
for travel for students to scientific conferences. GENE
-------
∂09-Dec-81 1503 CG
To: JMC at SU-AI, CLT at SU-AI, ZM at SU-AI, VRP at SU-AI
Here's the schedule for the MTC qual
Date: December 14
Place: Meet at John McCarthy's office
Students: Yellin, Malachi, Malik, Weening
Examiners : McCarthy, Manna, Pratt, Talcott, Goad
The examining committee is partitioned into two subcommittees:
(1) {Manna, Pratt}
(2) {McCarthy, Talcott, Goad}
Each student will be examined by only one of the subcommittees,
according to the following schedule:
Committee (1): Frank Yellin at 9:30, Yoni Malachi at 11:00
Committee (2): Jitendra Malik at 9:30, Joe Weening at 11:00
The examiners should meet at 9:00am at McCarthy's office.
Please send complaints or questions about this schedule to cg@sail.
∂09-Dec-81 1606 JMM MTC qual
To: CG
CC: JMC, CLT
I will not be taking the MTC qual .
Jitendra
∂09-Dec-81 1730 pratt@Shasta (SuNet) schedule
Date: 9 Dec 1981 17:23:14-PST
From: pratt at Shasta
To: CG@SU-AI, CLT@SU-AI, JMC@SU-AI, VRP@SU-AI, ZM@SU-AI
Subject: schedule
As I already told Chris, I'll be at a DARPA meeting Dec 14-16.
Vaughan
∂10-Dec-81 1010 Gene Golub <CSD.GOLUB at SU-SCORE> [SAMET at USC-ISIF: hello]
Mail-from: ARPANET site SU-SCORE rcvd at 10-Dec-81 1002-PST
Date: 10 Dec 1981 1001-PST
From: Gene Golub <CSD.GOLUB at SU-SCORE>
Subject: [SAMET at USC-ISIF: hello]
To: Faculty at SU-SCORE
Does anyone have a seminar for Hanan to speak in? The Colloquium that week
has a speaker. If so, communicate with Hanan directly. GENE
---------------
Mail-from: ARPANET site USC-ISIF rcvd at 3-Dec-81 2119-PST
Date: 3 Dec 1981 2119-PST
From: SAMET at USC-ISIF
Subject: hello
To: golub at SCORE
Gene,
Hello there. I am on sabbatical this year and am going to the orient for
about a month an a half in January. On my return there is a good possibility
that I will be coming through San Francisco. I would be interested in
visiting Stanford and would be quite pleased to give a colloquium.
The date would be February 16 and I would probably arrive on February 15.
A suitable topic is "Quadtrees for Image Processing." Please
let me know if this would be possible, etc.
Things have been quite hectic for me recently. I have a rather large
grant with Azriel Rosenfeld and am quite busy. I am enjoying though and am
glad about my career choice. Dianne is doing real well in ur department
but I guess I did not have to tell you that!
Best regards and hope to see you one of these days.
Hanan,
-------
-------
∂10-Dec-81 1013 DEK luncheon talk
In case you haven't seen the announcement, Vaughan Pratt is speaking
this noon to the Algorithms for Lunch Bunch (12:15 in room 301).
He is going to tell us about his most recent work on Dynamic Logic,
which (from the abstract at least) seems like it would be interesting
to you. The abstract is posted on the second floor by the mailboxes,
under Thursday.
∂10-Dec-81 1629 TW Comprehensive committee
To: "@COMP.DIS[1,TW]" at SU-AI
To the comprehensive committee:
It is time to prepare the winter comps. I would like to have a first
meeting next week at which we can get people started working on questions
and can work out a schedule for January. The meeting will be on Tuesday
at 1:30 in MJH 352 (next to the coffee area). If you can't make it then,
let me know. If you receive this and didn't realize you were on the
comprehensive committee, we should discuss it.
See you Tuesday. --t
∂10-Dec-81 2115 lantz@Shasta (SuNet) ARPA gateway
Date: 10 Dec 1981 21:03:20-PST
From: lantz at Shasta
To: equip
Subject: ARPA gateway
As some of you know, Network Graphics is putting up the money for an
ARPANET/Ethernet gateway, to the tune of over $20K. Because this hardware
will benefit CSD in general, I would like to be reimbursed when the
facilities funding comes through (or be allocated the same amount of money
to spend on NG-related things). There is already a budget item for
things like this; I'm just making it known that this gateway qualifies.
Keith
∂11-Dec-81 0230 Robert Elton Maas <REM at MIT-MC> Permutation group algorithm/program
Date: 11 December 1981 05:27-EST
From: Robert Elton Maas <REM at MIT-MC>
Subject: Permutation group algorithm/program
To: REM-DIARY-READERS at MIT-MC, MINSKY at MIT-MC, JMC at SU-AI,
LES at SU-AI
On Nov. 24 I finally found an algorithm for "compiling" a permutation
group so that membership in the group can be determined quickly/instantly.
This past evening (Dec. 10) I finished debugging a LISP program that
performs this algorithm. It takes only about 9.5 minutes of realtime
to diagonalize a permutation group on 20 letters. Having done so,
determining whether a given element is or is not in that group is trivial.
I would like to understand a bit more about your algorithm.
Questions:
1. How is the permutation group presented? By generators presented as
products of cycles?
2. What is the form of a "diagaonalized" permutation group?
∂11-Dec-81 0245 Robert Elton Maas <REM at MIT-MC> permutation-group algorithm
Date: 11 December 1981 05:42-EST
From: Robert Elton Maas <REM at MIT-MC>
Subject: permutation-group algorithm
To: JMC at SU-AI
Date: 11 Dec 1981 0236-PST
From: John McCarthy <JMC at SU-AI>
I would like to understand a bit more about your algorithm.
Questions:
1. How is the permutation group presented? By generators presented as
products of cycles?
Yes essentially. Internally it uses ASSOC lists, thus the permutation
((1 2 3) (4 7)) would be ((1 . 2) (2 . 3) (3 . 1) (4 . 7) (7 . 4)).
But it's trivial to convert between the two notations.
2. What is the form of a "diagaonalized" permutation group?
Analagous to upper-diagonal normal form for systems of linear equations.
The first equation lets you eliminate the first variable by expressing it
in terms of the 2nd thru nth, then the second equation lets you eliminate
the second by expressing in terms of 3rd thru nth, etc. until the nth
is eliminated.
My diagonalizations allows a permutation to be converted into another
(modoule the group) that fixes the first letter, then into one that fixes
first two letters, etc. until all letters are fixed (if the element is
in the group) or until at some ply you find there's no way to fix
that next letter you want to fix (if element isn't in group).
I understand the answer to the first question but not the second. Could
you give an example of a diagonalization?
∂11-Dec-81 0309 Robert Elton Maas <REM at MIT-MC>
Date: 11 December 1981 06:07-EST
From: Robert Elton Maas <REM at MIT-MC>
To: JMC at SU-AI
cc: REM at SU-AI
Date: 11 Dec 1981 0248-PST
From: John McCarthy <JMC at SU-AI>
I understand the answer to the first question but not the second. Could
you give an example of a diagonalization?
