perm filename S79.IN[LET,JMC]1 blob
sn#453688 filedate 1979-07-01 generic text, type C, neo UTF8
COMMENT ⊗ VALID 00442 PAGES
C REC PAGE DESCRIPTION
C00001 00001
C00044 00002 ∂02-Apr-79 1108 Bmoore at SRI-KL (Bob Moore) letter to study group particpants
C00045 00003 ∂02-Apr-79 1211 LES LLL temporary access to NS
C00046 00004 ∂02-Apr-79 1504 Feigenbaum at SUMEX-AIM LOTS visit for high level delegation from Israel
C00048 00005 ∂02-Apr-79 1514 DCO
C00050 00006 ∂02-Apr-79 1604 Bmoore at SRI-KL (Bob Moore) Letter to study group particpants
C00051 00007 ∂02-Apr-79 1743 Mark Crispin <Admin.MRC at SU-SCORE> Dialnet
C00053 00008 ∂02-Apr-79 1819 MRC RFC753.TXT[NET,MRC]
C00054 00009 ∂02-Apr-79 1833 Jrobinson at SRI-KL (Jane Robinson)
C00055 00010 ∂03-Apr-79 0256 REM via AMES-TIP Data-compression programming
C00057 00011 ∂03-Apr-79 0623 REM via AMES-TIP Pruning histogram in IRSM software/hardware
C00060 00012 ∂03-Apr-79 0825 Ijcai79 at SUMEX-AIM IJCAI REviews
C00062 00013 ∂03-Apr-79 0949 Bmoore at SRI-KL (Bob Moore) letter to CASBS study group participants
C00068 00014 ∂03-Apr-79 1128 Feigenbaum at SUMEX-AIM good news
C00069 00015 ∂03-Apr-79 1436 Mcgoveran at SUMEX-AIM Dean's Office message
C00070 00016 ∂03-Apr-79 1454 Gaschnig at SRI-KL Ph.D. student looking for job
C00073 00017 ∂03-Apr-79 1528 LGC via AMES-TIP Leadoff Colloquium Slot
C00075 00018 ∂04-Apr-79 1642 BS Ralph Gorin's Salary Review Date
C00076 00019 ∂05-Apr-79 0857 Feigenbaum at SUMEX-AIM Re: Ralph Gorin's Salary Review Date
C00077 00020 ∂05-Apr-79 1309 SL Meet with you?
C00078 00021 ∂05-Apr-79 1725 TW
C00103 00022 ∂05-Apr-79 1737 ARK S-1 Meeting Times
C00104 00023 ∂06-Apr-79 1656 CLT FOLISP
C00105 00024 ∂06-Apr-79 1716 ME
C00108 00025 ∂06-Apr-79 1721 ME
C00114 00026 ∂07-Apr-79 2242 RWG
C00115 00027 ∂08-Apr-79 0455 DON via Easy Street job slots
C00116 00028 ∂08-Apr-79 0538 DON antidisestablishmentarianism
C00118 00029 ∂08-Apr-79 1523 RPG New Official Policy
C00121 00030 ∂08-Apr-79 1636 Lederberg at SUMEX-AIM V. Turchin
C00123 00031 ∂08-Apr-79 1712 ME one segment programs
C00126 00032 ∂08-Apr-79 2034 PTZ purge
C00127 00033 ∂08-Apr-79 2037 MRC
C00129 00034 ∂08-Apr-79 2316 ME disk drives
C00130 00035 ∂08-Apr-79 2330 MRC messages
C00133 00036 ∂09-Apr-79 0058 ME
C00135 00037 ∂09-Apr-79 0949 PAT
C00136 00038 ∂09-Apr-79 1052 Pickering at SUMEX-AIM reprint copy
C00137 00039 ∂09-Apr-79 1455 Ijcai79 at SUMEX-AIM IJCAI Reviews
C00138 00040 ∂09-Apr-79 1638 DON
C00140 00041 ∂09-Apr-79 1718 Mark Crispin <Admin.MRC at SU-SCORE> [Yeager at SUMEX-AIM: Re: Sumex ''Dialnet'']
C00143 00042 ∂10-Apr-79 0941 Sacerdoti at SRI-KL (Earl Sacerdoti) reference request for David Wilkins
C00145 00043 ∂10-Apr-79 1538 ME
C00146 00044 ∂10-Apr-79 1701 LES Consulting agreement
C00150 00045 ∂11-Apr-79 0043 MLB Overlayed opcodes in FSIM.
C00152 00046 ∂11-Apr-79 0717 Feigenbaum at SUMEX-AIM Re: V. Turchin
C00153 00047 ∂11-Apr-79 1501 Ijcai79 at SUMEX-AIM reviews
C00155 00048 ∂12-Apr-79 1554 DCO
C00157 00049 ∂12-Apr-79 1900 JMC*
C00158 00050 ∂13-Apr-79 0057 REM Appendix to my to-publish paper on data-compression methods
C00160 00051 ∂13-Apr-79 0105 REM
C00161 00052 ∂13-Apr-79 0158 MRC April 6 Beijing Review
C00162 00053 ∂13-Apr-79 1119 Mcgoveran at SUMEX-AIM CPPC Luncheons
C00165 00054 ∂13-Apr-79 2047 100 : ME new file pack
C00167 00055 ∂14-Apr-79 0401 100 : Committee to Nuke the Whales
C00168 00056 ∂14-Apr-79 1300 RWW FOL FIXES
C00169 00057 ∂14-Apr-79 1654 ARK S-1 meeting notice
C00170 00058 ∂15-Apr-79 0452 REM at MIT-MC (Robert Elton Maas) Sorting and searching, data compression
C00174 00059 ∂15-Apr-79 1122 RWW
C00175 00060 ∂15-Apr-79 1452 JC Meeting
C00176 00061 ∂16-Apr-79 0221 MRC So much for "Free China"
C00178 00062 ∂16-Apr-79 0321 LLW Visiting Faculty Appointment for John Reiser
C00182 00063 ∂16-Apr-79 0957 PAT
C00183 00064 ∂16-Apr-79 1400 TW
C00184 00065 ∂16-Apr-79 1349 LES DECUS Meeting
C00185 00066 ∂16-Apr-79 2006 REM at MIT-MC (Robert Elton Maas)
C00187 00067 ∂16-Apr-79 2057 BPM
C00189 00068 ∂17-Apr-79 0154 LES ARPA proposal
C00190 00069 ∂17-Apr-79 0749 Mark Crispin <Admin.MRC at SU-SCORE> DECUS and Dialnet by May 15
C00193 00070 ∂17-Apr-79 1037 PAT visit to ARPA
C00194 00071 ∂17-Apr-79 1519 100 : REM via SU-TIP CRUNCH PATENT
C00195 00072 ∂17-Apr-79 1544 DEW
C00197 00073 ∂17-Apr-79 2015 Zenon.Pylyshyn at CMU-10A (X320ZP51) Your letter re conference in October
C00200 00074 ∂18-Apr-79 0038 MRC importing McMahon to help
C00202 00075 ∂18-Apr-79 0913 DCO
C00204 00076 ∂18-Apr-79 1113 Sacerdoti at SRI-KL (Earl Sacerdoti) Dave Wilkins recommendation
C00205 00077 ∂18-Apr-79 1420 DCL ARPA PROPOSAL and Project costs.
C00207 00078 ∂18-Apr-79 2033 DCL
C00209 00079 ∂19-Apr-79 1916 Davis at OFFICE-2 The TEX Formatter
C00213 00080 ∂20-Apr-79 0311 LLW S-1s at SAIL, Etc.
C00222 00081 ∂20-Apr-79 0324 LLW Proposals for FY80 S-1 Project-Related Work
C00224 00082 ∂20-Apr-79 0455 REM via AMES-TIP Huffman in person
C00225 00083 ∂20-Apr-79 1448 Sacerdoti at SRI-KL (Earl Sacerdoti) thanks
C00226 00084 ∂21-Apr-79 2317 LLW
C00228 00085 ∂22-Apr-79 0957 Feigenbaum at SUMEX-AIM Re: Shrinking CS206
C00229 00086 ∂23-Apr-79 0017 LLW
C00248 00087 ∂23-Apr-79 0918 DPB Shrinking 206.
C00250 00088 ∂23-Apr-79 0924 CSL.FB at SU-SCORE Digital Dataphone Service
C00251 00089 ∂23-Apr-79 1116 Bmoore at SRI-KL (Bob Moore) S and P problem
C00252 00090 ∂23-Apr-79 1633 PAT a faculty meeting
C00253 00091 ∂23-Apr-79 1634 PAT telegram
C00254 00092 ∂23-Apr-79 2153 Feigenbaum at SUMEX-AIM Re: Shrinking 206.
C00255 00093 ∂24-Apr-79 1000 JMC*
C00256 00094 ∂24-Apr-79 1038 ES Mark Thomas Clifton
C00257 00095 ∂24-Apr-79 1112 REP Elephant
C00258 00096 ∂24-Apr-79 1509 MLB
C00259 00097 ∂24-Apr-79 1602 PAT reservations
C00260 00098 ∂24-Apr-79 1947 ES Book for MTC
C00261 00099 ∂25-Apr-79 0013 LLW Automatic Distribution Privileges Availability
C00263 00100 ∂25-Apr-79 0838 DPB
C00264 00101 ∂25-Apr-79 0935 Bmoore at SRI-KL (Bob Moore)
C00265 00102 ∂25-Apr-79 1311 RWW ARPA proposal
C00266 00103 ∂25-Apr-79 1715 Bmoore at SRI-KL (Bob Moore)
C00268 00104 ∂26-Apr-79 0916 CSL.OWICKI at SU-SCORE
C00270 00105 ∂26-Apr-79 0934 Bmoore at SRI-KL (Bob Moore)
C00272 00106 ∂26-Apr-79 0951 RWW arpa proposal
C00274 00107 ∂26-Apr-79 1000 JMC*
C00275 00108 ∂26-Apr-79 1003 Mcgoveran at SUMEX-AIM FACULTY MEETING NEXT TUESDAY
C00278 00109 ∂26-Apr-79 1105 Bmoore at SRI-KL (Bob Moore)
C00279 00110 ∂26-Apr-79 1128 CSL.OWICKI at SU-SCORE
C00280 00111 ∂26-Apr-79 1605 ARK S-1 Meeting
C00281 00112 ∂26-Apr-79 1652 PAT
C00282 00113 ∂26-Apr-79 2050 MRC via SU-TIP Dialnet debugging
C00284 00114 ∂26-Apr-79 2329 RWW arpa proposal
C00285 00115 ∂27-Apr-79 0942 MAX Sato's paper
C00286 00116 ∂27-Apr-79 1221 DCO
C00288 00117 ∂27-Apr-79 1401 TOB visitor
C00289 00118 ∂27-Apr-79 1522 Morton at PARC-MAXC AI Review
C00290 00119 ∂27-Apr-79 1627 MAILER failed mail returned
C00291 00120 ∂27-Apr-79 1748 PAT restored essays
C00294 00121 ∂29-Apr-79 1338 PEG Compatible versions of PCPASC, SOPA, runtimes
C00296 00122 ∂29-Apr-79 2153 RWW
C00297 00123 ∂29-Apr-79 2224 RWW proposal
C00298 00124 ∂29-Apr-79 2227 RWW proposal
C00299 00125 ∂29-Apr-79 2239 RWW proposal
C00300 00126 ∂30-Apr-79 1113 PAT proofs for MI9
C00301 00127 ∂30-Apr-79 1834 Mark Crispin <Admin.MRC at SU-SCORE> Dialnet progress
C00302 00128 ∂30-Apr-79 1835 GUERNYLUSBY at USC-ISI May 3 visit to DARPA
C00304 00129 ∂30-Apr-79 1015 PWM SAIL visit of Dr. Peter Claringbold
C00305 00130 ∂30-Apr-79 0308 RWW
C00306 00131 ∂30-Apr-79 0306 RWW
C00309 00132 ∂01-May-79 0009 WD dog
C00310 00133 ∂01-May-79 0727 MRC Dialnet progress
C00312 00134 ∂01-May-79 0900 morris at PARC-MAXC (repeat) Talks on Semantics, 6/4-6/8
C00313 00135 ∂01-May-79 1008 DCO
C00315 00136 ∂01-May-79 1348 DEW oral exam
C00316 00137 ∂01-May-79 2004 LGC Analysis of Feasibility Concept
C00317 00138 ∂01-May-79 2129 WD via SU-TIP
C00318 00139 ∂02-May-79 0042 BCM noch ein Witz
C00319 00140 ∂02-May-79 0034 BCM politicheskaya shutka dla vas
C00320 00141 ∂02-May-79 0847 MCM Dialnet progress
C00321 00142 ∂02-May-79 0917 SSO via SU-SCORE Seminar
C00322 00143 ∂02-May-79 0930 DEW oral
C00323 00144 ∂02-May-79 1118 PAT III board meeting
C00324 00145 ∂02-May-79 1205 PAT III board meeting and Dave Wilkins orals
C00325 00146 ∂02-May-79 1330 DAN
C00326 00147 ∂02-May-79 1347 DEW oral exam
C00327 00148 ∂03-May-79 1029 REM at MIT-MC (Robert Elton Maas) Continued STOALL search, progress report
C00331 00149 ∂03-May-79 1052 PAT III board meeting
C00332 00150 ∂03-May-79 1226 PAT my absence
C00333 00151 ∂03-May-79 1517 MRC Dialnet alive!
C00336 00152 ∂03-May-79 1527 PAT
C00337 00153 ∂05-May-79 1009 MRC Tops-20 Dialnet documentation
C00338 00154 ∂05-May-79 1302 Jrobinson at SRI-KL (Jane Robinson) Margaret Young
C00340 00155 ∂05-May-79 2212 DD at MIT-AI (David Dyer)
C00342 00156 ∂06-May-79 1102 REM via AMES-TIP MACLISP TOTALLY BROKEN
C00344 00157 ∂06-May-79 1752 MAX GO TO JAPAN
C00345 00158 ∂07-May-79 1551 CSL.SALLY at SU-SCORE Space Needs in the Center for Integrated Systems
C00347 00159 ∂07-May-79 2335 CSL.OWICKI at SU-SCORE (Susan Owicki) Modal logic seminar
C00349 00160 ∂07-May-79 2336 CSL.OWICKI at SU-SCORE (Susan Owicki) [Shel Finkelstein <SJF at SU-AI>: Modal logic seminar]
C00353 00161 ∂08-May-79 0817 WD
C00354 00162 ∂08-May-79 0936 Mcgoveran at SUMEX-AIM Faculty Meeting for 5/14
C00356 00163 ∂08-May-79 1505 RPG Proposal
C00358 00164 ∂08-May-79 1748 Zenon.Pylyshyn at CMU-10A (X320ZP51) ai-Phil conference
C00360 00165 ∂08-May-79 2035 DCO
C00363 00166 ∂08-May-79 2334 Feigenbaum at SUMEX-AIM getting together with John
C00364 00167 ∂09-May-79 1056 PAT IBM San Jose
C00365 00168 ∂09-May-79 1057 PAT III board meeting (again!)