A diagonalization tells at the top level what generator of the group
to multiply in order to cancel the effect of moving the first letter
away from iself. The next-to-top level considers the subgroup that
fixes that first letter, and tells what to multiply to move the
second letter back to itself without moving the first. The third lvel
tells how to move back to third letter without moving first or second. Etc.
For example, here's such a diagonalization for a group on 5 letters:
Row label is letter you're trying to fix, column label is what it got
moved to and you want to to move it back from there to row label.
1 2 3 4 5
1 () (1 3 2) (1 2 3) (1 4) (2 5 3) (1 4 3 5)
2 () (2 3) (4 5) (2 5 3 4) (2 4 3 5)
3 () (3 5 4) (3 4 5)
4 () (4 5)
5 ()
The above table will allow reducing ANY permutation on those 5 letters
1,2,3,4,5 to the identity permutation (), thus the group is the
symmetry group of order 5! = 1*2*3*4*5 = 120. For example, starting
with (1 5 4 2 3) you want to return 5 back to 1 so you look in
row 1 column 5 and see (1 4 3 5), so (1 5 4 2 3) * (1 4 3 5) = (2 5 3 4),
then you want to return 5 back to 2 so you look in row 2 column 5
and see (2 4 3 5), so (2 5 3 4) * (2 4 3 5) = () the identity so you
stop. If there were any missing entries in the table and you hit one
of them during this algorithm, it would mean the permutation you
started with wasn't in the group. (All entries are present IFF the
group is the symmetry group on all the letters.)
I see how it works for the full symmetric group, and I tried it for
the Klein Vierergruppe consisting of (1), (1 2)(3 4), (1 3)(2 4) and
(1 4)(2 3). It was possible to fill in the first row of the table
but no more, since every permutation that keeps 1 fixed also keeps
the others fixed. The group generated by (1 2 3 4) was equally trivial.
What's the first non-trivial example? Are the alternating groups on
four and five letters good examples?
∂11-Dec-81 0415 Robert Elton Maas <REM at MIT-MC>
Date: 11 December 1981 07:13-EST
From: Robert Elton Maas <REM at MIT-MC>
To: JMC at SU-AI
Alternating groups have every entry except the bottommost present
(ignoring the main diagonal which consist of identity elements).
They aren't very interesting.
Rubics cube and the like generate nontrivial groups, much smaller than
the alternating group (like maybe 1/6 of it) but much much larger than
a tiny group such as Klein group.
The sliding-pegs-in-cross puzzle that RWG gave me last night turned out
to be A20 (i.e. order 20!/2) which was a good test of the program
working (see PERG1.TTY[1,REM]) but not an intersting result.
I want to install softwar to convert between cycle notation and
assoc-list notation before I tackle the Rubic cube et al.
∂11-Dec-81 0856 Ichiki at SRI-AI 12/17 TINLUNCH MEETING
Date: 11 Dec 1981 0854-PST
From: Ichiki at SRI-AI
Subject: 12/17 TINLUNCH MEETING
To: amsler, appelt, archbold, cheeseman, farley, gaschnig, grosz,
To: haas, hendrix, hobbs, kaplan, konolige, lewis, lowrance,
To: bmoore, michell, nilsson, pmartin, robinson, jrobinson,
To: jrosenschein, rperrault, sagalowicz, shieber, stan, stickel,
To: tyson, waldinger, walker, warren, wilkins, zadeh,
To: wde at SRI-KL, goguen at SRI-KL, csd.genesereth at SU-SCORE,
To: ah at SU-AI, drh at SU-AI, kaplan at PARC, kay at PARC,
To: broberts at BBNG, sidner at BBND, stefik at PARC,
To: wasow at PARC-MAXC, ml at SAIL, jed at SAIL, dlo at SAIL,
To: brachman at USC-ISI, utexas-karttunen, utexas-peters,
To: jjk at SAIL, briansmith at PARC, levesque at SRI-KL,
To: dekleer at PARC-MAXC, jmc at SAIL, montalvo at LBL-UNIX,
To: bobrow at PARC-MAXC, csd.hewitt at SCORE,
To: sandra at SRI-KL
Scheduled Person--Barbara Grosz, 12/17
The next TINLUNCH meeting will be held Thursday, 12/17 in Conference
Room A in the A Building from 12:00 till 1:00. The two papers to be
discussed are in the IJCAI-81 Proceedings:
KNOWLEDGE ACQUISITION IN THE CONSUL SYSTEM
By David Wilczynski
USC/Information Sciences Institute
4676 Admiralty Way
Marina del Rey, CA 90291
Pages 135-140
REPRESENTATION AND INFERENCE IN THE CONSUL SYSTEM
By William Mark
USC/Information Sciences Institute
4676 Admiralty Way
Marina del Rey, CA 90291
Pages 375-381
The two authors will not be here--there are extra copies of the papers
on top of Barbara's file cabinet.
-------
∂11-Dec-81 1222 CG
To: CLT, JMC, ZM
Here's yet another schedule for the MTC qual (with any luck, the last).
Date: December 14
Place: Meet at John McCarthy's office
Students: Yellin, Malachi, Weening
Examiners : McCarthy, Manna, Pratt, Talcott, Goad
The examining committee is no longer partitioned into two subcommittees;
the examination of each student will be conducted by the entire committee.
The examinations will start at 1:00pm. We expect that
each examination will take between one and one and one-half hours.
The exams are scheduled one hour and fifteen minutes apart,
with the expectation that the schedule will not be followed exactly.
The order in which the students will be seen is:
(1) Weening (1:00), (2) Malachi(2:15), (3) Yellin(3:30)
As usual, send complaints or questions about this schedule to cg@sail.
The examiners should meet at McCarthy's office at 12:45.
∂11-Dec-81 1215 David A. Moon <MOON at MIT-MC> LISPM Array Timings
Date: 11 December 1981 15:13-EST
From: David A. Moon <MOON at MIT-MC>
Subject: LISPM Array Timings
To: GLR at MIT-AI
cc: BUG-LISPM at MIT-MC, GJS at MIT-AI, LISPTiming at SU-AI
It is a known bug that arrays larger than the size of main memory
don't work very well, in particular they have to be paged in at
least twice to create them. Fixing this requires remodularizing
part of the microcode, which is why it hasn't been done yet. This
has been discussed over (bug lispm) several times in the past.
I don't see what purpose is served by timing things that are known
to be broken.
∂11-Dec-81 1224 CG correction #23
To: JJW at SU-AI, FNY at SU-AI, YM at SU-AI, ZM at SU-AI, JMC at SU-AI,
CLT at SU-AI
I neglected to delete Pratt from the list of examiners; he will not be
on the examining committe.
∂11-Dec-81 1224 PJH via ROCHESTER cobol
hi john, hope you are well. I and Rochester are now fairly well
acquainted and i think we will get along. (blech, isnt that cute?
but you sewhatimean.)
i now seem to have fairly good acess to the net, so this is intended to
be the first move in what i hope can become
a moderately regular correspondence. The point of this is to ask you
whether theree is any sustance to the vague idea i heard third
or fourth hand that ou were working on a symbolic language in which
business machines could talk to one another, and if so whether there
is any chance you might like a tempoary collaborator next summer,
as it seems that i might (subject to negotiation) be visiting SRI for
a month or so then, and I would
like very much to spend some or all of that time working near you, if that
could be fixed up. It sounds as though the COBOL idea might be able to
come up with the funds, is the point.