C00366 00169 ∂09-May-79 1115 PAT Macy's calculator
C00367 00170 ∂09-May-79 1643 PLS Potluck
C00368 00171 ∂10-May-79 0700 JB THESIS MANUSCRIPT.
C00369 00172 ∂10-May-79 1042 Mcgoveran at SUMEX-AIM Meeting with Ed and Ralph
C00370 00173 ∂10-May-79 1118 LES SAIL move
C00371 00174 ∂10-May-79 1403 MRC Arpanet directory
C00373 00175 ∂10-May-79 1528 Mcgoveran at SUMEX-AIM Visitng Scholars from China
C00374 00176 ∂10-May-79 1659 Mcgoveran at SUMEX-AIM Faculty Meeting on Monday, May 14
C00376 00177 ∂11-May-79 1054 DEW Hans Berliner's visit
C00377 00178 ∂11-May-79 1141 REP Summer Job
C00378 00179 ∂11-May-79 1549 DEW
C00379 00180 ∂11-May-79 1550 DEW room change for oral exam
C00380 00181 ∂11-May-79 1800 Raphael at SRI-KL (Bert Raphael) OPEN HICE
C00382 00182 ∂11-May-79 1804 Raphael at SRI-KL (Bert Raphael) OPEN HICE Times
C00383 00183 ∂11-May-79 1951 RWW prolegamena paper
C00384 00184 ∂12-May-79 1613 DEW message from Hans Berliner to arrange meeting
C00387 00185 ∂13-May-79 1503 DEW
C00388 00186 ∂13-May-79 1707 RWW dinner tonight
C00389 00187 ∂14-May-79 1025 Bmoore at SRI-KL (Bob Moore) terminals at CASBS
C00391 00188 ∂14-May-79 1318 DPB Jacks Hall third floor
C00394 00189 ∂14-May-79 1447 ARK S-1 Meeting
C00395 00190 ∂14-May-79 1456 100 : ljh
C00396 00191 ∂14-May-79 1531 Engelmore at SUMEX-AIM Discussion of your proposal
C00397 00192 ∂14-May-79 1751 LGC Engelmore's Visit
C00398 00193 ∂14-May-79 1900 LGC CBCL.PRO
C00402 00194 ∂15-May-79 1150 Elschlager at SUMEX-AIM CONCEPTS AS OBJECTS
C00404 00195 ∂15-May-79 1537 CET Help -- Colloquia Speakers Needed
C00405 00196 ∂15-May-79 1654 CLT concert
C00406 00197 ∂15-May-79 1802 DEW Remember my oral!
C00407 00198 ∂16-May-79 0942 DCO
C00410 00199 ∂16-May-79 1110 PAT
C00411 00200 ∂16-May-79 1115 PAT
C00412 00201 ∂16-May-79 1311 MRC reload this morning
C00413 00202 ∂16-May-79 1420 TW via SU-SCORE Summer guest
C00414 00203 ∂16-May-79 1519 Engelmore at SUMEX-AIM Visit to SAIL
C00415 00204 ∂17-May-79 1119 CET MTC Qual
C00416 00205 ∂17-May-79 1220 TW
C00418 00206 ∂17-May-79 1348 PAT
C00431 00207 ∂17-May-79 1414 TW AI qual
C00433 00208 ∂17-May-79 1431 HVA Patte's Memo to JMC
C00434 00209 ∂17-May-79 1544 MAX Information about dinner
C00435 00210 ∂17-May-79 1612 PAT
C00437 00211 ∂18-May-79 1637 Mcgoveran at SUMEX-AIM Status of PRC Scholars: Summer Quarter and Beyond
C00438 00212 ∂18-May-79 1647 CET Bill Miller
C00439 00213 ∂18-May-79 1656 CET
C00440 00214 ∂18-May-79 2332 ME
C00441 00215 ∂21-May-79 1106 PN Help for Users
C00442 00216 ∂21-May-79 1302 DCL Delivery of XGP output to Jacks Hall
C00453 00217 ∂21-May-79 1917 MRC Dialnet
C00454 00218 ∂21-May-79 2026 Zenon.Pylyshyn at CMU-10A (X320ZP51) Access to conputer
C00466 00219 ∂22-May-79 0119 MRC
C00467 00220 ∂22-May-79 0128 LLW
C00471 00221 ∂22-May-79 2137 Zenon.Pylyshyn at CMU-10A (X320ZP51) Terminals etc
C00473 00222 ∂22-May-79 2248 DCL Lab ARPA budget
C00474 00223 ∂23-May-79 0038 HPM
C00475 00224 ∂23-May-79 0042 HPM
C00476 00225 ∂23-May-79 0926 FB VAX OS
C00478 00226 ∂23-May-79 0934 FB VAXes
C00479 00227 ∂23-May-79 0937 FB displays
C00480 00228 ∂23-May-79 1335 Bennett at SUMEX-AIM China
C00483 00229 ∂23-May-79 1453 DCO
C00485 00230 ∂23-May-79 1640 REM via SU-TIP Status of data-compaction R+D
C00489 00231 ∂23-May-79 2247 Buchanan at SUMEX-AIM IJCAI invitation for Ma Xiwen
C00491 00232 ∂23-May-79 2302 DCO
C00492 00233 ∂23-May-79 2348 JED the picnic wants YOU!
C00493 00234 ∂24-May-79 0007 REM via AMES-TIP New demo/R+D program for compression
C00496 00235 ∂24-May-79 0057 LES Manna segment of ARPA proposal
C00497 00236 ∂24-May-79 0837 Feigenbaum at SUMEX-AIM Re: Rosse memo
C00498 00237 ∂24-May-79 1127 MAX
C00499 00238 ∂24-May-79 1328 TOB
C00501 00239 ∂24-May-79 1335 HVA Martin Davis
C00502 00240 ∂25-May-79 0146 LES computer time for music
C00505 00241 ∂25-May-79 0752 Feigenbaum at SUMEX-AIM (Response to message)
C00506 00242 ∂25-May-79 1536 DEW
C00507 00243 ∂25-May-79 2018 FB meeting
C00508 00244 ∂26-May-79 0917 REM via AMES-TIP Addenda on c.r.
C00510 00245 ∂26-May-79 1750 Mark Crispin <Admin.MRC at SU-SCORE> [PBARAN at USC-ISI: PAN - a tiny electronic mail system]
C00519 00246 ∂26-May-79 2317 MRC PCNET mail
C00520 00247 ∂27-May-79 1835 ARR Unused files in [PAS,S1]
C00521 00248 ∂27-May-79 1840 ARR my previous message
C00522 00249 ∂28-May-79 0848 CSL.FB at SU-SCORE The Dover Printer
C00524 00250 ∂28-May-79 1430 PAM
C00525 00251 ∂29-May-79 0759 Feigenbaum at SUMEX-AIM (Response to message)
C00526 00252 ∂29-May-79 0755 FB Dover and XGP
C00528 00253 ∂29-May-79 0804 FB Subject of Friday's meeting
C00529 00254 ∂29-May-79 0832 Feigenbaum at SUMEX-AIM meeting Friday
C00530 00255 ∂29-May-79 1319 Feigenbaum at SUMEX-AIM meeting Friday
C00531 00256 ∂29-May-79 1358 ARK via SU-TIP S-1 Meeting Reminder
C00532 00257 ∂29-May-79 1537 MAX
C00533 00258 ∂29-May-79 1558 100 : ljh Message
C00534 00259 ∂29-May-79 2336 LGC via AMES-TIP Engelmore Visit
C00535 00260 ∂30-May-79 0258 ME
C00536 00261 ∂30-May-79 1055 PAT
C00537 00262 ∂30-May-79 1136 ARR S-1 meeting today
C00538 00263 ∂30-May-79 1151 JMC call
C00539 00264 ∂30-May-79 1203 Morris at PARC-MAXC Testing, 1, 2, 3, ...
C00540 00265 ∂30-May-79 1209 Morris at PARC-MAXC Xerox PARC Semantics Week Lectures
C00547 00266 ∂30-May-79 1417 TW via SUMEX-AIM Assignments for Qual
C00552 00267 ∂30-May-79 1546 DCL Lab computer costs
C00553 00268 ∂30-May-79 1549 HVA MARTIN DAVIS
C00554 00269 ∂30-May-79 1811 MRC
C00555 00270 ∂31-May-79 1926 ME 1-controller system
C00556 00271 ∂01-Jun-79 0742 PHY
C00557 00272 ∂01-Jun-79 1028 REM at MIT-MC (Robert Elton Maas)
C00562 00273 ∂01-Jun-79 1125 Feigenbaum at SUMEX-AIM divorce settlement
C00563 00274 ∂01-Jun-79 1133 Feigenbaum at SUMEX-AIM reminder:meting
C00564 00275 ∂01-Jun-79 1203 TW via SU-TIP Schedule for exams next Thursday
C00566 00276 ∂01-Jun-79 1343 Morris at PARC-MAXC Semantics Week schedule change
C00568 00277 ∂02-Jun-79 0015 LGC IJCAI Passage
C00575 00278 ∂02-Jun-79 1139 DCL ARPA proposal
C00576 00279 ∂02-Jun-79 1435 YYY via SU-TIP IBM visit
C00577 00280 ∂02-Jun-79 1548 Mark Crispin <Admin.MRC at SU-SCORE> G.xxx people
C00578 00281 ∂02-Jun-79 1604 MRC via SU-SCORE
C00579 00282 ∂04-Jun-79 0158 MINSKY at MIT-AI (Marvin Minsky)
C00580 00283 ∂04-Jun-79 0204 Feigenbaum at SUMEX-AIM mccarthy's views on terminal system
C00584 00284 ∂04-Jun-79 1555 Zenon.Pylyshyn at CMU-10A (X320ZP51) Center
C00586 00285 ∂04-Jun-79 2000 DCL Cost of operating AI lab computer
C00588 00286 ∂04-Jun-79 2040 LGC Proposal Revision
C00589 00287 ∂04-Jun-79 2210 LES SAIL computer cost
C00590 00288 ∂05-Jun-79 0130 JB
C00592 00289 ∂05-Jun-79 0141 JB YES.
C00593 00290 ∂05-Jun-79 0853 SELFRIDGE at BBN-TENEXD tech comittee notes
C00603 00291 ∂05-Jun-79 0910 Feigenbaum at SUMEX-AIM Acting Assistant Professor appointment
C00604 00292 ∂05-Jun-79 1106 DEW Comments on my thesis
C00605 00293 ∂05-Jun-79 1341 DCL
C00607 00294 ∂05-Jun-79 1642 DPB End of quarter department meetings
C00611 00295 ∂06-Jun-79 0215 LLW Visiting Appointment for John Reiser
C00612 00296 ∂06-Jun-79 0925 LES Reiser appointment
C00613 00297 ∂06-Jun-79 1347 Feldman at SUMEX-AIM concept reification
C00615 00298 ∂06-Jun-79 1419 PERLMAN at BBN-TENEXA Any new ideas on internet congestion?
C00621 00299 ∂06-Jun-79 1608 ARK S-1 Meeting Notice
C00622 00300 ∂06-Jun-79 1700 JMC*
C00623 00301 ∂06-Jun-79 2038 RWW arpa proposal
C00624 00302 ∂06-Jun-79 2206 LGC ARPA Proposal
C00630 00303 ∂06-Jun-79 2217 ARK Corrected S-1 Meeting Notice
C00631 00304 ∂07-Jun-79 0652 DCO
C00632 00305 ∂07-Jun-79 1417 PLS
C00633 00306 ∂07-Jun-79 1500 LES
C00634 00307 ∂07-Jun-79 1626 ARR Give a subject to your mail.
C00635 00308 ∂07-Jun-79 1810 REM at MIT-MC (Robert Elton Maas) Better than Huffman-code in yet another way
C00638 00309 ∂08-Jun-79 0059 LLW Offer of Appointment to Reiser
C00640 00310 ∂08-Jun-79 1105 LES Reiser appointment
C00641 00311 ∂08-Jun-79 1513 LGC Perlis
C00642 00312 ∂08-Jun-79 1636 REM at MIT-MC (Robert Elton Maas)
C00644 00313 ∂08-Jun-79 1810 REM at MIT-MC (Robert Elton Maas)
C00647 00314 ∂08-Jun-79 1816 PBARAN at USC-ISI (Response to message)
C00648 00315 ∂08-Jun-79 1820 REM at MIT-MC (Robert Elton Maas)
C00650 00316 ∂08-Jun-79 2234 DCO
C00653 00317 ∂09-Jun-79 0245 CERF at USC-ISI Re: Any new ideas on internet congestion?
C00655 00318 ∂09-Jun-79 1629 JMC
C00656 00319 ∂10-Jun-79 1709 ARR nd MSG files
C00657 00320 ∂11-Jun-79 1241 Feigenbaum at SUMEX-AIM moving the TIP
C00658 00321 ∂11-Jun-79 1750 JMC remaining on estate
C00659 00322 ∂11-Jun-79 1834 ADMIN.GORIN at SU-SCORE Re: moving the TIP
C00660 00323 ∂11-Jun-79 2240 LES Moving the TIP
C00661 00324 ∂12-Jun-79 1439 PAT lecture
C00662 00325 ∂12-Jun-79 1443 PAT King paper
C00663 00326 ∂12-Jun-79 1600 JMC*
C00664 00327 ∂13-Jun-79 0243 LLW Reiser Bio
C00668 00328 ∂13-Jun-79 1139 CLT meeting of the minds
C00669 00329 ∂13-Jun-79 1400 JMC*
C00670 00330 ∂13-Jun-79 1720 LES Discussion of move to Jacks Hall
C00671 00331 ∂13-Jun-79 2258 DCO
C00672 00332 ∂14-Jun-79 0600 PMF Leave of absence
C00674 00333 ∂14-Jun-79 0845 CET Rich Pattis grade
C00675 00334 ∂14-Jun-79 1353 DEW
C00676 00335 ∂14-Jun-79 1427 PERLMAN at BBN-TENEXA source notification meeting
C00677 00336 ∂14-Jun-79 1658 Ijcai79 at SUMEX-AIM IJCAI-79 Travel
C00682 00337 ∂15-Jun-79 0315 CERF at USC-ISI Re: source notification meeting
C00683 00338 ∂15-Jun-79 0824 Ijcai79 at SUMEX-AIM IJCAI-79 Invited Speakers &Panels
C00686 00339 ∂15-Jun-79 1032 Mcgoveran at SUMEX-AIM NSF Proposal Snag
C00688 00340 ∂15-Jun-79 1031 DEW arpanet loss loses
C00689 00341 ∂15-Jun-79 1118 DEK tip first, lab afterwards
C00690 00342 ∂15-Jun-79 1137 RPG Move
C00692 00343 ∂15-Jun-79 1139 PERLMAN at BBN-TENEXA mailing list
C00693 00344 ∂15-Jun-79 1138 RPG Previous message
C00694 00345 ∂15-Jun-79 1212 REM via AMES-TIP Moving TIP before SU-AI
C00697 00346 ∂15-Jun-79 1237 REM
C00699 00347 ∂15-Jun-79 1358 ES ARPA net
C00700 00348 ∂15-Jun-79 1412 PAM
C00701 00349 ∂15-Jun-79 1411 DBG TIP
C00702 00350 ∂15-Jun-79 1431 BPM ARPAnet access
C00704 00351 ∂15-Jun-79 1541 SMG via HARV-10 arpanet
C00705 00352 ∂15-Jun-79 1545 DCL ARPAnet availability
C00707 00353 ∂15-Jun-79 1617 ARR Inconvenienced by not having ARPA
C00711 00354 ∂15-Jun-79 2219 HPM TIP usage
C00712 00355 ∂15-Jun-79 2227 MRC TIP going away
C00714 00356 ∂16-Jun-79 0624 BIS Losing NET Connection
C00715 00357 ∂16-Jun-79 0806 BH via MIT-AI moving the TIP
C00716 00358 ∂16-Jun-79 0812 FB Tip move inconvenience
C00717 00359 ∂16-Jun-79 1259 WP Moving the TIP before the lab.
C00718 00360 ∂16-Jun-79 1334 DEA Loss of ARPA net
C00719 00361 ∂16-Jun-79 1423 RPG TIP Move and MacLisp maintainance
C00721 00362 ∂16-Jun-79 2111 DON
C00722 00363 ∂16-Jun-79 2113 DON
C00723 00364 ∂16-Jun-79 2308 DON dictionary
C00724 00365 ∂17-Jun-79 0937 MINSKY at MIT-AI (Marvin Minsky)
C00725 00366 ∂17-Jun-79 1023 BIS Manipulator safety in JAX Hall
C00726 00367 ∂18-Jun-79 1038 JKS tip move
C00727 00368 ∂18-Jun-79 1100 JMC*
C00728 00369 ∂18-Jun-79 1154 RSC LISP Verifier meeting
C00730 00370 ∂18-Jun-79 2054 MXB via SRI-KL account
C00731 00371 ∂18-Jun-79 2334 CLT NIHIL EX NIHIL
C00732 00372 ∂19-Jun-79 0329 DON via Easy Street
C00733 00373 ∂19-Jun-79 1344 Bmoore at SRI-KL (Bob Moore) Why don't we get together...
C00734 00374 ∂19-Jun-79 1408 Feigenbaum at SUMEX-AIM confusion on the 20 vs VAX front
C00738 00375 ∂19-Jun-79 1708 Bmoore at SRI-KL (Bob Moore) Terminals at CASBS
C00739 00376 ∂19-Jun-79 1728 PHW at MIT-AI (Patrick H. Winston) Ma Xiwen
C00740 00377 ∂19-Jun-79 2332 LLW SAIL Moving Date
C00744 00378 ∂19-Jun-79 2332 FB 10's and 20's versus VAXes
C00746 00379 ∂20-Jun-79 0503 Mark Crispin <Admin.MRC at SU-SCORE> Dialnet work
C00752 00380 ∂20-Jun-79 0554 PJH via LONDON
C00756 00381 ∂20-Jun-79 1206 MFB kl → m.j.hall
C00757 00382 ∂20-Jun-79 1219 PAT
C00758 00383 ∂20-Jun-79 1250 DEK KL move date
C00760 00384 ∂20-Jun-79 1327 BPM SAIL move
C00762 00385 ∂20-Jun-79 1336 RPG Move
C00763 00386 ∂20-Jun-79 1353 RAK Moving date
C00764 00387 ∂20-Jun-79 1619 LGC ARPA Access to SAIL
C00765 00388 ∂20-Jun-79 1812 Mark Crispin <Admin.MRC at SU-SCORE> SAIL Arpanet downtime
C00767 00389 ∂20-Jun-79 1827 Mark Crispin <Admin.MRC at SU-SCORE>
C00770 00390 ∂20-Jun-79 1918 HPM M.J.H. move
C00774 00391 ∂20-Jun-79 2037 BIS Best Regards
C00775 00392 ∂20-Jun-79 2144 RPG Solution
C00776 00393 ∂21-Jun-79 0824 Ijcai79 at SUMEX-AIM IJCAI Travel:Visas
C00777 00394 ∂21-Jun-79 0950 Bmoore at SRI-KL (Bob Moore)
C00778 00395 ∂21-Jun-79 0952 Dbrown at SUMEX-AIM (Response to message)
C00779 00396 ∂21-Jun-79 1000 CLT*
C00780 00397 ∂21-Jun-79 1207 ARK S-1 Meeting Notice
C00781 00398 ∂21-Jun-79 1247 PAT todorovich
C00782 00399 ∂21-Jun-79 1249 RPG
C00783 00400 ∂21-Jun-79 1253 PN Pail-5,Ptrans
C00784 00401 ∂22-Jun-79 1212 ARK S-1 meeting: location corrected
C00786 00402 ∂22-Jun-79 1327 LES Dastek visit
C00788 00403 ∂22-Jun-79 1330 HVA (1) Sunday, July 8th (2) Carolyn Talcott-Summenr Quarter
C00790 00404 ∂22-Jun-79 1458 CSD.JJF at SU-SCORE 1980 AI Qual
C00792 00405 ∂22-Jun-79 1628 MDD SAIL MEMO
C00793 00406 ∂22-Jun-79 1808 MDD future addresses
C00794 00407 ∂22-Jun-79 1904 REM Shannon's original paper
C00796 00408 ∂22-Jun-79 2232 Feigenbaum at SUMEX-AIM AI Qual for next year
C00797 00409 ∂23-Jun-79 1034 DEW lab moving date
C00798 00410 ∂23-Jun-79 1656 Rindfleisch at SUMEX-AIM DEC GIFT VAX - LICENSE ARRANGEMENTS
C00802 00411 ∂23-Jun-79 2325 Feigenbaum at SUMEX-AIM (Response to message)
C00803 00412 ∂23-Jun-79 2335 Feigenbaum at SUMEX-AIM a bit of jmc history
C00805 00413 ∂24-Jun-79 0905 FB Digital gift
C00808 00414 ∂24-Jun-79 1355 Feigenbaum at SUMEX-AIM VAX VLSI/CAD facility proposal
C00821 00415 ∂24-Jun-79 2104 CLT predpar substitution seems to have the catching bv
C00822 00416 ∂25-Jun-79 0421 MRC via Dialnet Dialnet TELNET
C00823 00417 ∂25-Jun-79 0742 PJH via LONDON dates and turner
C00826 00418 ∂25-Jun-79 1103 Moore at SRI-KL (J Moore) TAK over the Integers
C00830 00419 ∂25-Jun-79 1102 RWW bugs
C00831 00420 ∂25-Jun-79 1335 Bmoore at SRI-KL (Bob Moore) Hayes's idea about Ray Turner
C00834 00421 ∂25-Jun-79 1416 CLT equiv
C00835 00422 ∂26-Jun-79 0109 DLW at MIT-AI (Daniel L. Weinreb)
C00836 00423 ∂26-Jun-79 0137 LES
C00837 00424 ∂26-Jun-79 1336 RWW
C00842 00425 ∂26-Jun-79 1429 RWW bug
C00845 00426 ∂26-Jun-79 1608 RWW BUG
C00846 00427 ∂27-Jun-79 0927 DEW thesis
C00848 00428 ∂27-Jun-79 1022 DEW
C00849 00429 ∂27-Jun-79 1335 PN Proposal for Version markers in Pascal programs
C00851 00430 ∂27-Jun-79 1418 REM via SU-TIP
C00852 00431 ∂27-Jun-79 1445 DEK moving day
C00853 00432 ∂27-Jun-79 2238 RPG
C00854 00433 ∂27-Jun-79 2329 RPG
C00855 00434 ∂28-Jun-79 0435 PJH via LONDON martindavis
C00856 00435 ∂28-Jun-79 1558 Bmoore at SRI-KL (Bob Moore) Ascribing Mental Qualities to Machines
C00861 00436 ∂28-Jun-79 1721 PAT Edinburgh flights
C00863 00437 ∂28-Jun-79 1900 JMC*
C00864 00438 ∂29-Jun-79 0157 Feigenbaum at SUMEX-AIM AI Lab's design project budget
C00866 00439 ∂29-Jun-79 1023 Jrobinson at SRI-KL (Jane Robinson) Dinner tomorrow nnight
C00867 00440 ∂29-Jun-79 1420 ARK S-1 Meeting Notice
C00868 00441 ∂29-Jun-79 1645 Buchanan at SUMEX-AIM Request for help at IJCAI
C00870 00442 ∂29-Jun-79 1903 LCW SAIL Move
C00871 ENDMK
C⊗;
∂02-Apr-79 1108 Bmoore at SRI-KL (Bob Moore) letter to study group particpants
Date: 2 Apr 1979 1108-PST
From: Bmoore at SRI-KL (Bob Moore)
Subject: letter to study group particpants
To: jmc at SAIL
Have you finshed with it yet? Should we send it out? Also, how about adding
Barbara Partee to the list of potential invitees?
Bob
-------
The letter is ok to send without change, and I agree that Barbara Partee
would be a good potential invitee. Do you want to send it from SRI?
From here; Patte would do it for you? Or from CASBS?
∂02-Apr-79 1211 LES LLL temporary access to NS
For your information, a group at LLL which is involved in monitoring the
Pennsylvania situation requested access to News Service. I put them on
as "LLL" for the month of April.
∂02-Apr-79 1504 Feigenbaum at SUMEX-AIM LOTS visit for high level delegation from Israel
Date: 2 Apr 1979 1503-PST
From: Feigenbaum at SUMEX-AIM
Subject: LOTS visit for high level delegation from Israel
To: admin.gorin at SCORE
cc: jmc at SAIL
Ralph,
Pls hold open the time from 2:30pm to 5 or 5:30 for a visititation
from the Technion in Israel (Vice Chancellor for Academic Affairs,
plus Director of Computer Center, etc. ) whp have heard from others that
LOTS is sensational and wqnt to see it for themselves and talk to the
folks who made it.
Ed
p.s. I forgot to mention the date!!! It is May 7.
-------
∂02-Apr-79 1514 DCO
To: "@GROUP.DIS[1,DCO]" at SU-AI
PROGRAM VERIFICATION AND ANALYSIS SEMINAR
There will be no seminar this week, but next week...
PLACE: ERL 237
TIME: 2:30 p.m., March 10
TITLE: An Overview of "Scottery"
SPEAKER: Frieder von Henke, Stanford University
ABSTRACT:
This seminar is an introduction to Dana Scott's logic for reasoning
about partial recursive functions. The logic has been further developed and
implemented as the "Logic for Computable Functions " by R. Milner and others.
We will discuss the underlying theory (fixed point theory, "Scottery"), LCF, and
the denotational style of defining the semantics of programming languages (due
to Strachey and Scott).
The first seminar will be a tutorial and overview. Further seminars
will go into more detail on specific areas of interest to participants.
∂02-Apr-79 1604 Bmoore at SRI-KL (Bob Moore) Letter to study group particpants
Date: 2 Apr 1979 1604-PST
From: Bmoore at SRI-KL (Bob Moore)
Subject: Letter to study group particpants
To: jmc at SAIL
I don't think it matters very much where this letter is sent from. If it
is convenient for Patte to do it, that would be fine with me.
Bob
-------
OK, please send Patte your updated list.
∂02-Apr-79 1743 Mark Crispin <Admin.MRC at SU-SCORE> Dialnet
Date: 2 Apr 1979 1731-PST
From: Mark Crispin <Admin.MRC at SU-SCORE>
Subject: Dialnet
To: JMC at SU-AI
Sumex is going to try to set up an experiment to talk with us soon. They
have a cabinet full of VADIC 3400s they aren't using! They will connect
some up (or try to) as soon as Tom Rindfleisch gets back from vacation to
ok the idea.
They will have buffering problems since their implementation is a user mode
one; they only have 119 bytes of buffer space and so have to run a lot more
than we do in order to keep up with our 263 byte packets. With any luck
though we will be able to demonstrate that communication is possible.
-------
∂02-Apr-79 1819 MRC RFC753.TXT[NET,MRC]
To: JMC, LES
Please read this file. It is the Arpa bureaucrats official new idea
on how internetwork messages should look.
∂02-Apr-79 1833 Jrobinson at SRI-KL (Jane Robinson)
Date: 2 Apr 1979 1833-PST
From: Jrobinson at SRI-KL (Jane Robinson)
To: mccarthy at SU-AI
It was ok. The baked potato just went into the next day's soup. How
are you now, is Tuesday's event still on, can I help by
bringing e.g. a salad???And if so, at what time?
Jane
-------
No help needed. We expect you at 7.
∂03-Apr-79 0256 REM via AMES-TIP Data-compression programming
To: RWW, JMC
After giving up on Mainsail due to many bugs and no support of SU-AI version,
and hassling with Maclisp much, I managed to find a way to get Maclisp to
do storage allocation correctly. It works great at MIT-MC. But none of
the versions of Maclisp here work (all but one lack dynamic array-space
allocation, and the one that has dynamic array-space allocation has a bug
in array-space allocation and a bug in FILEPOS, the former I've been able
to program around but the latter is sorta essential), so I guess for the
time being development of the new SSORT routine (both as stand-alone and
as essential part of internal workings of data-compression stuff, as well
as a testbed for storage allocation in Maclisp) is halted here and can
only be done at MIT-MC. Boy am I ever pushing these "great" programming
languages to their limit and making them prove themselves or die trying!
∂03-Apr-79 0623 REM via AMES-TIP Pruning histogram in IRSM software/hardware
To: JMC, RWW
Search depth for histogram, once context is found, is often quite short.
Thus linear search used in IC2 wins. (Example, search depth average 2.5
when crunching nutrition database file, compression ratio 4.9 to 1.)
Now for the super conclusion. Since only a tiny part, the very most
common next-characters in each given context, are needed most of the time,
I think we can super-win the memory/time tradeoff by swapping the rest
of the histogram out to slower memory, or even by flushing it entirely
and just using an escape token for rare next-characters. This should
allow the given algorithm and contexts to fit in much less memory,
allowing many more contexts to be squeezed into the same memory, allowing
better compression and search depth reduction to less than 2.0, allowing
further pruning of histograms... Particular contexts that are never
selected because whenever they match a longer context also matches and
thus usurps each instance, can get pruned down to have nothin but the
escape node! Commonly-selected contexts can be expanded slightly, like
to five or so entries plus the escape, and of course the null-context
can be kept large (perhaps a hundred next-characters). Shrinking useless
contexts to escape-only can partly offset the lack of context-tree
optimization that happens with adaptive algorithm.
Now as soon as I finish getting Maclisp to do what I want, and finish
fully testing it to be sure, maybe I'll start programming some of these
new ideas, advancements upon IC2.FAI.
∂03-Apr-79 0825 Ijcai79 at SUMEX-AIM IJCAI REviews
Date: 3 Apr 1979 0823-PST
From: Ijcai79 at SUMEX-AIM
Subject: IJCAI REviews
To: mccarthy at SAIL, newell at CMU-10A, Norman at PARC-MAXC
WE NEED YOUR IJCAI REVIEWS!