Anyway, have a good xmas out of the snow (24 inches yesterday)
PAT
∂11-Dec-81 1243 Robert Elton Maas <REM at MIT-MC>
Date: 11 December 1981 15:39-EST
From: Robert Elton Maas <REM at MIT-MC>
To: JMC at SU-AI
So, feel free to send me any permutation-groups you want to know about,
up to about 50 letters (the complete Rubic Cube group is 48 letters
and is of order 24! * 24! / 12 if I remember correctly from lengthy
work by hand, the one I ran last night is 20 letters and took 9.5 minutes
to run and is of order 20! / 2 according to my program).
I'm going to the graphic-gathering polyhedra-puzzle meeting tomorrow
afternoon at PA cultural center and I'm gonna collect puzzles that are
equivalent to permutation groups and feed them all into my algorithm.
∂11-Dec-81 1839 Mike Slocum <MAS at S1-A> final.f81
Date: 11 Dec 1981 1819-PST
From: Mike Slocum <MAS at S1-A>
Subject: final.f81
To: jmc at SU-AI
-----------------------------------------------------
Thank's for the final...
The class was great, but many more examples of some simple EKL
proofs would have helped, (in the exact EKL syntax). The manual
also needs some work (eg. more examples, of more features).
mas.
∂12-Dec-81 0749 Robert Elton Maas <REM at MIT-MC> Update on PERG
Date: 12 December 1981 10:47-EST
From: Robert Elton Maas <REM at MIT-MC>
Subject: Update on PERG
To: RWG at MIT-MC, JMC at SU-AI
I've added code to convert to/from cycle notation. This makes the
trace output more readable, and makes it easier to enter new problems
in cycle notation. I guess i'll try Rubic's cube now, the simplified
versions first, then if I'm stil awake the full thing...
∂14-Dec-81 1056 FFL
To: JMC, FFL
John Yen, a student at the U. of Santa Clara, who says he spoke with you
a couple of weeks ago, called to ask if he could see you this Tuesday, Dec. l5,
at about 3 p.m. I am to let him know. 246 3952 or 984 458l.
∂14-Dec-81 1124 Gene Golub <CSD.GOLUB at SU-SCORE> Lunch
Mail-from: ARPANET site SU-SCORE rcvd at 14-Dec-81 1121-PST
Date: 14 Dec 1981 1120-PST
From: Gene Golub <CSD.GOLUB at SU-SCORE>
Subject: Lunch
To: faculty at SU-SCORE
The final lunch for the quarter will take place tomorrow.
As a special treat, Zohar Manna will be present; he's here
for two weeks. GENE
-------
∂14-Dec-81 1534 Gene Golub <CSD.GOLUB at SU-SCORE> Lunch on Dec 15 (Second Notice)
Mail-from: ARPANET site SU-SCORE rcvd at 14-Dec-81 1527-PST
Date: 14 Dec 1981 1524-PST
From: Gene Golub <CSD.GOLUB at SU-SCORE>
Subject: Lunch on Dec 15 (Second Notice)
To: faculty at SU-SCORE
cc: CSD.GOLUB at SU-SCORE
The lunch on Tuesday will take place in MJH 146. Jim Gibbons who is
the chairman of the CIS executive committee will talk about the
"property rights" of CIS. The lunch begins at 12:15.
GENE
PS We'll be having extra sandwiches this time and goodies!
-------
Let me mention that a discussion of "property rights in CIS" is for
me a strong incentive to not come.
∂15-Dec-81 0118 Robert Elton Maas <REM at MIT-MC> PERG and Rubic
Date: 15 December 1981 04:16-EST
From: Robert Elton Maas <REM at MIT-MC>
Subject: PERG and Rubic
To: BIL at MIT-MC, JMC at SU-AI
I tested my algorithm on Rubic's cube subgroups:
(1) Ignoring where corners end up, including all physical motions
but where edge cubes go only -- order = 24*22*20*...*10*8*6*4 = 12!*2↑11
(2) Ignoring where edges end up, including all physical motions
but where corner cubes go only -- first run encountered data error,
fixed it, now it gives order = 24*21*18*15*12*9*6 = 8! * 3↑7
Now that I know the data is correct, gonna try the whole cube...
∂15-Dec-81 0341 Robert Elton Maas <REM at MIT-MC>
Date: 15 December 1981 06:39-EST
From: Robert Elton Maas <REM at MIT-MC>
To: HOEY at MIT-MC, BIL at MIT-MC, JMC at SU-AI
Well, my algorithm worked for the full Rubic cube group.
5 generators (6th is a dunzell), order = (8! 3↑8 12! 2↑12) / 12
Throwing in 3 kinds of disassembly (swap faces of an edge cube <2>,
rotate faces of a corner cube <3>, and swap two edge cubes <2>)
and it gave the expected 8! 3↑8 12! 2↑12.
Now to tackle the RUbic Pyramid and any othe problems anybody wants
to throw at my algorithm....
∂16-Dec-81 1015 FFL
To: JMC, FFL
Elliott Bloom's secretary called to remind you on your lunch date with
him at noon at Faculty Club Thursday - tomorrow.
∂16-Dec-81 1109 Ichiki at SRI-AI TINLUNCH MEETING IN BUILDING -A-
Date: 16 Dec 1981 1107-PST
From: Ichiki at SRI-AI
Subject: TINLUNCH MEETING IN BUILDING -A-
To: tlgrp:
Scheduled Person--Barbara Grosz, 12/17
The next TINLUNCH meeting will be held Thursday, 12/17 in Conference
Room A in the A Building from 12:00 till 1:00.
NOTE: For those of you who do not know how to get to the
Administration (A) Building, it is east of the Engineering (E)
Building, on Ravenswood Avenue with a horseshoe driveway. Look for
the flag in the middle of the driveway.
The two papers to be discussed are in the IJCAI-81 Proceedings:
KNOWLEDGE ACQUISITION IN THE CONSUL SYSTEM
By David Wilczynski
Pages 135-140
REPRESENTATION AND INFERENCE IN THE CONSUL SYSTEM
By William Mark
Pages 375-381
The two authors will not be here--there are extra copies of the papers
on top of Barbara's file cabinet.
-------
∂16-Dec-81 1345 Gene Golub <CSD.GOLUB at SU-SCORE> Applicant
Date: 16 Dec 1981 1341-PST
From: Gene Golub <CSD.GOLUB at SU-SCORE>
Subject: Applicant
To: Jmc at SU-AI
Philip Wadler wants to spend two years at Stanford upon completion
of his degree. Do you remember him? any comments?
GENE
-------
∂16-Dec-81 1631 Marilynn Walker <CSD.MWALKER at SU-SCORE> TGIF FOR DENNY BROWN
Date: 16 Dec 1981 1627-PST
From: Marilynn Walker <CSD.MWALKER at SU-SCORE>
Subject: TGIF FOR DENNY BROWN
To: CSD-Faculty: ;
cc: csd.mwalker at SU-SCORE
Come one, Come all!