If it more convenient for you, you may return your
comments in a msg. Send to IJCAI79@SUMEX-AIM.
Please return your reviews this week.
Thank you -- The IJCAI79 Program Committee
-------
∂03-Apr-79 0949 Bmoore at SRI-KL (Bob Moore) letter to CASBS study group participants
Date: 3 Apr 1979 0949-PST
From: Bmoore at SRI-KL (Bob Moore)
Subject: letter to CASBS study group participants
To: pat at SAIL
cc: jmc at SAIL
Here is the text of the letter that John and I want to send to the other
members of the AI/Phil study group, with the current list of suggested
workshop participants at the end. I am sending through the ID mail the
list of addresses for the study group members that I got from the Center
in case you need it. I believe that Dan Dennet and Zenon Pylyshyn are
not currently at their home instituions as indicated on the list, however.
I know that Pylyshyn is spending the year at MIT, and I have heard that
Dennet is spending the year at Oxford. You might want to check whether
the Center has more current addresses for them.
Bob
------------------
Dear XXXXX:
In connection with the forthcoming study of AI and
Philosophy at the Behavioral Sciences Center, we believe
that it would be desirable to have a small conference or
workshop on this topic that would involve a larger group of
scholars than just those in the study group. We would like
to solicit your reactions to this suggestion and ideas on
how it should be organized. A two day meeting (probably a
Friday and Saturday) sometime in October seemed most
appropriate to us. We suggest having invited papers
distributed in sufficient time for the workshop participants
to have read them, but also to select one participant to
serve as official commentator for each paper. The format of
the workshop would then be to have a brief presentation of a
paper, followed by the remarks of the commentator, the
response of the author, and general discussion. This format
would allow for three or four papers presented per day, or a
total of six or eight for the workshop. Since this is to be
an interdisciplinary workshop, it seems desirable that the
authors and commentators be from different disciplines.
We envision officially inviting about twenty people from
outside the study group, with an informal invitation being
extended to the local Philosophy and AI communities. We
have attached a list of prospective participants. We would
like to hear your general reaction to this idea, plus any
ideas you have regarding the date or format of the workshop,
specific topics to be discussed, and potential participants,
speakers, or commentators. At the moment all ideas are
quite tentative, so please do not hesitate to recommend
changes to our suggestions.
Sincerely,
John McCarthy
Robert Moore
SUGGESTED PARTICIPANTS FOR WORKSHOP ON AI AND PHILOSOPHY
ARTIFICIAL INTELLIGENCE
David Marr MIT
Mitch Marcus MIT
Jon Doyle MIT
Brian Smith MIT
Terry Winograd Stanford
Richard Weyhrauch Stanford
Lewis Creary Stanford
Barbara Grosz SRI
Jerry Hobbs SRI
Jane Robinson SRI
Herbert Simon CMU
Bill Woods BBN
Drew McDermott Yale
PHILOSOPHY
Patrick Suppes Stanford
Julius Moravcsik Stanford
John Perry Stanford
Noam Chomsky MIT
Jerry Fodor MIT
Ned Block MIT
Hubert Dreyfus Berkeley
David Kaplan UCLA
Hilary Putnam Harvard
Aaron Sloman Sussex
Barbara Partee UMass
-------
∂03-Apr-79 1128 Feigenbaum at SUMEX-AIM good news
Date: 3 Apr 1979 1126-PST
From: Feigenbaum at SUMEX-AIM
Subject: good news
To: jmc at SAIL
John,
I made a strong case last month for a big salary increase for you to correct
the present unsatisfactory situation, and learned today that the case was
successful.
Ed
-------
∂03-Apr-79 1436 Mcgoveran at SUMEX-AIM Dean's Office message
Date: 3 Apr 1979 1412-PST
From: Mcgoveran at SUMEX-AIM
Subject: Dean's Office message
To: TAR at SAIL, JMC at SAIL
Julia Johnson called me and asked that I pass on the following message:
Please call her at 72275 to arrange a meeting with the dean to chat.
Bob Tarjan had a meeting for April 2, and John McCarthy had a meeting
set for April 3.
Mary
-------
∂03-Apr-79 1454 Gaschnig at SRI-KL Ph.D. student looking for job
Date: 3 Apr 1979 1453-PST
From: Gaschnig at SRI-KL
Subject: Ph.D. student looking for job
To: buchanan at SUMEX-AIM, feigenbaum at SUMEX-AIM,
To: mccarthy at SU-AI
Mail from USC-ISIC rcvd at 30-Mar-79 1559-PST
Date: 30 Mar 1979 1559-PST
From: GERHART at USC-ISIC
Subject: a good student
To: gaschnig at SRI, simon at CMU-10A
cc: gerhart
Lee Erman referred me to you for the following problem:
There's a very bright Ph.D. student at Duke named Dana Nau, who's
looking for a job and not finding the response we would expect in
today's market. He's exceptionally strong mathematically and his
thesis contains results on tree searching. He did a classy master's
thesis in medical applications and is good on all the formal sides of
computer science. He is knowledgeable about AI, but not strictly AI.
He won't deal with problems that don't have sufficiently high and
challenging mathematical content. With Duke's unestablished
reputation for high-quality Ph.D. students and his undergraduate
origins from Missouri, (not to mention his eclectic interests) this
student may be over-looked because he doesn't fit the mold. If you
know of any positions or have any advice to pass on, I would
appreciate it. His advisor at Duke is Alan Biermann.
-------
---------------------------
I don't know this fellow, but I thought I'd pass the info on to you
in case you're interested.
Cheers,
John
-------
∂03-Apr-79 1528 LGC via AMES-TIP Leadoff Colloquium Slot
Have you any interest in being the speaker at a first meeting of the SAIL
Colloquium next Monday afternoon (9 April)?
It's possible, but I'd prefer you got someone else.
∂04-Apr-79 1642 BS Ralph Gorin's Salary Review Date
To: Feigenbaum at SUMEX-AIM, JMC at SU-AI
CC: BS at SU-AI
There was some confusion in the Provost's Office about Ralph's salary review
date. All is o.k. now, and everyone is agreed that it will be 2/1/80 (not
3/1/80, as I mentioned to you this morning, Ed).
For budget planning purposes, Pat Devaney would like to know the percentage
of increase you probably will recommend for Ralph next February. Is 8%
a realistic guess?
Betty
∂05-Apr-79 0857 Feigenbaum at SUMEX-AIM Re: Ralph Gorin's Salary Review Date
Date: 5 Apr 1979 0846-PST
From: Feigenbaum at SUMEX-AIM
Subject: Re: Ralph Gorin's Salary Review Date
To: BS at SU-AI, JMC at SU-AI
In response to the message sent 4 Apr 1979 1646-PST from BS at SU-AI
Given the intense pressure on salaries for such people, a better guess
for Ralph's next year's salary increase would be 10% or more.
Ed
-------
∂05-Apr-79 1309 SL Meet with you?
John - Whereas you are close enough to the bull's eye to figure centrally in
the matter of my transfer to full time in the CSD, you seem to have had a
masterfully successful mechanism for minimizing your involvement in the
process. Given that it is now agreed that I will make the move, I would
like the chance to sit down and exchange a few words of "hello" with you.
Would you like to suggest a time. Presently, any time April 16, 18, or 19
is o.k. with me. - Sid
∂05-Apr-79 1725 TW
We are in the last stages of preparing the AI Qual syllabus for this
spring. The entire file is on line as SYL.PUB[1,TW] at SAIL and
<WINOGRAD>SYL.PUB at SUMEX. We would like your comments on any part
of it. However, there are specific areas which we think you are most
likely to want to comment on (and in some cases which we have not
yet filled in). I am enclosing in this message the general description
of the topic list and the full list of topics. Those entries
which seem closest to your interest are marked with a star. Please send me
any changes as soon as possible. I will send something out to the
students within a week, whether it reflects your views or not.
-------------------------
The syllabus this year is an updated version of the one prepared in 1975.
Like that one, it is organized to present a picture of the range of
knowledge expected of Ph.D. candidates in Artificial Intelligence, rather
than specifying a fixed list of readings. There are a number of different
dimensions along which we could divide up the material. The attempt in
the earlier version to establish a thorough categorization has been
replaced this year with a less formal, more realistic organization. We
have listed a number of "topics" with a short paragraph describing the
necessary reading for each. These topics overlap in various ways, and
reflect idiosyncratic views of how things divide up without any attempt to
provide a consistent classification. Hopefully the set of references
included with each item will make it possible for students to select a
reasonable number of readings which will fill any knowledge gaps. The
long reading list is intended as a source of details on individual
references, not as a necessary set of things to read. It includes a rough
indication of what sort of understanding is most important for each
reference -- whether there is a general perspective, one or more specific
concepts, and/or a body of detail with which students are expected to be
fluent. These indications are of course based on the prejudices and
peculiarities of the committee making up the syllabus, and should not be
taken as representing the views of anyone else (including the members of
individual exam committees). Use your judgement.
-------------------------------
.head |Topics to be studied|
As mentioned above, this is not intended as a complete or structured
classification. It is a list of answers to the vague question "What
kinds of things should AI students know about?". There is no significance
to the ordering.
* General Perspective
Weak Methods
* Application of Formal Logic to AI
Knowledge Engineering - Expert Systems
* Knowledge Representation Formalisms
* Game Playing
* Planning and Common-sense Reasoning
Mathematical theorem proving and discovery
Natural Language
Speech understanding
Vision
Physical manipulation
* Automatic Programming and Program Verification
Learning and Inductive Inference
Psychological Models
Automata and Formal Language Theory
* Programming Languages for AI
* Philosophical Implications
* Political and Social Implications
* History and politics of the field
.subhead |General Perspective|
There are several recent books on AI which attempt to provide an
overview. Of these, [Boden AI] and [Winston AI] are the best
starting point. Shorter articles which provide long-term perspective
are [Minsky STEPS], [Feigenbaum IFIP], and [Nilsson OVERVIEW].
In addition, the AI handbook will provide an overview of lots
of AI issues, and copies of some parts are available.
.subhead |Weak Methods|
Classicaly, AI has been associated with a set of methods for
problem solving and search which have been called "weak methods".
These include early work such as GPS [Newell, Shaw, and Simon],
notions of heuristic search as described in [Nilsson
AI] and [Handbook SEARCH], and more theoretical ideas abot
problem spaces discussed in [Newell ILL] and at length in
[Newell & Simon HPS]. Students are expected to be familiar enough
with the technical details to demonstrate how the techniques
operate, but will not be asked to prove theorems or remember complex
results.
.subhead |Application of Formal Logic to AI|
Much work in AI has made use of formal logic, and in particular of
the first order predicate calculus. Some of the issues of how
logic can be applied are discussed in [McCarthy & Hayes] and in
[Hayes DEFENCE]. Many of the current issues are discussed in
[McCarthy 5IJCAI]. The relevance of theorem proving to problem
solving is discussed in [Green TP], and the techniques of
resolution theorem proving are described in [Nilsson AI]. As
with weak methods, students are expected to be familiar with
the basic mechanisms (e.g. be able to demonstrate a simple proof
by resolution, or explain the issues in unification) but
will not be required to know sophisticated technical results
(e.g. prove the completeness of resolution with the X heuristic).
A tutorial on some of the relevant mathematics is in [Manna MTC].
.subhead |Knowledge Engineering - Expert Systems|
Much of current AI work is being subsumed under the heading of "knowledge
engineering". [Bernstein KBS] is a general survey of knowledge based
systems. [Winston AI] gives a general idea of systems of this kind done
at MIT, and [Feigenbaum IJCAI5] describes the general approach. A number
of expert systems have been built in the past few years. Students should
be familiar with the general capabilities and design. Areas include:
Algebraic Manipulation [Martin and Fateman MACSYMA]; Chemistry [Feigenbaum
et. al. DENDRAL]; Medicine [Shortliffe MYCIN], [Pople ??]; Molecular
Genetics [??]; and Mineral Exploration [Duda et al PROSPECTOR].
.subhead |Knowledge Representation Formalisms|
A number of current AI research projects are centered around the
development of knowledge representation languages. Some of the early
issues in representation are discussed in [Amarel ACTIONS] and [Bobrow
DIMENSIONS]. More recent discussions include [Winograd FRAME], [Hayes
DEFENCE], and [Winograd EXTENDED]. Students should be familiar with at
least the following general approaches: Procedural Embedding [Hewitt
APPRENTICE]; Semantic Networks [Quillian SEMANTIC MEMORY], [Woods LINK],
[Hendrix, PARTITION], [Brachman CONCEPT]; Conceptual Dependency [Schank in
S&C]; Frames, Scripts, etc. [Minsky FRAMES], [Schank and Abelson SCRIPTS];
Production Systems [Davis & King OVERVIEW]; Description Languages [Bobrow
& Winograd KRL]
.subhead |Game Playing|
One of the earliest and most publicized areas of AI research has been game
playing programs, such as those for checkers [Samuel in C&T] and Chess
[Greenblatt FJCC]. Students should be familiar with the basic techniques
(e.g. Minimax and alpha-beta) and some of the more subtle problems (e.g.
the horizon effect [Berliner IJCAI3] and the use of patterns [Simon
1973]). Basic techniques are taught in [Nilsson AI] and [Slagle AI].
.subhead |Planning and Common-sense Reasoning|
Much of the AI work related to robotics dealt with the planning of action
sequences. GPS [Newell, Shaw and Simon] was the early classic, and other
well known systems are STRIPS [Fikes, Hart, Nilsson] and NOAH [Sacerdoti
NONLINEAR]. An early discussion of the problems is in [McCarthy
ADVICE-TAKER].
.subhead |Mathematical theorem proving and discovery|
Mathematics has always been an important domain for AI. The Geometry
theorem prover [Gelernter] was an early program. More recent efforts in
mathematical theorem proving are discussed in [Bledsoe MAN-MACHINE]. The
mechanization of mathematical discovery is discussed in [Lenat IJCAI5].
Important background is provided by the discussions of mathematics in
[Polya <all?>]
.subhead |Natural Language|
Much of the research on natural language is summarized in the handbook
articles [Handbook NL] and the drafts of [Winograd LANGUAGE]. Students
should be aware of the general content of these and familiar with the
issues which arise in parsing and reasoning with natural language. This
includes a level of understanding of parsing techniques (CFG, ATN, TG,
etc.) similar to that of heuristic search and theorem proving discussed in
sections above. Early work in NL is described in [Simmons SURVEY 1965]
[Simmons SURVEY 1970], [Minsky SIP (browse)], and [Weizenbaum ELIZA].
[Handbook NL] gives a summary of work in Machine Translation. Many papers
reflecting current research interests are found in [Schank and Nash-Webber
TINLAP-1] and [Waltz TINLAP-2].
.subhead |Speech understanding|
Work in Speech systems is well summarized in the handbook,
and in [Reddy in IEEE], [Woods IEEE], [Walker IJCAI5], [Newell SPEECH],
[Erman & Lesser IJCAI4] and [Erman & Lesser IJCAI5].
.subhead |Vision|
<this paragraph needs to be written by someone who knows more than I
do. Here are the entries from the old one as updated by Jim>
.begin nofill
General [Winston AI (parts of chs 3, 4, 8)]
Pattern Recognition [Selfridge and Neisser in C&T] [Hunt AI]
Perceptrons [Hunt AI]
Representation [Marr VISUAL]
Signal Processing
Scene Analysis [Agin and Binford CURVED] [Yakimovsky & Feldman AIJ]
[Waltz in MIT74]
.end
.subhead |Physical manipulation|
<this paragraph needs to be written by someone who knows more than I
do. Here are the entries from the old one as updated by Jim>
.begin nofill
Hands [Finkel et. al. AL] [Stanford Robotics films]
Mobility
.end
.subhead |Automatic Programming and Program Verification]
<this paragraph needs to be written by someone who knows more than I
do. Here are the entries from the old one as updated by Jim>
.begin nofill
[Buchanan and Luckham AUTOMATING] [Sussman and Goldstein in MIT74]
[Waldinger&Manna] [Floyd INTERACTIVE] [Green IJCAI5] [Barstow IJCAI5]
.end
.subhead |Learning and Inductive Inference|
<this paragraph needs to be written by someone who knows more than I
do. Here are the entries from the old one as updated by Jim>
.begin nofill
[Sussman in MIT74] [Goldstein in MIT74] [Buchanan et.al.
THEORY-FORMATION] [Winston in MIT74] [Fikes et.al.GENERALIZED]
[Smith et al IJCAI5]
.end
.subhead |Psychological Models|
Many AI programs have been intended as models of human information
processing. Information processing psychology in general is described in
[Newell and Simon HPS (Sections 1 and 5 especially)]. An early program
which modelled human verbal behavior was [Feigenbaum EPAM]. More recent
models are found in [Anderson & Bower HAM (chapters 4 and 7)], [Anderson
??], [Norman & Rumelhart EXPLORATIONS], and [Collins & Quillian USER]. A
famous early program which took one form of intelligence tests was [Evans
ANALOGY], and a controversial model of paranoia is described in [Colby
SIMULATIONS]. Relevant classics from psychology include [Bartlett
REMEMBERING] and [Miller MAGICAL].
.subhead |Automata and Formal Language Theory|
Although this is not AI per se, it forms an important part of the
background. Relevant work is in [Minsky COMPUTATION], [Manna MTC] and the
material on perceptrons in [Hunt AI] (extra detail in [Minsky & Papert
PERCEPTRONS]).
.subhead |Programming Languages for AI|
.begin nofill;skip
List Processing -- LISP
String processing -- SNOBOL
Associative mechanisms -- LEAP/SAIL
Active data structures -- SIMULA/SMALLTALK/ACTORS
Pattern Matching [Bobrow & Raphael]
Data Structures [Knuth Vol.I]
PLANNER, CONNIVER, QA4, etc. [Bobrow & Raphael]
Production Systems [Davis & King OVERVIEW]
.end
The candidate is expected to be familiar enough with some AI language (e.g.
LISP or SAIL) to demonstrate the ability to write simple programs. He or she
should also know enough about the features of the more specialized languages
(CONNIVER, QA4, the LEAP and multiple process features in SAIL) to discuss
the kinds of problems for which they are useful, and the limitations they
force the programmer into. It is not necessary
to know syntactic details of these features.
.subhead |Philosophical Implications|
There has been a continuing discussion about the philosophical implications
of intelligent machines. The original paper often cited is
[Turing TEST]. The best known book criticizing AI from a philsophical
point of view is [Dreyfus CAN'T]. A response to an earlier version
of the same material is in [Papert DREYFUS]. Many of the general
issues are discussed in [The Rock Test, others in Anderson MINDS] and
in [Dennett BRAINSTORMS]. There is an accessible discussion of the issue in
[Boden AI]. A recent controversy dealing with AI and the philosophy
of science is in [Dresher & Hornstein SUPPOSED], [Winograd CONTESTED],
and [Dresher and Hornstein RESPONSE].
.subhead |Political and Social Implications]
One of the pioneers in the field discussed these issues in [Weiner HUMAN],
and more recently there has been a discussion raised by [Weizenbaum
REASON], and responses like [McCarthy ??]. [Boden AI] also discusses some
of the important questions. [Firschein & Coles IJCAI3 survey] attempts to
predict some of the future applications of AI. An important controversy
in British support for AI is in [Lighthill AI] and responses in the same
booklet and in [McCarthy LIGHTHILL].
.subhead |History and politics of the field|
.begin fill
A critical part of being able to do and evaluate research in a field
is having some perspective on what things have been done and why --
in particular, what lessons have been learned and how can they
be applied to keep from repeating mistakes. In addition, students
need to know where and how things are published in the field if
they hope to keep up with current work. It is difficult to give specific
references, but the general books on AI [e.g. Boden, Winston, Jackson,
Raphael] each give some perspective. It is also useful to look at
some of the early collections of papers, such as [Feigenbaum & Feldman].
In particular, students should have some familiarity with the early
history of AI -- its connections to cybernetics and machine translation in
the 50's. It is interesting to note some of the early optimistic attempts
at "self-organizing systems" and work on perceptrons and neural nets.
Today, there is much discussion about the emergence of a discipline called
"cognitive science" which includes work now considered artificial
intelligence, psychology, linguistics, and philosophy. Students should
have done some thinking about the relationships between these disciplines,
especially where they adopt differing methodologies in looking at the same
phenomena.
The following journals contain material relevant to AI, and it is useful
to have a general idea of what kinds of things each of them contains.
.begin nofill
Journal of AI
SIGART (ACM special interest group on Artificial Intelligence)
SIGCAS (ACM special interest group on Computers and Society)
Machine Intelligence (1 - ?)
IJCAI proceedings (1 - 5)
TINLAP proceedings (1 - 2)
CACM (computer science, some AI)
JACM (computer science, some AI)
Cognitive Psychology (some AI)
American Journal of Computational Linguistics (some AI)
Cognitive Science
The Behavioral & Brain Sciences
Special interest conferences: cybernetics, natural language, robotics
.end
∂05-Apr-79 1737 ARK S-1 Meeting Times
To: S1 at SU-AI
S-1 meetings are tentatively scheduled for alternate Wednesdays at 2:30 pm
in the Serra House Conference room. If this time is inconvenient, send me
a message indicating what times are acceptable. I expect the first meeting
to be on Wednesday, April 18, but another announcement will be sent out when
this has been made definite.
Arthur
∂06-Apr-79 1656 CLT FOLISP
To: JMC, RWW, LCA, SEK, CLT
The file FOLISP.DOC[FOL,CLT] contains
(1) list of files used to build the current FOLISP
(2) pointers to summary files
(3) explanation of how to use FOLISP to do proofs
(4) list of exercises
∂06-Apr-79 1716 ME
You have exceeded your disk quota.
The files listed below have been purged to reduce your disk
area to your quota of 2000
Before purging, your files occupied 2470
BACKUP.TMP[S79,JMC]
KNOW.LST[S78,JMC]
DIALNE.LST[DIA,JMC]
CRYPT.DMP[ 2,JMC]
LFOL.DMP[F78,JMC]
REV1.DMP[F78,JMC]
CODE.DMP[ 2,JMC]
MENTAL.XGP[F76,JMC]
CHEMEM.XGP[W79,JMC]
CONCEP.XGP[E76,JMC]
FIRST.XGP[W79,JMC]
ELEPHA.XGP[S79,JMC]
ELEPHA.XGP[LET,JMC]
PUTNAM.XGP[LET,JMC]
CRYPT.FAI[ 2,JMC]
CHRI.LE1[ 2,JMC]
VIS[ 2,JMC]
VIS1[ 2,JMC]
VITASK[ 2,JMC]
MIDF76[ 2,JMC]
PARC.2[ 2,JMC]
EC[ 2,JMC]
BDAY.LIT[LIT,JMC]
NOTES[LOT,JMC]
SPACE[LOT,JMC]
LENG.LE1[LOT,JMC]
BLURB[LOT,JMC]
DEC20[LOT,JMC]
TASK[LOT,JMC]
P[LOT,JMC]
TERMIN[LOT,JMC]
BUDGET.2[LOT,JMC]
FRANKL.LE1[LOT,JMC]
LOTS.BLB[LOT,JMC]
MEASUR[LOT,JMC]
TERMIN.2[LOT,JMC]
BOARD[LOT,JMC]
LOTS.REP[LOT,JMC]
SHAW.LE1[LOT,JMC]
HOTER.BAY[PUB,JMC]
WEIZEN.REP[PUB,JMC]
HOTER.REC[PUB,JMC]
HOTER.OUT[PUB,JMC]
WEIZEN.LED[PUB,JMC]
HOTER.POX[PUB,JMC]
HOTER.WD[PUB,JMC]
WEIZEN.REV[PUB,JMC]
HOTER.BGB[PUB,JMC]
WEIZEN.ANS[PUB,JMC]
SUBSCR[PUB,JMC]
GLUGG2[LIT,JMC]
CARGO.LIT[LIT,JMC]
IBM.SNG[LIT,JMC]
BIB.PUB[SEN,JMC]
TECHNO[SEN,JMC]
CALIF[SEN,JMC]
FEB14.NOT[SEN,JMC]
QUOTE.2[SEN,JMC]
STATEM.LET[SEN,JMC]
POSTER[SEN,JMC]
ASHLEY.COV[SEN,JMC]
SIGN[SEN,JMC]
ENDORS[SEN,JMC]
LONG[SEN,JMC]
LEGISL[SEN,JMC]
FEB8.78[SEN,JMC]
ENERGY.ST1[SEN,JMC]
MORTEM[SEN,JMC]
SUNDES[SEN,JMC]
MEET.F78[SEN,JMC]
GOSPER[SEN,JMC]
NOON15[SEN,JMC]
CALVO.L1[SEN,JMC]
TECHNO.NOT[SEN,JMC]
CGOLMA[W77,JMC]
SABBAT.LIT[LIT,JMC]
CHRIST[LIT,JMC]
CONFIG[LOT,JMC]
VIETNA.NS[S78,JMC]
ALGOL.SNG[LIT,JMC]
TIMES.LE5[LET,JMC]
FILMAN.RE3[LET,JMC]
LUCKHA.RE1[LET,JMC]
FEICHT.LE2[LET,JMC]
F77.IN[LET,JMC]
∂06-Apr-79 1721 ME
To: S1 at SU-AI
You have exceeded your disk quota.
The files listed below have been purged to reduce your disk
area to your quota of 7000
Before purging, your files occupied 9350
030LNK.TMP[PAS, S1]
OUT.TMP[FOR, S1]
PPSAV.TMP[PCP, S1]
TRIAL.TMP[FOR, S1]
INDEX.TMP[SMA, S1]
INDEX2.TMP[SMA, S1]
POA.LST[MIC, S1]
DSKPOA.LST[MIC, S1]
EOA.LST[MIC, S1]
IOA.LST[MIC, S1]
DOA.LST[MIC, S1]
F.LST[MIC, S1]
DSKP.LST[MIC, S1]
E.LST[MIC, S1]
I.LST[MIC, S1]
P.LST[MIC, S1]
D.LST[MIC, S1]
FINE.LST[FIN, S1]
ADN.LST[AMB, S1]
SAIL11.LST[ 11, S1]
T6.LST[FOR, S1]
TRIAL.LST[FOR, S1]
AGOTO.LST[FOR, S1]
OPS.REL[OPS, S1]
BOOTST.REL[PAS, S1]
UCPASC.REL[PAS, S1]
TEST.REL[PAS, S1]
RFINE.REL[FIN, S1]
REDISP.REL[FIN, S1]
CALIAD.REL[FIN, S1]
QREPL.REL[FIN, S1]
TTYCOR.REL[FIN, S1]
AUX.REL[FIN, S1]
CURSOR.REL[FIN, S1]
FASM.REL[PRG, S1]
FILESC.REL[FIN, S1]
FINE.REL[FIN, S1]
AMODUL.REL[DOC, S1]
PASLIB.REL[PAS, S1]
D.BIN[MIC, S1]
SAIL11.BIN[ 11, S1]
MICRO.DMP[ SW, S1]
PLOT.DMP[ SW, S1]
RIX.DMP[DMP, S1]
GLOBAL.DMP[MIC, S1]
EXR.DMP[OLD, S1]
PCP.DMP[MEM, S1]
PCPW4.DMP[DMP, S1]
LINK.DMP[DMP, S1]
SOPAW5.DMP[PAS, S1]
PCPW5.DMP[DMP, S1]
SOPAE1.DMP[DMP, S1]
SOPAW5.DMP[DMP, S1]
FSIM.DMP[OLD, S1]
FASM.DMP[OLD, S1]
FASM.DMP[PAS, S1]
FSIM.DMP[PAS, S1]
11LOAD.DMP[ 11, S1]
PCPASX.DMP[DMP, S1]
PC.DMP[MEM, S1]
RFINE.DMP[FIN, S1]
SOPS09.DMP[DMP, S1]
FASM.DMP[PRG, S1]
SOPAW6.DMP[DMP, S1]
XGPLOT.DMP[DMP, S1]
PLOT.DMP[DMP, S1]
PCPW5.DMP[PAS, S1]
SOPAW4.DMP[DMP, S1]
PCPE03.DMP[DMP, S1]
SOX.DMP[DMP, S1]
SOPEX1.DMP[DMP, S1]
MICRO.DMP[MIC, S1]
PCPE04.DMP[DMP, S1]
DIAL.DMP[ 11, S1]
SOPU14.DMP[DMP, S1]
PCFOR0.DMP[DMP, S1]
SOPAF6.DMP[FOR, S1]
PASCAL.DMP[PAS, S1]
PCROSS.DMP[PAS, S1]
UCFOR1.DMP[DMP, S1]
FINE.DMP[FIN, S1]
AMODUL.XGP[DOC, S1]
FCDS.XGP[DOC, S1]
ETIME[MK1, S1]
IEINTR.DOC[MK1, S1]
FDRW.NUM[MK1, S1]
EDRW.NUM[MK1, S1]
S1PDRW.NUM[MK1, S1]
IDRW.NUM[MK1, S1]
DELETE.D[DOC, S1]
PREF.D1[DOC, S1]
D2.PUB[DOC, S1]
D3.PUB[DOC, S1]
TITLE.D3[DOC, S1]
DSAPP.D3[DOC, S1]
OP.2[DOC, S1]
DOC.3[DOC, S1]
DSYS1.PUB[DOC, S1]
DC2.WD[MK2, S1]
DC3.WD[MK2, S1]
DC4.WD[MK2, S1]
DCAM.WD[MK2, S1]
TITLE.DS1[DOC, S1]
DS2.PUB[DOC, S1]
TITLE.DS2[DOC, S1]
MFLOW1.DRW[MEM, S1]
CU1.TXT[MEM, S1]
BOARD.DRW[MEM, S1]
CU2.TXT[MEM, S1]
CU3.TXT[MEM, S1]
SUMM.D1[DOC, S1]
ABTRCT.DS2[DOC, S1]
VECCOD[DOC, S1]
TITLE.CS1[DOC, S1]
TITLE.CS2[DOC, S1]
SWSLD.PUB[DOC, S1]
CIT.ABS[DOC, S1]
EXPROG.D1[DOC, S1]
OS0SIM.0[DOC, S1]
SPEC11.0[DOC, S1]
RQR.D1[DOC, S1]
PARCH.D1[DOC, S1]
EBOXT[DOC, S1]
DRW.D3[DOC, S1]
MUX25.DRW[ SW, S1]
PRCRC1.DRW[ SW, S1]
13B16M.DRW[ SW, S1]
ARB2.DRW[ SW, S1]
MICRO.FAI[ SW, S1]
LAYOUT.TXT[ SW, S1]
BLANK.DRW[ SW, S1]
MUXREG.DRW[ SW, S1]
ARB1.DRW[ SW, S1]
PRCCTL.DRW[ SW, S1]
MUXMU1.DRW[ SW, S1]
MUXMU2.DRW[ SW, S1]
1F16.DRW[ SW, S1]
CYCLE.TXT[ SW, S1]
PARTS.DIS[ SW, S1]
ORDER[ SW, S1]
PRIOR.DRW[ SW, S1]
LCW.MSG[ SW, S1]
PRCRC2.DRW[ SW, S1]
BLOCK.DRW[ SW, S1]
ARB3.DRW[ SW, S1]
OPS.FOO[OPS, S1]
MCLOCK.DRW[ SW, S1]
PLT.CMD[ SW, S1]
NOTES[ SW, S1]
SW1.DRW[ SW, S1]
SW2.DRW[ SW, S1]
ECLM1.DRW[ SW, S1]
ECLM2.DRW[ SW, S1]
ECLF1.DRW[ SW, S1]
ECLMEM.DRW[ SW, S1]
ECLM3.DRW[ SW, S1]
USER.S1[SYS, S1]
MATHDR.RQR[DOC, S1]
BODLP1.827[DOC, S1]
FIXF.PUB[FOR, S1]
FONTS.PUB[FOR, S1]
FASM.FAI[OLD, S1]
∂07-Apr-79 2242 RWG
while at louie's, i figured out the key property of the randomizer fcn
inpollard's method: let p be a (presumably unknown) factor of the number you're
try to factor, n. then you want x ≡ y mod p to imply f(x)≡f(y), so that if x
and y are i iterations apart, x-y ≡ f(x)-f(y) ≡ f(f(x))-f(f(y)) ≡ ... ≡ 0 mod p,
so you will find a multiple of p no matter when you look back, as long as it is
i iterations back. i can't offhand think of any such but rational fcns mod n.