To bid Denny a kind and fond farewell...
When: Friday, December 18, 1981
Time: 4:30 p.m.
Where: The Oval (weather permitting)
-------
∂16-Dec-81 1828 Gene Golub <CSD.GOLUB at SU-SCORE> Applicant
Date: 16 Dec 1981 1827-PST
From: Gene Golub <CSD.GOLUB at SU-SCORE>
Subject: Applicant
To: faculty at SU-SCORE
cc: csd.dietterich at SU-SCORE, jf at SU-AI
Philip Wadler of CMU wishes to be considered for a position here.
He'll be in the Bay Area until Jan 20 ( his parents live in San Jose).
He could give a talk on his thesis: LISTLESSNESS IS BETTER THAN LAZINESS.
(Describes algorithm for autoatically transforming functional programs
to be more efficient, similar to lazy evaluation at compile time)
Does anyone have a free seminar slot for Wadler to speak?
Please let me know as early as possible. GENE
-------
∂17-Dec-81 0346 Mike Slocum <MAS at S1-A> Homework number 4.
Date: 17 Dec 1981 0333-PST
From: Mike Slocum <MAS at S1-A>
Subject: Homework number 4.
To: jmm at SU-AI, jmm at SU-AI
CC: jmc at SU-AI, MAS at S1-A
-----------------------------------------------------
Jitendra:
I finished as much of homework 4 as I had time for,
(tho incidently at LEAST 80% of my time was spent
struggling with the syntax of the new version of
ekl, and the semantics of getting things to work
as expected). As you will note, only the first part
of the assignment is done, and I'm not sure it really
proves anything.
The EKL PPR file is: HW4.PPR[1,mas] on the S1-A
machine.
Thanks for all your help, hopefully next time things
will go more smoothly.
Mike Slocum
Lawrence Livermore Lab.
L-535 4224923 mas%s1-a
∂17-Dec-81 1051 FFL
To: JMC, FFL
A Visiting Professor in Mech. Engr., Prof. Holst, called to say he had had
a letter from a colleague in Stockholm, Prof. Schnittner(sp?), who is
coming here with 12 students the second week of February. He had visited
at the DC Power Lab and met you, and would like to met with you with his
students during the time they are here. Prof. Holst asks that you call him
about this at 7-0960.
∂17-Dec-81 1303 Gene Golub <CSD.GOLUB at SU-SCORE> Sabbatical Leave
Date: 17 Dec 1981 1301-PST
From: Gene Golub <CSD.GOLUB at SU-SCORE>
Subject: Sabbatical Leave
To: faculty at SU-SCORE
Is anyone going on sabbatical during the next academic year?
There is a possibility I might want to invite a visitor with
the thought of eventual hiring. GENE
-------
∂17-Dec-81 1338 FFL
To: JMC, FFL
Ursula Maydell of Edmonton called. Will call again about 3 p.m.
∂17-Dec-81 1429 MAS
To: "@TOM.DIS[1,MAS]" at SU-AI
The dinner party will not be at Tom's house. It will be at:
580 Arastradero Road, The Penthouse (PH on elevator)
1/2 block from El Camino toward Foothill on the right.
Park in the street. Sunday at 6 PM.
∂17-Dec-81 1624 Gene Golub <CSD.GOLUB at SU-SCORE> Faculty
Date: 17 Dec 1981 1541-PST
From: Gene Golub <CSD.GOLUB at SU-SCORE>
Subject: Faculty
To: faculty at SU-SCORE
cc: csd.bscott at SU-SCORE, csd.mwalker at SU-SCORE
The next Faculty meeting will be held Jan 5, 1982. If you have any
agenda items, please send them to me. GENE
-------
∂17-Dec-81 1643 TW Comprehensive committee
To: "@COMP.DIS[1,TW]" at SU-AI
CC: csd.walker at SU-SCORE, csd.napier at SU-SCORE
The comprehensive committee met on Tuesday December 15, and made the
following decisions:
1. The comp will be held on February 6
2. We will hold weekly meetings from the beginning of the quarter until
the comp is over and graded. Tentatively the time for those
meetings is Tuesdays 2:30-4. If this is not possible,
alternatives are Mon. 1:15-3 and Fri. 1:30-3.
3. Our first meeting (regardless of the decision in item 2) will be
on Friday, January 8 at 1:30
4. Before that meeting, each person is to have generated a set of potential
questions and gotten them to my secretary (Rosemary Napier, room 340)
to be duplicated. She should have them all by Wed. Jan. 6. Everyone
is welcome to make up problems in all areas, but we have assigned
primary responsibility for areas as follows:
AA: Knuth, Spencer
AI: McCarthy, Lenat, Greiner
Hardware: Lantz, Wiederhold
MTC: McCarthy, Weening
NA: Herriott, student to be announced
Systems: Wiederhold, Winograd, Harkness
You are hereby requested to do the following:
By Wed., December 23: Let me know if the time for meetings is OK
By Wed., Jan 6: Have copies of proposed questions in to Rosemary
On Fri., Jan 8: Show up for the meeting in room 252 at 1:30.
Have a happy holiday,
-t
∂17-Dec-81 1811 CLT
∂17-Dec-81 1640 JMC
∂17-Dec-81 1429 MAS
To: "@TOM.DIS[1,MAS]" at SU-AI
The dinner party will not be at Tom's house. It will be at:
580 Arastradero Road, The Penthouse (PH on elevator)
1/2 block from El Camino toward Foothill on the right.
Park in the street. Sunday at 6 PM.
I will probably still be in Martinez Sunday evening.
I am going to the library, now.
∂18-Dec-81 1104 CLT
Here is my cousins address and phone number
Mrs. H.C. Fowler
1330 Vine Avenue, Martinez
415-228-1964
See you Sunday or Monday
∂18-Dec-81 1318 Gene Golub <CSD.GOLUB at SU-SCORE> Gift to Department
Date: 18 Dec 1981 1315-PST
From: Gene Golub <CSD.GOLUB at SU-SCORE>
Subject: Gift to Department
To: faculty at SU-SCORE
cc: su-bboard at SU-SCORE
I am very pleased to say that Dr. Cuthbert Hurd has donated 100
shares of IBM stock to establish the Cuthbert C. Hurd Scholarship
in Computer Science. The annual interest will be used to help
graduate students in the department attend meetings.
GENE GOLUB
-------
∂19-Dec-81 1505 Gifford at PARC-MAXC Recursive Programs as Functions in a First Order Theory
Date: 19 Dec 1981 14:34 PST
From: Gifford at PARC-MAXC
Subject: Recursive Programs as Functions in a First Order Theory
To: jmc@sail
cc: Gifford
I have draft of the above paper dated February 12, 1979. Is there a newer
version available? If so, is there a press file online for it?
Thanks,
Dave Gifford
∂19-Dec-81 1606 Gene Golub <CSD.GOLUB at SU-SCORE> Wadler Visit
Date: 19 Dec 1981 1604-PST
From: Gene Golub <CSD.GOLUB at SU-SCORE>
Subject: Wadler Visit
To: faculty at SU-SCORE
cc: wadler at CMU-10A
We had hoped to have Phil Wadler speak at the CS Colloquium on Jan 5
but this is not an especially good date for him since he needs to be
in Portland on Jan 6. Does anyone have a seminar/colloquium spot for
Wadler the week of Jan 11? If nothing else he can speak on 1/5/82.
Gene
-------
∂20-Dec-81 0238 Jerry E. Pournelle <POURNE at MIT-MC>
Date: 20 December 1981 05:36-EST
From: Jerry E. Pournelle <POURNE at MIT-MC>
To: JMC at SU-AI
Re: Abandoned mines:
As a matter of fact, yes. There are some fairly good locations
and low prices, too.
Talked to Heinlein today. He'd love to hel with Shackleton; I
didn't really give him all lowdown, but he is interested in
seeing things happen that will lead to space colonies. At his
age he doesn't expect to live to see them, so what he wants is
to be part of doing something that will lead to them.
I fear he's not feeling too well.
Jerry
∂20-Dec-81 0952 JMM exams
I have finished grading 1 question and am in the midst of the second.I am in my
office and will be there through most of the day.
∂20-Dec-81 1307 JMM
Are you planning to come down to Margaret Jacks sometime ? I have finished grading
2 questions .
∂20-Dec-81 1321 RWW
Is carolyn around? I'd like to talk to her.
She's in Martinez with sick cousin. She may be back today but more likely
tomorrow.
∂20-Dec-81 1506 Gifford at PARC-MAXC PS: Recursive Programs as Functions in a First Order Theory
Date: 20 Dec 1981 15:06 PST
From: Gifford at PARC-MAXC
Subject: PS: Recursive Programs as Functions in a First Order Theory
To: jmc@sail
cc: Gifford
I'm also interested in the book you are writing with Carolyn, "Lisp with Proofs".
Is there a draft that I might look at?
Thanks,
Dave Gifford
∂20-Dec-81 1736 RWG
∂20-Dec-81 1407 JMC
She's in Martinez with sick cousin. She may be back today but more likely
tomorrow.
Thanks
∂21-Dec-81 1042 LGC cc of msg to REG on SAIL phone hookup
∂21-Dec-81 1035 LGC SAIL Phone Hookup /cc jmc
To: REG
The cost of a foreign-exchange Mountain View phone for me would be in the
neighborhood of $125.00 per month ($6.40 or $6.75 per mile per month for
15 - 18 miles, depending on who you talk to, plus regular phone rental and
unlimited service. The estimated cost of a private tie line is in the
neighborhood of $75.00 per month. On this basis, JMC prefers that I have
a private tie line, and is willing to pay for it. -- Lew
∂21-Dec-81 1231 RWW
Directions to the church
Foothill Congregational Church
Orange Avenue, Los Altos
From Palo Alto go toward San Jose on Foothill expressway.
Go to Main Street Los Altos turn right (away from the town) and then
immediately left. follow the left hand on the wall rule to stay close
foothill. There are three churches. Foothill is the middle one.
The service will be at 10:30am sharp!! Please try to arrive by 10:15.
Thanks. I am glad you are comming.
PS: Don't forget the dinner at 7pm at our house.
∂21-Dec-81 1436 Guy.Steele at CMU-10A Results of Common LISP Meeting
Date: 21 December 1981 1727-EST (Monday)
From: Guy.Steele at CMU-10A
To: alan at MIT-AI, brooks at MIT-AI, rlb at MIT-MC, gsb at MIT-ML,
hic at MIT-AI, gjc at MIT-MC, David.Dill at CMU-10A,
Scott.Fahlman at CMU-10A, chiron at CMU-20C, jkf at UCB-C70,
rpg at SU-AI, Joseph.Ginder at CMU-10A, zubkoff at CMU-20C,
Walter.VanRoggen at CMU-10A, Gail.Kaiser at CMU-10A, rg at MIT-AI,
hedrick at rutgers, eak at s1-a, jlk at MIT-MC, jmc at SU-AI,
dm at utah-20, moon at MIT-AI, scherlis at CMU-20C, rms at MIT-AI,
vanmelle at PARC-MAXC, masinter at PARC-MAXC, weinreb at MIT-AI,
jonl at MIT-MC, rz at MIT-MC
Subject: Results of Common LISP Meeting
Message-Id: <21Dec81 172702 GS70@CMU-10A>
I have now compiled all the results of the meeting, including
late comments, votes, and all decisions, into two new documents.
One is called "Decisions on the ...". It is the document
"Votes on the ..." which was distributed at the meeting, augmented
with the meeting contents. It is about 250 pages long. The other
is a summary document called "Revisions to the ...". It is about
25 pages long, and gives just the title and result for each issue.
That alongside the Votes document is probably enough to see what
happened; the Decisions document contains the total flamage.
The PRESS files for these are called DECISIONS.PRESS and REVISIONS.PRESS.
Right now they are on directory <SLISP.SLM> @ CMUA, and I will soon
transport them to directory GLS; @ MIT-AI, whence they may be transported
by others thither and yon.
--Guy
∂21-Dec-81 1449 Guy.Steele at CMU-10A Correction: files on MIT-MC, not MIT-AI
Date: 21 December 1981 1737-EST (Monday)
From: Guy.Steele at CMU-10A
To: alan at MIT-AI, brooks at MIT-AI, rlb at MIT-MC, gsb at MIT-ML,
hic at MIT-AI, gjc at MIT-MC, David.Dill at CMU-10A,
Scott.Fahlman at CMU-10A, feinberg at CMU-20C, jkf at UCB-C70,
rpg at SU-AI, Joseph.Ginder at CMU-10A, zubkoff at CMU-20C,
Walter.VanRoggen at CMU-10A, Gail.Kaiser at CMU-10A, rg at MIT-AI,
hedrick at rutgers, eak at s1-a, jlk at MIT-MC, jmc at SU-AI,
dm at utah-20, moon at MIT-AI, scherlis at CMU-20C, rms at MIT-AI,
vanmelle at PARC-MAXC, masinter at PARC-MAXC, weinreb at MIT-AI,
jonl at MIT-MC, rz at MIT-MC
Subject: Correction: files on MIT-MC, not MIT-AI
Message-Id: <21Dec81 173738 GS70@CMU-10A>
I will put the files on MIT-MC, not MIT-AI. There is
more disk space there. This may not happen for a day or so
after all. Watch for the files to appear!
--Guy
∂21-Dec-81 1530 Gene Golub <CSD.GOLUB at SU-SCORE> Au Pair Girl
Date: 21 Dec 1981 1527-PST
From: Gene Golub <CSD.GOLUB at SU-SCORE>
Subject: Au Pair Girl
To: faculty at SU-SCORE
If anyone is interested in a French au pair girl, please contact me.