∂08-Apr-79 0455 DON via Easy Street job slots
To: ME, JMC
I agree with your suggestion that two-segment jobs be made into single
segments where possible, but I point out that it's not all winnage.
Every such program, especially SAIL programs, will take up more disk
space due to duplication of the library routines. I'll modify DIRED
and FIND, as commonly used programs, to be single-segment, but less
commonly used SAIL programs (like PACK) I figure to leave as is.
∂08-Apr-79 0538 DON antidisestablishmentarianism
∂07-Apr-79 1523 JMC algorithm for find
I am surprised that FIND took 15 seconds to not find "antidisestablishmentarianism".
I would have thought that very long words would take less time than
the usual 10 seconds.
--------------------
Well, here I was all prepared to explain why long words would take longer
than short ones to set up the search, and no longer (or shorter) to perform
the actual search, but I couldn't imagine that it took THAT much longer, so
I tried it myself. As you can see below, the time taken was 6.605 ms of
processor time. Searching for a short word usually takes about 6.5 ms, so
this is not unreasonable. I suspect the extra 5 seconds you reported were
"wholine" time -- the compute time reported in the wholine is much more
variable, being dependent on system load and other fluctuating factors.
.df antidisestablishmentarianism
0 hits on key "antidisestablishmentarianism".
Exit
↑C
.ti
Total,Increment,Wait,PS,EBOX msec,Incr=
5:21'04, 0:09'00, 17:16'52,48900,151073,6605
.r ppsav
∂08-Apr-79 1523 RPG New Official Policy
To: ME, JMC
CC: LES, DCL
As you know, there exist several 1 segment versions of MacLisp at the
Lab: one which puts the logical high segment just above the low segment, growing
above that until 256k, and another which puts the logical high segment at the top
of the address space, managing the intermediate memory itself.
Neither of these two versions have been well tested, and it was
anticipated that there was code that might not work under the first of these two
options. Thus, I would, offhand, tend to rely on the 256k version if forced to
use a one segment MacLisp.
However, the real problem lies with reliability in another dimension.
Namely, as you know, I will be working on the S1 LISP this summer, and apparently
will be under their employ for the regular part of the year as well. At this
time it isn't clear to what extent I will be maintaining MacLisp while not
under direct Lab support, except insofar as any bugs pertain to my own thesis
work, and that support would be probably only for the two segment version. In
addition, I am strongly disinclined at this point to accept any further
responsibility for untested programs. So, if there was a push towards
legislating that the one segment versions become standard, I would like to point
out that potential users of these versions are doing so with the likelihood of
minimal support until June, and no support thereafter.
-rpg-
∂08-Apr-79 1636 Lederberg at SUMEX-AIM V. Turchin
Date: 8 Apr 1979 1527-PST
From: Lederberg at SUMEX-AIM
Subject: V. Turchin
To: FEIGENBAUM
cc: mccarthy at SAIL
I'll be sending you a cv and bibliog. shortly on Valentin Turchin. He got
out ovf the USSR recently in spite of having taken a very active and
courageous role in re implementation of the Helsinki treaty / and Amnesty
International. He has a temporary situation at Courant Institute, needs a
more settled research position. His English is quite reasonable; but he has
published maninly in Russian -- esp. on a REFAL language (I think a symbolic
interpreter for algebraic expr..) John Backus reputedly knows something of
his work.
Do either of you?
In any case, please look out for the hardcopy stuff I will send you soon,
which includes a recent (1/79) SIGPLAN summary of REFAL.
Josh
-------
∂08-Apr-79 1712 ME one segment programs
To: RPG
CC: JMC, LES, DCL
The official recommendation of running one-segment programs instead of two
is a recommendation only; it applies particularly where it is trivial to
make the program run in one-segment, without introducing (additional)
bugs. For instance, SAIL programs should be run in one segment. Although
it would be nice to be able to run MacLisp in one segment, no undue
pressure should be felt to sacrifice reliability in order to do that.
Conversion to one segment MacLisp can continue with whatever effort there
is available, but we certainly don't want to force people to use a version
that doesn't work, just to save a job slot. Of course, if several people
simultaneously are running MacLisp, and each has a different upper
segment, then there will be more pressure to use a one-segment version,
but the 256K version is acceptable unless we reach the point where we run
out of core, but I don't expect that to happen. And if several people are
sharing a MacLisp upper segment, then that is only one job slot that could
be saved (at the cost of a lot of core), so there is little pressure to
use an undebugged version in that case.
∂08-Apr-79 2034 PTZ purge
To: S1 at SU-AI
I was mistaken about all file having been given the same purge priority.
I understood the documentation file to say that that would happen with an
incomplete purge line, and then misread the list of what had been purged.
Sorry.
Polle
∂08-Apr-79 2037 MRC
To: LES, JMC
Can't something be done about S-1's disk usage? They are always grossly
over allocation (right now they are only 1,000 kword over). Here is an
abridged extract of a message about that...
**********************************************************************
∂08-Apr-79 2022 PTZ the recent purge and OPTION.TXT[DOC,S1] purge line
To: S1 at SU-AI
S1 files were hard hit by the purge yesterday (EJG is in the process of
restoring them now).
**********************************************************************
The disk space we got is just going to go away again when they restore
all their files. I don't like all this system downtime being taken up
by purges; 5 so far this year! But it's been necessary just to have any
disk space to work with at all.
∂08-Apr-79 2316 ME disk drives
∂08-Apr-79 2257 JMC
What is present situation of new disks?
ME - One of them is waiting to get its heads aligned. The Ampex guy
came to do it Friday, but his drive-testing unit was broken so he
couldn't do anything. Presumably he'll come back as soon as he gets
his equipment repaired (or gets another good one). The other disk
drive seems to work OK.
∂08-Apr-79 2330 MRC messages
Well foo. I was sick last night and most of today. I'm still pretty
gronked. With any luck, I MAY get to some Dialnet work tonight. Or, I
might end up sick for a week.
This sort of pressure and having to answer it isn't doing much towards
speeding up Dialnet. I can understand why you are upset about the time it
is taking but you should consider that this isn't the sort of thing that
can be hacked together with a lick and a promise if you want a truly
cuspy, well-engineered protocol and implementation.
Almost every day I learn something new about the internals of WAITS and
TOPS-20 which require a minor rethink&rewrite of my code. About once a
week with TOPS-20 and once a month with WAITS I learn something new that
requires a major modification. My knowledge acquisition has been faster
than I have been able to keep up with in coding. On TOPS-20, I had NO
documentation about its internals other than the code. And, unlike WAITS,
I did not have any people to ask questions of. Ralph and JQ don't know
anything about the stuff I needed to know. Neither does anybody else in
this area. I'm now the local expert in talking to the front end and in
device service routines. I sure could have used me to ask questions of
about six months ago.
-- Mark
∂09-Apr-79 0058 ME
You are exceeding your disk quota.
Files that occupy space beyond your quota are subject to purging,
** ESPECIALLY if there are less than 1000 free disk blocks in the system! **
If you don't delete some of your files, the purger will.
Your disk quota is: 2000
Your files occupy 2414
∂09-Apr-79 0949 PAT
∂06-Apr-79 1615 JMC
To: PAT
CC: MAX
Please find out how Ma can get a Stanford ID card?
---
I've already set the wheels in motion...
∂09-Apr-79 1052 Pickering at SUMEX-AIM reprint copy
To: S1 at SU-AI
Date: 9 Apr 1979 1045-PST
From: Pickering at SUMEX-AIM
Subject: reprint copy
To: s1 at SAIL, mjb at SAIL
cc: wiederhold
please let me (or jayne) know if you are interested in a copy of a recent
article in Astronautics & Aeronautics entitled "A Trend Toward Inexpensive
Computer Simulations", which discusses the S-1 among other large processes.
Gio
-------
∂09-Apr-79 1455 Ijcai79 at SUMEX-AIM IJCAI Reviews
Date: 9 Apr 1979 1328-PST
From: Ijcai79 at SUMEX-AIM
Subject: IJCAI Reviews
To: Dear Referee:
PLEASE - We need your IJCAI reviews now!
Please submit any outstanding reviews you have to the
Program Coommmitteee by this Wednesday. You may
send them in a msg to ijcai79@sumex-aim, if that is more
convenient.
Thank you for your help. - The IJCAI-79 Program Committee
-------
∂09-Apr-79 1638 DON
To: JMC
CC: ME
∂09-Apr-79 1633 JMC
To: DON, ME
Is there any way of running DFIND out of E?
DON - Not currently; what did you have in mind? Something like E's MAIL
commands? I could probably arrange for (D)FIND to do something like, if
entered at the RPG startup, using a file pointed to by certain ACs to
determine the search parameters (instead of reading from the terminal).
Unlike the ⊗X MAIL command, however, it would entail exiting from E.
But let's hear what Marty has to say about it from the "other end".
From the user's point of view, a need for verifying a spelling arises
while writing. I would like to write "⊗XDFIND advertise" and have it
confirm for me that "advertise" is spelled with an "s" rather than a "z".
∂09-Apr-79 1718 Mark Crispin <Admin.MRC at SU-SCORE> [Yeager at SUMEX-AIM: Re: Sumex ''Dialnet'']
Date: 9 Apr 1979 1711-PST
From: Mark Crispin <Admin.MRC at SU-SCORE>
Subject: [Yeager at SUMEX-AIM: Re: Sumex "Dialnet"]
To: Les at SU-AI, JMC at SU-AI
Thought you would be interested... Taking time to do a good design
does pay off...
---------------
Mail from SUMEX-AIM rcvd at 9-Apr-79 1424-PST
Date: 9 Apr 1979 1419-PST
From: Yeager at SUMEX-AIM
Subject: Re: Sumex "Dialnet"
To: Admin.MRC at SU-SCORE
cc: YEAGER
In response to your message sent 1 Apr 1979 1149-PST
Hi Mark, I'm just testing a new MSG. It is now available NETWORK wide
from BBNA. It can be retrieved by FTPing from BBNA after logging in
anonymous with any old password. It resides on <MSG> and there are
versions for the various monitors. So, you can let any of you friends
know if they don't already.
TTYFTP is coming along. I've just about finished making the SERVER FTP,
and should be done by the end of the week.
I really like your design. The NOP every five seconds or so saves a
lot of code. One does not have to worry if the window is full. Things
move right along! Your algorithm is quite elegant...
Bill
-------
---------------
-------
∂10-Apr-79 0941 Sacerdoti at SRI-KL (Earl Sacerdoti) reference request for David Wilkins
Date: 10 Apr 1979 0936-PST
From: Sacerdoti at SRI-KL (Earl Sacerdoti)
Subject: reference request for David Wilkins
To: jmc at SAIL
cc: Sacerdoti, hart
John,
David Wilkins has applied for a position in SRI's Artificial
Intelligence Center. Would you please send me a short note assessing
his ability to perform innovative research in the SRI environment?
In particular, to what degree is he a self-starter, how well does he
work in a group, how well can he communicate his ideas, and what is
his potential as a group leader? Finally, how does he rank compared
to the to other students you've dealt with in the last few years?
Thanks in advance for your time. --Earl
-------
∂10-Apr-79 1538 ME
To: JMC, LES
I am having MJC change the default mode in SNAIL from /SAISEG to
/NOSAISEG, in order to make SAIL programs by default run in one segment.
(It's not done yet, but will be sometime soon.) Let me know if you have
any objections.
∂10-Apr-79 1701 LES Consulting agreement
To: TED, JMC, HVA, REG, ROB
Here is a draft consulting agreement for Ted Panofsky. Comments solicited.
OFFICE MEMORANDUM * STANFORD UNIVERSITY * OFFICE MEMORANDUM
STANFORD ARTIFICIAL INTELLIGENCE LABORATORY
9 April 1979
To: John McCarthy, Director, SAIL
From: Ted Panofsky & Les Earnest
Subject: Consulting agreement
This memo describes the terms of a mutually beneficial consulting
agreement between Ted Panofsky and the A.I. Lab. This agreement holds
for a period of six months beginning 1 March 1979 and may be
renegotiated thereafter.
Ted Panofsky will provide consulting services to the Laboratory upon
request by one of the following persons: Les Earnest, Ralph Gorin, John
McCarthy, Robert Poor. When working in this capacity, he will log in on
the SAIL computer as TED. Each month, he will itemize the days and
hours spent working for the Lab and also specify who requested the
work. Ted's account will be credited for this work at a rate of $30 per
hour.