Gene
-------
∂21-Dec-81 1608 vanMelle, Masinter Re: Results of Common LISP Meeting
Date: 21 Dec 1981 16:08 PST
Sender: vanMelle at PARC-MAXC
Subject: Re: Results of Common LISP Meeting
In-reply-to: Guy.Steele's message of 21 December 1981 1727-EST (Monday)
To: Guy.Steele at CMU-10A, alan at MIT-AI, brooks at MIT-AI, rlb at MIT-MC,
gsb at MIT-ML, hic at MIT-AI, gjc at MIT-MC, David.Dill at CMU-10A,
Scott.Fahlman at CMU-10A, chiron at CMU-20C, jkf at UCB-C70, rpg at SU-AI,
Joseph.Ginder at CMU-10A, zubkoff at CMU-20C, Walter.VanRoggen at
CMU-10A, Gail.Kaiser at CMU-10A, rg at MIT-AI, hedrick at rutgers, eak at
s1-a, jlk at MIT-MC, jmc at SU-AI, dm at utah-20, moon at MIT-AI, scherlis at
CMU-20C, rms at MIT-AI, weinreb at MIT-AI, jonl at MIT-MC, rz at MIT-MC
cc: vanMelle, Masinter
From: vanMelle, Masinter
We prepared a report on Common Lisp and the November meeting
for circulation to the Interlisp community. A draft of the report is on
[parc-maxc]<netlisp>comlisp.report
accessible with login as user anonymous password anything. We would
appreciate any comments you have before it is circulated more widely,
especially if you disagree strongly with any representation we have made.
Bill & Larry
∂21-Dec-81 2125 Gene Golub <CSD.GOLUB at SU-SCORE> Faculty meeting Jan 5, 1982
Date: 21 Dec 1981 2120-PST
From: Gene Golub <CSD.GOLUB at SU-SCORE>
Subject: Faculty meeting Jan 5, 1982
To: Faculty at SU-SCORE
Our first faculty meeting of the New Year will take place Jan 5
at 2:30. I will ask the chairmen of our various committees to make
a report. GENE
-------
∂22-Dec-81 1000 JMC*
John Hansen about Pratt
∂22-Dec-81 1317 Marvin Minsky <MINSKY at MIT-AI> AAAI Tutorial Chairman
Date: 22 December 1981 16:14-EST
From: Marvin Minsky <MINSKY at MIT-AI>
Subject: AAAI Tutorial Chairman
To: MINSKY at MIT-AI, GJS at MIT-AI, PHW at MIT-AI, sacerdoti at SRI-KL,
rick at RAND-AI, jmc at SU-AI, tw at SU-AI, reddy at CMU-10B,
simon at CMU-10A, newell at CMU-10A, athompson at USC-ECL,
bobrow at PARC-MAXC, winograd at PARC-MAXC, dwaltz at BBND,
webber at BBND, woods at BBND, erman at USC-ISIF,
balzer at USC-ISIF, buchanan at SUMEX-AIM, engelmore at SUMEX-AIM,
feigenbaum at SUMEX-AIM, feldman at SUMEX-AIM,
aaai-office at SUMEX-AIM, bledsoe at UTEXAS-20, grosz at SRI-AI,
nilsson at SRI-AI, walker at SRI-AI
Chuck Rich is new AAAI Tutorial Chairman. His net-name is RICH@MIT-AI.
He already has a good working relation with Lou Robinson.
∂22-Dec-81 1348 Ichiki at SRI-AI TINLUNCHES FOR 1/7 AND 1/14
Date: 22 Dec 1981 1349-PST
From: Ichiki at SRI-AI
Subject: TINLUNCHES FOR 1/7 AND 1/14
To: tlgrp:
The TINLUNCH speaker originally scheduled for 1/7 is Norman Haas.
That TINLUNCH is being cancelled due to that Thursday being so soon
after our return from the Holidays, and Norman Haas no longer being
here.
The TINLUNCH speaker for 1/14 is JERRY HOBBS (Jerry, note that is YOU!).
The meeting will again be held in Conference Room A of Building A; more
information forthcoming next year (1982).
-------
∂23-Dec-81 0900 JMC*
654-7458
∂23-Dec-81 1412 TW meeting time
To: "@COMP.DIS[1,TW]" at SU-AI
Several of you have not answered the message about meeting times.
Please do. If you did, please excuse the lazy use of the full address list.
-t
∂24-Dec-81 0900 JMC*
654-7458
∂24-Dec-81 0922 Gene Golub <CSD.GOLUB at SU-SCORE>
Date: 24 Dec 1981 0920-PST
From: Gene Golub <CSD.GOLUB at SU-SCORE>
To: JMC at SU-AI
In-Reply-To: Your message of 24-Dec-81 0110-PST
Thanks for your quick response.
GENE
-------
∂24-Dec-81 1832 Lowell Wood <LLW at S1-A> Forward Motion
Date: 24 Dec 1981 1833-PST
From: Lowell Wood <LLW at S1-A>
Subject: Forward Motion
To: jmc at SU-AI
CC: LLW at S1-A, RAH at S1-A, danny at MIT-AI,
minsky at MIT-AI
∂23-Dec-81 0046 JMC via SU-AI
Any movement on Shackleton?
[John: Rod has most of the transit vehicle design scoping and mass
budgeting done. I'm grinding away on the organizational options,
structures, costs and schedules, and I presume (because I haven't heard
from him for a while) that Danny is productively essaying on what's to be
done once Rod's systems get us there. I believe that we'll all still
shooting for an exchange of pretty complete drafts right after the New
Year's Day weekend. I braced Jack Schmitt privately over dinner last week
on what should be done during the next couple of decades, and he
enthusiastically embraced the `maximally austere' approach to absolutely
everything; aside from unmistakable vestiges of the classic NASA Pyramid
Mentality and a desire to colonize the Martian surface ASAP, I think he'll
be a pillar of strength for our efforts. Lowell]
∂25-Dec-81 0900 JMC*
ursula
∂26-Dec-81 1546 ZM
To: JJW
CC: JMC
Next logic qual -- last week of February. Zohar
∂27-Dec-81 1042 Marvin Minsky <MINSKY at MIT-MC>
Date: 27 December 1981 13:40-EST
From: Marvin Minsky <MINSKY at MIT-MC>
To: jmc at SU-AI, rah at S1-A, llw at S1-A
Those SPACE people on the network are talking about skyhooks.
Should we tell them?
∂27-Dec-81 1656 TOB Clara Torda
John
Do you know anything about her? Who does?
tom
Clara Torda is retired from some New York academic institution. She asked
permission to audit CS206 in Fall 1980 and CS226 in Winter 1981. Her
login QCT is left over from then. I have no idea who pays for her
computer time. It's not me, and I hope it's not you. She has a book on
memory and the brain which I'll be glad to lend you if you want to take
the trouble to look at it. Is she a problem? How come you ask me?
∂27-Dec-81 1732 Lowell Wood <LLW at S1-A> Pearls Before Swine
Date: 27 Dec 1981 1732-PST
From: Lowell Wood <LLW at S1-A>
Subject: Pearls Before Swine
To: minsky at MIT-AI
CC: LLW at S1-A, RAH at S1-A, jmc at SU-AI
∂27-Dec-81 1043 Marvin Minsky <MINSKY at MIT-MC>
Date: 27 December 1981 13:40-EST
From: Marvin Minsky <MINSKY at MIT-MC>
To: jmc at SU-AI, rah at S1-A, llw at S1-A
Those SPACE people on the network are talking about skyhooks.
Should we tell them?
[Marvin: A closer look indicates that they are babbling, not talking,
about skyhooks (except of course for HPM). We haven't tried to keep the
Starbridge non-public (as indicated by the `preprint' we distributed at
Pournelle's meeting last January), but I would suggest that our
self-respect would compel us to `go public' in a forum with a less
fragrant reputation than SPACE. Rod's within a couple of months of
getting a completed document out, I estimate, and I therefore counsel
continued patience. On a related topic, I haven't seen hide-nor-hair
of Danny recently--is he grinding diligently on his share of the
Shackelton write-up? Lowell]
∂27-Dec-81 1847 Lowell Wood <LLW at S1-A> Needed Item
Date: 27 Dec 1981 1847-PST
From: Lowell Wood <LLW at S1-A>
Subject: Needed Item
To: danny at MIT-AI
CC: LLW at S1-A, RAH at S1-A, jmc at SU-AI, minsky at MIT-AI
Dear Danny:
It has become very clear that we need a `moon-mover' (the local analog of an
`earth-mover' which can accomplish three major tasks with minimum hassle:
1) Scooping out one or more hemi-cylindrical depressions in which to
install our Kevlar habitation balloons (so that their undersides can be
more readily, more uniformly and less riskily load-bearing).
2) Depositing a least 0.3 (and preferably 1) meter of soil over the most
exposed (upper) portion of these balloons (post-inflation), for thermal
and UV/EUV isolation and meteorite and hard solar/cosmic ray shielding.
(A foot is about a thermal skin depth for the local day/nite, and gives
decent decoupling from hard radiation; FRPC occupational exposure standards
would suggest about 10 feet for the nastiest solar storm ever recorded,
and the `maximum austerity' standard corresponds to about a meter: you
take several dozen rem if caught in the nastiest-ever storm. It's also a
little easier on your thermal control system to be three skin depths isolated
from the local surface.)
3) Hauling local material to (and then from) a processing site (which
initially will be just a gas bake-out station). (Jack Schmitt is adamant
that virtually all soil sample contain an STP cc of hydrogen per cc of
material, 90% of which comes off by 500 C and all of it by 1000 C; he also
cheerily added that you get all the sulfur which you could possibly want,
in the process.)
Some sort of reasonably articulated conveyor `belt' with a `maw' or
`feeder' attachment at its ingestion end would seem to be satisfactory for
the job, but you may have better ideas. It needs to be able to chew
through a large volume of material before its useful life is over, so that
its ingestion unit must be able to go a least 1 (and preferably 3) meters
below the local surface, and it needs to be able to somehow cover a large
area semi-systematically without a lot of human supervision
(raster-scanning?; diverging spirals?).
Obviously, the `something' we need should be highly reliable, readily
unjammed/repaired, low in power demand and mass, and moderately high in
capacity. Depending to some extent on whether we use direct solar or
photovoltaic energy to do the baking (which clearly takes much more energy
per unit mass than does the tranportation), we'll have something of the
order of 10 kW during the day for processing (and none at night--the
initial nighttime energy budget is remarkably tight, just because the
night is so long). This will permit us to process about 20 grams of
material per second, assuming that we take it all the way to 1000 C and
have only 50% heat recovery through counter-current mass flows (thereby
producing about 3 moles of hydrogen gas per hour, if Schmitt is right).
If we want to initially cover a 5 meter diameter, 20 meter long balloon,
we'll have about 200 square meters equivalent of surface, which 20
grams/second of delivery will cover at the rate of 1 cm depth per day. It
would be nice to be an order of magnitude faster (though you can dodge
both thermal and UV/EUV problems indefinitely with a negligible mass of
superinsulation), just to get some non-trivial meteoric protection fairly
soon--you really have to cover with most of the 30 cm minimal layer depth
in one shot, because the superinsulation crushes with the first cm or so
of junk on top of it and your thermal insulation due to it mostly goes
away. Rod will be delivering us some anticipated holing rates due to
direct meteor `shine', but Schmitt raised the hitherto (by me, at least)
unanticipated problem of secondary `shine' from material spattered out of
the surface nearby by direct impacts--the indirect high-speed mass flux
near the surface could easily be an order of magnitude greater than the
direct one. How one digs the balloon emplacement hemi-cylindrical volume
in a period during which one could reasonably live on the lander looks
quite challenging--the mass to be moved is comparable to the maximum (of 1
meter minimum thickness) which we would like to throw over the top of the
balloon, once it's emplaced.
Anyway, a power-efficient gadget which could haul around somewhere between
20 and 200 grams/second of material for a total operating life of at least
100 million seconds (implying that, if it `mines' to a mean depth of 2
meters, it needs to range over a total of 1000-10000 square meters, and
will gather/scatter material yielding up 1-10 tonnes of water during its
3-year projected lifetime) along the lines sketched above would be useful to
the point of being essential. What can you come up with that does this
job and folds up into negligible volume and has even less mass?
Just as a background note, Rod and I have bargained to a compromise of 3
kWe time-averaged power supply during the night initially, which requires
the storage of about 300 kg of hydrogen and oxygen and the service of a 90
kg fuel cell (7 kWe average capacity, w/ 12 kWe 15 minute peak). I'm
determined to get at least one spare fuel cell included in the package (as
well as the electrolytic cell capacity to generate fuel for it during the
day), not only for first-contingency reasons, but because extracting the
first ton of local water will give us another 10 kW of nighttime power
supply. The lander's fuel tanks provide free, high-grade storage for the
liquified fuel cell feedstreams, and the skillful lander pilot gets a
bonus-- the more quickly he sets the lander down, the more hydrogen and
oxygen are left in the fuel tanks to initially power his little ecosystem
during the long nites, and the more interesting power-using activities in
which he gets to engage for these periods. Full exercise of these
considerations will serve to upgrade the `quality of life' for half the
time from a little above subsistence to something really interesting.
Sorry for the last-minute interjection. Hope the rest of the writing's
going well!
Lowell
∂28-Dec-81 1527 FFL
To: JMC, FFL
Please call Stanley Brodsky at SLAC. Says it is important.
854-3300 X2644.
∂28-Dec-81 1528 FFL Call from student at H-P in Colorado Springs, John Romano
To: JMC, JMM, FFL
Wanted to talk with one of you. Says he has been unable to complete work
for CS206 on Lots and has some other questions. Says neither he nor otherss
taking the course kthere have received any material back. I did not know
if there was any reason for him to expect to get material back from us. If
there is, then I will check with TV network. He will call back next
week during the afternoon.
∂28-Dec-81 1623 FFL
To: JMC, FFL
Your optician called to say he has the frame for your glasses now.
∂28-Dec-81 1656 FFL
To: JMC, FFL
The name is GABOR BELOVARI, Dept. of Mathematical Sciences, Kent State, Kent,
Ohio 44242.
Please send a note to Prof. Ursula Maydell, Computer Science Dept.
University of Alberta, Edmonton, Albert, CANADA with Belovari's name
and address. If he sent me a curriculum vitae, send a copy of that
too.
∂28-Dec-81 1906 TOB
John
I asked because she has submitted a draft of a paper for publication
at MIT, I think for Robotics Research. It seemed obscure and I was
asked to get a calibration on her to see whether it is worthwhile
to put effort into deciding how to get it rewritten.