Ted may also use the Lab's computer system for other design projects of
his choice. When performing these functions, he will log in as TP. The
existing computer accounting system will be used to evaluate computing
services consumed in this way, using the conversion 1 doubloon = $1,000.
For prime time (weekday afternoons), this is equivalent to the following
formula:
$.50/job hr. + 1.00/cpu min. + .05/K min. + .0002/dskop;
where "K min." refers to core memory utilization while running and
"dskops" refers to the number of disk operations performed. On morning,
evenings, and weekend afternoons, these charges are cut in half; at
other times, they are reduced to one-third.
At the end of any given month in which Ted's cumulative credits exceed
the Lab's charges for the TP account, Ted may request a cash payment for
the difference or any part thereof. At the end of any month where the
Lab's account is ahead, the Lab may request that Ted provide more design
services or that he curtail his use of the computer.
The agreement with Ted looks ok to me.
∂11-Apr-79 0043 MLB Overlayed opcodes in FSIM.
To: "@S1.DIS[P,DOC]" at SU-AI
This message is addressed to users of FSIM.
Because it is inconvenient to add new opcodes whenever we need them, PMF
and I have been forced to occasionally patch the simulator to make an
obscure, unused or unimplemented instruction call the code for a new
instruction. For example, we recently usurped RPFILE to implement the new
SEXCH (sorted exchange) instruction.
Since there is a small but finite probability that someone might be
surprised by these "overlayed" opcodes, I have provided a facility for the
responsible programmers to cause FSIM to print a warning the each time an
overlayed instruction is executed for the first time. FSIM.0[DOC,S1] has
been updated accordingly. Hopefully this will be sufficient warning.
- Marc
∂11-Apr-79 0717 Feigenbaum at SUMEX-AIM Re: V. Turchin
Date: 11 Apr 1979 0716-PST
From: Feigenbaum at SUMEX-AIM
Subject: Re: V. Turchin
To: Lederberg
cc: mccarthy at SAIL
In response to your message sent 8 Apr 1979 1527-PST
Josh, I don't know Turchin--except for having seen his name in the
newspapers. I will look for the material . Good you alerted me. I hardly
ever read hard-copy mail these days.
Ed
-------
∂11-Apr-79 1501 Ijcai79 at SUMEX-AIM reviews
Date: 11 Apr 1979 1456-PST
From: Ijcai79 at SUMEX-AIM
Subject: reviews
To: jmc at SU-AI
cc: ijcai79
John,
We are trying to get the last outstanding IJCAI reviews in so that
we can make our final decisions by the end of the week. We have sent
you 5 papers for you and your colleagues to look at:
J.Doyle & J. McDermott -- Non-monotonic logic
J.M. Hullot -- Pattern Matching
F.Ouabdesselam -- Natural Deduction
M.Shimura -- New Modal Logic
'A.Cohn -- Many Sorted Logic
Would you please give me "bottom-line" accept/reject decisions
as soon as possible and follow these up with written comments in the
next week? I am returning comments to the authors so they can
improve the papers or at least understand the reasons for your
judgment, so I still would like the comments as well as the judgment.
many thanks,
Bruce
-------
∂12-Apr-79 1554 DCO
To: "@GROUP.DIS[1,DCO]" at SU-AI
PROGRAM VERIFICATION AND ANALYSIS SEMINAR
PLACE: ERL 237
TIME: 2:30 p.m., April 17
TITLE: Logic for Computable Functions
SPEAKER: Frieder von Henke, Stanford University
ABSTRACT:
The Logic for Computable Function (LCF) has been proposed by Dana
Scott as a calculus for reasoning about partial recursive functions. LCF
is based on the lambda-calculus, enriched by conditionals and a fixed
point operator, and is embedded in a Gentzen-type natural deduction
system. The logic has been further developed and implemented by Robin
Milner; it has been applied mainly to formalizing, and reasoning about,
programming language syntax and semantics, and properties of programs.
This seminar is a tutorial on the various forms of the logic and
its implementation. Specifically, we will discuss features of the
Edinburgh LCF system. This system includes a programming language that
serves as a metalanguage for the logic. It allows a user to program
proof strategies but protects him against faulty proofs.
∂12-Apr-79 1900 JMC*
pay Colwell
∂13-Apr-79 0057 REM Appendix to my to-publish paper on data-compression methods
To: "@CRUNCH.DIS[1,REM]" at SU-AI
The file 79412B.APX gives details of the left-context modified-Huffman
data-compression method as implemented in my programs of 1976, namely
CRU1, CRU2 and CRU3. It is meant to accompany 79219A.WRU as an appendix
or maybe edited into it as a new section somewhere before the "future work"
section. I've run it through SPELL but haven't proof-read it yet -- I just
typed it in, entirely, in one session at the keyboard tonite, while up at
the SU-AI lab. As usual, you may FTP and create one hardcopy for your
personal, but please refer other interested parties to me rather than
sub-distribute listings or file-pointers yourself. Although no patentable
secrets are given, I'd like to retain the right to publish first.
∂13-Apr-79 0105 REM
You got the very first listing of 79412B.APX, just before SPELL, sorry
for spelinerrors.
∂13-Apr-79 0158 MRC April 6 Beijing Review
In this issue is an interesting re-evaluation of Confucius. My copy
is in the lounge.
∂13-Apr-79 1119 Mcgoveran at SUMEX-AIM CPPC Luncheons
Date: 13 Apr 1979 1117-PST
From: Mcgoveran at SUMEX-AIM
Subject: CPPC Luncheons
To: faculty.list:
Memo from Shirley Lewis of the Career Planning and Placement Center(CPPC).
As you may know, during each academic year the CPPC has numerous companies
interviewing students for permanent employment upon graduation. Occasionally,
the company will request luncheon arrangements with the appropriate faculty
and CPPC staff during their stay.
The CPPC would like for as many of the professors as possible to participate in
the various luncheons throughout the year. These luncheons often serve as a
means to provide good public relations between Stanford's individual depart-
ments, and companies. Often, because of this, we build a stronger possibility
of gaining or strengthening an affiliate.
The period of recruitment fruns from October through November, January through
March, and approximately three weeks in April. would you please supply the
CPPC with the names of specific professors who would be interested in
occasionally going to the various luncheons (approximately 2-3 times per
academic year).
Thank you for your time regarding this matter. If you have any questions,
please phone me at 7-3963.
-------
I might be interested depending on what companies.
∂13-Apr-79 2047 100 : ME new file pack
To: JMC, LES
I'll put up a new system pack Sunday evening. No time tonight (even for purge),
since ROB (and I) spent most of the time checking out the disk drives.
∂14-Apr-79 0401 100 : Committee to Nuke the Whales
1) Support Our Boys in Diablo Canyon
2) Let's Terraform Colorado!
∂14-Apr-79 1300 RWW FOL FIXES
To: "@FOL.DIS[FOL,RWW]"
I have fixed TTY command so that it is usable. It will get better.
I have also found the ill UUO bug and fixed MMFOL.
∂14-Apr-79 1654 ARK S-1 meeting notice
To: S1 at SU-AI
CC: pickering at SUMEX-AIM
There will be an S-1 meeting on Thursday, 19 April at 1:15 pm (in the
Serra House Conference room, I think). Note that this meeting has been
rescheduled to avoid conflict with Hector Garcia's oral.
Arthur
∂15-Apr-79 0452 REM at MIT-MC (Robert Elton Maas) Sorting and searching, data compression
Date: 15 APR 1979 0751-EST
From: REM at MIT-MC (Robert Elton Maas)
Subject: Sorting and searching, data compression
To: DEK at SU-AI, RWW at SU-AI
CC: JMC at SU-AI, REM at SU-AI
It occurs to me that my IRSM method has application in storing
inverted lists (indexes, lists of all documents having a given keyword).
You store a bit map of 1 for each document having the keyword and 0 for
each document not having it, but you data-compress this bit map by
using IRSM. If exactly half the documents have the given keyword, no
compression is obtained, and you just get the full bit map as the most
efficient way to store the info. If extremely few documents have the
given keyword, the compressed bit map has about the same size as a
simple list of document numbers. If extremely few documents DON'T have
the given keyword, the c.b.m. is like a list of exceptions. In intermediate
cases, like more than a thousand documents have the keyword but less than
ten percent of the total, IRSM may be considerably more efficient (less
storage for the inverted index) than any other storage method for inverted
indexes. The advantage, beyond simple optimality, is that a single method
works for all densities of 1 bits, rather than special cases for each
range of density. For such a simple Markov model, p(1) a constant for
each keyword, p(0)=1-p(1), no other possible values of token compressed,
the compressor and decompressor can be made quite fast, and can be put
in hardware easily. Selling such a program or device to Lockheed for
use in Dialog might be a way to make a quick 100 grand.
Your ideas seems good, but alas, my own experience in trying to peddle
ideas leaves me very skeptical about making a quick 100 grand or even
a quick one grand. The resistance to new ideas is such that it is
even difficult to give them away.
∂15-Apr-79 1122 RWW
To: "@FOL.DIS[FOL,RWW]"
Meeting
APR 17: 2:45-4:00 Bldg 120 room 123; meeting of FOL group
Jim Ward will tell us about his proof of theorem 3.9 in Mendelson.
Ben will tell us what he's up to.
We will discuss (if we have time) merits of various languages recently
developed by Feferman and Barwise.
∂15-Apr-79 1452 JC Meeting
Jim Rosse asks if you and Les can be at the meeting tomorrow
morning at 9:00 in the conference room at H&S. I have left
word for Les, who is away for the day, but Joan said probably
he could. Feigenbaum is confirmed. I told Rosse that you had
declined, but he felt that at least one of you should be there
and better both.
John
One of us will be there.
∂16-Apr-79 0221 MRC So much for "Free China"
To: JMC, DBG
TAIPEI, Taiwan (AP) - The Taiwan Garrison Command's military court
sentenced one man to death today, two to life imprisonment and 10
others to terms of eight to 15 years for conspiring to overthrow the
government and sedition.
A spokesman said Wu Chun-fa, 55, was sentenced to death and Li
Yung-ho, 50, and Lin Yung-hsiao, 48, were given life sentences.
The court also sentenced a 76-year-old former magistrate, Yu
Teng-fa, to eight years and his son, Yu Jui-yen, 53, to two years for
spreading communist propaganda and not reporting a communist agent to
the government. But the son's sentence was suspended because he is in
poor health, the spokesman said.
---
Not obvious. It seems like suppression of dissent, because the story
does not mention their being agents of the mainland government, which
also shoots Kuomintang agents, but such a charge may have been omitted
from the story. I fear that the Kuomintang government uses a genuine
threat to their survival as a reason for unjustified suppression, but
I don't know it for a fact.
∂16-Apr-79 0321 LLW Visiting Faculty Appointment for John Reiser
To: JMC, LES
CC: LLW
I spoke with you, John, a few months ago, and with you, Les, a couple of
weeks ago about a visiting faculty appointment for John Reiser; your
responses were both of encouraging flavors, and it was left up to me to
suggest when would be an appropriate time for the associated paperwork to
be ground through by Stanford.
As a result of recent conversations with John by others and by myself, the
appropriate time quite definitely seems to be now. Reiser is presently
winding up the work which has occupied him for more than a year, and is
now in the process of deciding whether to finish seeking his management's
permission to collaborate on multiprocessor operating systems and
lanaguages research with SU/CS and the S-1 Project, or to undertake some
other, more applied research of relatively keen interest to his
management. He has stated quite unequivocally in these recent discussions
that a visiting faculty appointment at SU/CS was tantamount to a sine qua
non for the former option to materialize (more so in the eyes of his
management than in his own), and that he would be quite happy to offer a
seminar series or whatever as a quid pro quo.
Since as you remarked, John, a visiting faculty appointment at zero time
costs nothing, and involves nothing other than pushing paperwork through
your local bureaucracy, I will certainly appreciate your commencing the
paperwork at the present time. Reiser indicated that he would be asking
for permission to spend about half his total professional effort with us
for the next year, starting this Spring. LLL would pay his travel and
subsistence expenses during this period, and BTL would continue to pay his
full salary.
John's addresses and phone numbers are as follows:
Business: Dr. John Fredrick Reiser
Technical Staff Member
Bell Telephone Laboratories
Crawford Corner Road
Holmdel, NJ 07733 201-949-3942
Home: 53 South Street
Red Bank, NJ 07701 201-741-3128
Thanks very much for the help. I trust that Reiser's being in California
during this next 12 months will be a marked benefit to both SU/CS
Department and to the S-1 Project!
Lowell
Ed: How does this strike you? It seems to me to have mainly advantages.
∂16-Apr-79 0957 PAT
∂13-Apr-79 1554 JMC
Please get me "Integer Programming" by T.C. Hu from the library.
---
Is checked out to someone, Richard has put a recall on it and will
call me when it is returned.
∂16-Apr-79 1400 TW
I am working on some chapters of my book for which your papers
on circumscription induction and ascribing mental qualities to
machines are relevant. I have adopted a general policy of making
references only to works easily available to students. I assume
that the version of the c.i. paper in the AI journal on non-monotonic
logic that Danny is doing is best. Is there anything for the
other one? --terry
The paper on ascribing mental qualities will appear as an AI memo
and in a symposium volume. Patte can give you the precise reference.
∂16-Apr-79 1349 LES DECUS Meeting
To: MRC at SU-AI
CC: JMC at SU-AI, REG at SU-AI
Regarding your proposed trip to the Spring DECUS meeting, I wish to attach
a condition: you may go if Dialnet demonstrably works between two sites
by May 15. This is to include at least file transfer and mail and preferably
telnet as well.
∂16-Apr-79 2006 REM at MIT-MC (Robert Elton Maas)
Date: 16 APR 1979 2235-EST
From: REM at MIT-MC (Robert Elton Maas)
To: JMC at SU-AI, RWW at SU-AI
CC: rem at SU-AI, rwg at SU-AI
Yesterday when I checked my us-mail I found the NTIS published search that
I had ordered (data compression from Ei, over 200 abstracts, mostly on
facsimilie images (digitized TV frames), EEG and ECG, voice but some text
compression. I'll be checking into some of these articles in the upcoming
weeks. There was one report on updating Huffman codes incrementally, but
in the ASILOMAR conference proceedings which I remember trying to get
several years ago without success. Nothing close to my left-context method
of 1976, although a couple papers with ambiguous abstracts might be
similar, and nothing at all like my IRSM invention of 1978 nor even the
mathematical precursor to it of 1971-77.
∂16-Apr-79 2057 BPM
To: JMC, LLW
"People in the News"
4:04 pm, 16 April 1979
STANFORD, Calif. (AP) - Nobel prize-winning physicist Edward Teller
says he is ''giving serious consideration'' to a request by Sen.
S. I. Hayakawa, R-Calif., that he run against California Democratic
Sen. Alan Cranston in 1980. Cranston is approaching the end of his
second six-term in the Senate.