Is there anyone here who knows her well?
Tom
∂28-Dec-81 1624 JMC
Clara Torda is retired from some New York academic institution. She asked
permission to audit CS206 in Fall 1980 and CS226 in Winter 1981. Her
login QCT is left over from then. I have no idea who pays for her
computer time. It's not me, and I hope it's not you. She has a book on
memory and the brain which I'll be glad to lend you if you want to take
the trouble to look at it. Is she a problem? How come you ask me?
I think the paper should be judged on its merits independently of its
source.
∂29-Dec-81 1223 Mike Farmwald <PMF at S1-A>
Date: 29 Dec 1981 1221-PST
From: Mike Farmwald <PMF at S1-A>
To: jmc at SU-AI
I didn't make it to IBM due to both a mixup and time constraints.
I'll call John Cocke and talk to him. I have a new parallel machine design
which you might be interested in. It's main goal is to make task creation
as cheap as possible. I believe I have a way of making task creation
cost essentially zero (i.e. creating a task can be done in addition to
doing a useful operation). Also process-to-process communication costs
are very small.
∂29-Dec-81 1625 Gene Golub <CSD.GOLUB at SU-SCORE> Philip Wadler
Date: 29 Dec 1981 1621-PST
From: Gene Golub <CSD.GOLUB at SU-SCORE>
Subject: Philip Wadler
To: faculty at SU-SCORE,
RECRUITMENT: ;
cc: ADMIN: ;,
ADMIN: ;
Philip Wadler will be giving the colloquium on Tuesday, Jan 5.
He is also being considered for an appointment. Please let me
know whether you are able to see him on the Tuesday. Remember
we have a faculty meeting at 2:30. GENE
-------
∂29-Dec-81 2319 Mike Slocum <MAS at S1-A> Fall cs206 final...
Date: 29 Dec 1981 2314-PST
From: Mike Slocum <MAS at S1-A>
Subject: Fall cs206 final...
To: jmm at SU-AI
CC: jmc at SU-AI, MAS at S1-A
-----------------------------------------------------
Do you know when I hear about my final grade ??
(I assume you got my final okay)
Mike Slocum
As I remember, you weren't on the list of those to whom I was supposed
to give grades. Jitendra will check what we are supposed to have done.
The material exists to give a grade, however.
∂30-Dec-81 1117 Guy.Steele at CMU-10A Text-file versions of DECISIONS and REVISIONS documents
Date: 30 December 1981 1415-EST (Wednesday)
From: Guy.Steele at CMU-10A
To: common-lisp at SU-AI
Subject: Text-file versions of DECISIONS and REVISIONS documents
Message-Id: <30Dec81 141557 GS70@CMU-10A>
The files DECISIONS DOC and REVISIONS DOC on directory GLS;
at MIT-MC are available. They are text files, as opposed to
PRESS files. The former is 9958 lines long, and the latter is
1427.
--Guy
∂31-Dec-81 1152 RPG
decisio.pre[com,lsp
∂31-Dec-81 1235 FFL
To: JMC, FFL
Adele Goldberg called. 494 4385
∂01-Jan-82 1000 JMC*
Call Hurd about Ultra.
∂01-Jan-82 1428 YK takasu + igarashi
do you have a current address of these guys?
Richard
Prof. Satoru Takasu
Research Institute of Mathematical Sciences
Kyoto University
Kyoto, JAPAN
Prof. Shigeru Igarashi
Computer Science Department
Tsukuba University
Tsukuba, JAPAN
∂01-Jan-82 1500 JMC*
Pacific Fire Extinguisher called at 2:32 am about D.C. Power Lab.
∂01-Jan-82 1600 Guy.Steele at CMU-10A Tasks: A Reminder and Plea
Date: 1 January 1982 1901-EST (Friday)
From: Guy.Steele at CMU-10A
To: common-lisp at SU-AI
Subject: Tasks: A Reminder and Plea
Message-Id: <01Jan82 190137 GS70@CMU-10A>
At the November meeting, a number of issues were deferred with the
understanding that certain people would make concrete proposals for
consideration and inclusion in the second draft of the manual. I
promised to get the second draft out in January, and to do that I need
those proposals pretty soon. I am asking to get them in two weeks (by
January 15). Ideally they would already be in SCRIBE format, but I'll
settle for any reasonable-looking ASCII file of text approximately in
the style of the manual. BOLIO files are okay too; I can semi-automate
BOLIO to SCRIBE conversion. I would prefer not to get rambling prose,
outlines, or sentence fragments; just nice, clean, crisp text that
requires only typographical editing before inclusion in the manual.
(That's the goal, anyway; I realize I may have to do some
industrial-strength editing for consistency.) A list of the outstanding
tasks follows.
--Guy
GLS: Propose a method for allowing special forms to have a dual
implementation as both a macro (for user and compiler convenience)
and as a fexpr (for interpreter speed). Create a list of primitive
special forms not easily reducible via macros to other primitives.
As part of this suggest an alternative to FUNCTIONP of two arguments.
MOON: Propose a rigorous mathematical formulation of the treatment
of the optional tolerance-specification argument for MOD and REMAINDER.
(I had a crack at this and couldn't figure it out, though I think I
came close.)
GLS: Propose specifications for lexical catch, especially a good name for it.
Everybody: Propose a clean and consistent declaration system.
MOON/DLW/ALAN: Propose a cleaned-up version of LOOP. Alter it to handle
most interesting sequence operations gracefully.
SEF: Propose a complete set of keyword-style sequence operations.
GLS: Propose a set of functional-style sequence operations.
GJC/RLB: Polish the VAXMAX proposal for feature sets and #+ syntax.
ALAN: Propose a more extensible character-syntax definition system.
GLS: Propose a set of functions to interface to a filename/pathname
system in the spirit of the LISP Machine's.
LISPM: Propose a new error-handling system.
LISPM: Propose a new package system.
∂03-Jan-82 1824 Gene Golub <CSD.GOLUB at SU-SCORE> Cancellation of tenured faculty meeting
Date: 3 Jan 1982 1824-PST
From: Gene Golub <CSD.GOLUB at SU-SCORE>
Subject: Cancellation of tenured faculty meeting
To: tenured-faculty: ;
cc: CSD.GOLUB at SU-SCORE
There will not be a tenured faculty meeting on THursday, Jan 7.
I will be attending my neice's funeral; she was killed in a collision
on 280 by a drunken driver headed in the wrong direction.
Let's plan on meeting Jan 14 at 2:30. GENE
-------
-------
∂04-Jan-82 1108 FFL
To: JMC, FFL
Please call Bob Russel at 654-7458.
No need for letter to Russell; we discussed the matter on the phone.
∂04-Jan-82 1147 Irmgild Schack <CSD.IRMGILD at SU-SCORE> Tuesday Faculty Luncheon
Date: 4 Jan 1982 1141-PST
From: Irmgild Schack <CSD.IRMGILD at SU-SCORE>
Subject: Tuesday Faculty Luncheon
To: CSD-Faculty: ;
This week's faculty luncheon is cancelled. The next lunch-meeting
will be Tuesday, January 12.
GENE
-------