Teller, 71, a senior research fellow at Stanford University's Hoover
Institution and best known as ''father of the H-bomb,'' said in a
recent interview he had not made up his mind about running and did not
know when he would make a decision. But he added:
''I respect Cranston, but I am very opposed to his stand on vital
policy questions.'' For example, he said he against Cranston's energy
policies, including Cranston's opposition to nuclear energy.
∂17-Apr-79 0154 LES ARPA proposal
To: DCL, TOB
CC: JMC, HVA
Its that time again. Bill Carlson suggests we go for two-year funding
this time. JMC plans to visit ARPA around May 3, so it is highly
desirable that he have a draft of the entire proposal by then; preferably
a few days earlier so that we can put it together smoothly.
Since it has been just a year since our last proposal, the text need not
be totally original. I will get together with you to review budget needs.
∂17-Apr-79 0749 Mark Crispin <Admin.MRC at SU-SCORE> DECUS and Dialnet by May 15
Date: 17 Apr 1979 0749-PST
From: Mark Crispin <Admin.MRC at SU-SCORE>
Subject: DECUS and Dialnet by May 15
To: Les at SU-AI
Cc: JMC at SU-AI, Admin.Gorin at SU-SCORE
That is an incredibly bad idea. It took over a year for Dialnet to come up
on WAITS (and longer if you include that WAITS DLNSER is still evolving),
and on WAITS I had the advantage of a system that compiles and links in
a couple of minutes real time, people whose brains I could pick for
questions, and a system whose bug traps are designed for monitor debugging.
On Tops-20, there is first a lot more to learn. I am picking up a lot
very fast, but I also know there is a lot more. At GSB, it takes about
20-30 minutes real time to compile, link, and reload the monitor. SCORE is
somewhat faster, but still nowheres near SAIL in fast system turnaround.
I don't know of anybody here who knows enough about Tops-20 device service
routines to help me; certainly nobody around Stanford. The only person I
know who has done anything close to this is Mike McMahon at MIT (formerly
at SRI). Finally, Tops-20 is hard to debug. The bug trap mechanism is
scientifically designed to screw you unless you carry a complete listing
of Tops-20 with you or have one on site. The reason is that it is
oriented around "crash dumps" more than around online crash debugging.
All this has taken a good deal of time. The code is written and has been
written for a long while now (since last fall). The problem is that just
when I think I can start using it I find out about something new which
requires me to rewrite a large section of code. I think I've done pretty
damn well considering the amount of help I've gotten. I doubt very much
that anybody else could do much better given what I've had to work on.
-------
∂17-Apr-79 1037 PAT visit to ARPA
Carlson's secretary called. She wants to set up a schedule for your visit
on May 3 and would like you to call her. Judy @ 202/694-5917.
∂17-Apr-79 1519 100 : REM via SU-TIP CRUNCH PATENT
To: RWW, JMC, REM
Via NTIS bibliography, reading some articles it pointed to, I found
two nice things this afternoon:
(1) Paper published 78.Feb saying "If fractions of a bit could be
generated in storage medium then exact match to entropy could be
obtined..." - I plan to write to this guy and tellhim fractions can!
(2) Reference to somebody who actually got a patent on a data-cmpression
algorithm or device. I plan to track down this patent by author name
and find how it is indexed in patent system!!
∂17-Apr-79 1544 DEW
To: JMC at SU-AI, berliner at CMU-10A, lenat at SUMEX-AIM,
buchanan at SUMEX-AIM, DEW at SU-AI
Oral exam finalization and progress report
My oral exam will be at 2:15 p.m. on Wednesday, 16 May 1979 in ERL 410.
Please put this on your calendar and confirm it with me.
For those of you who wanted it in the morning: the Graduate Studies Office
has a new rule saying all orals must be at 2:15. When I said that one
of my committee members might only be available in the mornings, they told
me to find a new committee member!
Progress report: The first two chapters will be reorganized. Terms will
be used more precisely and defined (with examples) earlier. Comparisons
with Lenat's AM and KAISSA will be added. The reorganization will be a
fairly major project, so I have given up hope of having everything signed,
sealed and delivered by June 6. Expected graduation date is now July
or August of 1979.
∂17-Apr-79 2015 Zenon.Pylyshyn at CMU-10A (X320ZP51) Your letter re conference in October
Date: 17 Apr 1979 2313-EST
From: Zenon.Pylyshyn at CMU-10A (X320ZP51)
Subject: Your letter re conference in October
To: JMC @ SU-AI
Message-ID: <17Apr79 231314 ZP51@CMU-10A>
The idea of an interdisciplinary conference sounds fine -- though it might
be most useful if it were kept strictly limitied to the invitees and the
Center people (I suspect this is what you had in mind). I have already
expressed my positive views on the topic to Ken Klivington at Sloan when
he asked for my opinion on Lindsey's application for money. I will give
some thought to format and participants in the next few days. The only
thoughts that come to mind immediately are the names of two people not on
your list that I think would contribute immeasurable* Al Newell and Scott
Fahlman. I just heard John Searle: he might be OK though totally negative
(he is a good representative of the Dreyfus view, but somewhat more
articulate. There must be more philosophers, but I can't pick their names
out just now. I do know that many of the people on your list are very
unlikely to come (e.g. Chomsky, Fodor). Also I would very much like to
see some cognitive psychology oriented theorists, though the only one I
can think of is Newell. I will append more as it occurs to me.
-------
Cheers,
Zenon
I'll pass this on to Bob Moore, who will be doing the organizing. I talked
with Chomsky when he lectured here at Stanford, and although we argued
fiercely about Vietnam, we found ourselves in agreement about many aspects
of the relevant philosophy, and he said he might come to the conference.
Anyway he wanted to be invited. Newell, Fahlman and Searle can certainly
be invited.
∂18-Apr-79 0038 MRC importing McMahon to help
To: JMC, LES
I just finished talking to Mike McMahon about his coming here to help out
with Dialnet. He has agreed to help, and will have an answer for me about
travelling here sometime later this week or earlier next week. In the
meantime, he has taken a copy of the latest source and will be going over
it for what I should take care of in the meantime.
So let's start the ball rolling. I told him the time was more or less at
his convenience, except that if at all possible we'd like to get things
taken care of by May 15. He didn't seem to think that would be a problem.
∂18-Apr-79 0913 DCO
To: "@GROUP[1,DCO]" at SU-AI
PROGRAM VERIFICATION AND ANALYSIS SEMINAR
PLACE: ERL 237
TIME: 2:30 p.m., April 24
TITLE: Understanding Denotational Language Definitions
SPEAKER: Wolfgang Polak, Stanford University
ABSTRACT:
Denotational semantics has frequently been critized as being
baroque and incomprehensible. This talk attempts to explain how to read
and write denotational language definitions. The talk will be very
practically oriented and take most of the underlying theory for granted.
Denotational definitions use a small set of standard techniques and
concepts, such as continuations environments. Their use will be
demonstrated with the definition of a simple example language including
features such as expressions with side effects, block structure, and
jumps.
∂18-Apr-79 1113 Sacerdoti at SRI-KL (Earl Sacerdoti) Dave Wilkins recommendation
Date: 18 Apr 1979 1111-PST
From: Sacerdoti at SRI-KL (Earl Sacerdoti)
Subject: Dave Wilkins recommendation
To: jmc at SAIL
cc: sacerdoti
John,
Sorry to dun you, but we'd like to make Dave an offer & need a second
recommendation for our own paperwork. Please send me your comments, however
brief. Thanks. --Earl
-------
∂18-Apr-79 1420 DCL ARPA PROPOSAL and Project costs.
To: JMC
CC: LES, TOB
John,
Glad to hear about SCORE.
I think our immediate problem is to spread the lab. overhead over more
of the machine users, e.g., MUsic, S1 and CS.
We discussed this and I think you agree in principle.
I have asked LES to cons up a "dummy" budget based on last year's project
personnel, whatever changes are to be made to systems and administration
personnel, and next year's salaries.
This way we can get some idea of what we are facing vis a vis spreading the
costs.
I have asked LES to do this by Friday; I shall be in Washington all next week.
-David
∂18-Apr-79 2033 DCL
To: JMC
CC: LES
∂18-Apr-79 1550 JMC conversation with Lowell Wood
To: LES, DCL
He returned my call. We can expect the Mark I sometime
next winter, but it won't be usable as a time-sharing machine.
The second Mark 2A is scheduled for NBS, and we will get one
of the following batch. Time-sharing is one-and-a-half to two
years away on his schedule. ARPA money could advance that, but
Lowell is skeptical as to whether Carlson has money to back up
his ideas. I told him that I thought the S-1 project should
pay a larger share of the operating expenses of the KL, and we
agreed to discuss it next week. We need cost information by
then.
REPLY:
John, this is progress!
Cordell doesnt want to buy time. BUT I DIDNT HAVE ANY FIGURES.
Could we have the dummy budget by Friday so I can see it before I go to
washington.
-David
∂19-Apr-79 1916 Davis at OFFICE-2 The TEX Formatter
Date: 19 Apr 1979 1847-PST
From: Davis at OFFICE-2
Subject: The TEX Formatter
To: jmc at SU-AI
cc: davis
Dear Professor McCarthy,
I am a graduate student in the Mathematics department at Stanford,
and I have finished nearly all of the research for my thesis, which I
am now in the process of writing up. I have looked over the
documentation for Knuth's TEX formatter, and I would love to use it to
produce the final copy of my thesis.
My roommate, Martin Frost, checked with Les Earnest about the
possibility of doing this. Les said that this is usually not allowed,
but suggested that I talk to you.
The thesis, unfortunately, is in applied mathematics (it is on an
extension of the Bernstein polynomials to higher dimensions), and has
little to do with Computer Science, let alone Artificial Intelligence.
On the other hand, I would be a very small drain on your resources -- I
work full time for Tymshare on Doug Engelbart's NLS system, and I could
easily do all of the entry and editing on my machine. What I would
need would be an account with a few hundred working pages and
essentially zero permanent pages which would be used only at night or
on weekends.
I do not know how long this will take -- my experience indicates
that such things always take longer than one thinks, but I will simply
be writing it up since the research is complete. The rough draft is
nearly done, and there seem to be no major problems. My guess is that
it will take about three months.
I would love to do it using our own NLS system, but of course, we
have nothing that even begins to compare with Knuth's TEX for
formatting mathematical text. Also, I would like to do it to prove to
my advisor (Professor McGregor) that computers are not completely evil.
Thank you for considering my request. Let me know if I can give
you any more information.
-- Tom Davis
Please tell Les that I said it would be ok for you to use the machine
to produce your thesis. My motivation is also to get the mathematicians
interested in using computers more.
∂20-Apr-79 0311 LLW S-1s at SAIL, Etc.
To: JMC
CC: LLW
Dear John:
A slight updating of my remarks of yesterday:
1) Our present expectation is that the Mark I could be transferred to SAIL
for S-1 Project-related software development, per our previous agreement,
sometime this coming Winter. Project finances permitting, I expect that
we will be able to make enough Mark IIAs to also be able to make a similar
long-term loan of one to SAIL for Project support purposes commencing
sometime this coming Spring; however, since I don't yet have firm input re
FY79 Project financing from Washington, I can't make a firm commitment on
this Mark IIA loan schedule.
2) Contrary to what I told you yesterday, it is the intention of the local
OS development effort to bring up a version of UNIX with time-sharing
features by the time the Mark IIA prototypes become available next Winter.
Any Mark IIA that we would loan to SAIL would therefore have a fairly
well-characterized time-sharing system available for it, moreover
pre-tested by the local user community. It would presumably be rather
vigorously maintained thereafter, for similar reasons.
3) The Project has NO interest in operating a production line to make Mark
IIAs 'for export,' and it would be difficult for anyone (e.g. DARPA, whose
past support of the Project has not been completely overwhelming) to
induce us to develop a great deal of interest in such an undertaking. Our
present commitments extend to building one for NBS (which has an
application for it which personally interested several of us) and one for
loan to SAIL, and no more. We do not presently view the Mark IIA as a
trivially producible system (in contrast to the Mark III, whose
manufacture has a stamping-out-cookies flavor from the current vantage
point), and we intend to make no effort to accommodate Johnny-come-lately
admirers of the Project with Mark IIAs dropped on their doorstep at our
incremental production cost. (Such new-found friends may queue up now to
buy into the Mark III generation.) Only if a really urgent Government
interest were to be served would we be amenable to pointlessly (from our
standpoint) stretching out our Mark IIA experience by making
'machines-for-sale.'
I would not expect this stance on our part to adversely impact any SAIL
interest in developing INTERLISP (or anything else)--a single Mark IIA
should offer you several times the computing power of the KL-10, in a
worst case comparison, and perhaps an order-of-magnitude gain, on the
average. Its address space is 3 orders of magnitude more capacious than a
-10's (thanks largely to your forceful inputs to its design in its
formative stages), and can be physically realized at a current (and
exponentially decreasing) cost of only 20 K$/megaword. I really don't
understand what 2-4 Mark IIAs could do for any SAIL-sized computing
community that a single one couldn't do quite as nicely (at least until
the Mark III generation comes along, 2 years hence).
4) In response to your last question yesterday, our present thinking
regarding paying for our share of the KL-10 operating costs next fiscal
year really hasn't solidified yet. I do note that the total fraction of
system resources used by everyone Les associates with the S-1 Project
(including people like Marty Frost, who works for the Project in only a
titular sense) has averaged 12% since the beginning of the year, and that
less than 5% has been used by LLL staff members (the other 7% being
charged by Stanford faculty and students working on Project sub-contracts
part-time, and presumably on their own research the rest of the time
charged to the Project). Our current expectation is that our use of the
fairly rapidly evolving capabilities at LLL will be such that we will
average 2-3% of the total SAIL usage by next Fall, and will continue to
decline in fractional usage during the coming fiscal year. What the
sub-contract-related use will be during next year is currently quite up in
the air, both because we don't know what work will be proposed and how
computing-intensive it will be, and because we are unaware of how much
other SU/CS assets (e.g., SCORE) will be used in the course of this work.
I therefore estimate that the total S-1-Project-related usage of the SAIL
facilities would be rather stiffly upper-bounded by 8% over the FY80
period. I am therefore currently planning to budget at least this
fraction of the SAIL KL-10 system operating costs (which I estimate to be
at most 200 K$ annually) as a support cost item in the continuation of the
SAIL sub-contract from the S-1 Project into the coming fiscal year. If
you have any problems with this, please let me know.
Please give me a ring at your convenience, if you would care to discuss
any of the foregoing at greater length.
Warmest regards,
Lowell
We shall find out what Carlson had in mind - i.e. whether he was thinking
of Mark 2a or Mark 3 or was awaare of the distinction. What he had in mind,
however, was not giving Stanford 2 to 4 S-1s for Stanford's use, but rather
having Stanford operate an ARPAnet facility for general Interlisp work.
The questions would be the color of his money, whetherhis time scale
matches that of the Mark 3, and what is in it for Stanford and for the
S-1 project. I'll be back to you on the other problems.
∂20-Apr-79 0324 LLW Proposals for FY80 S-1 Project-Related Work
To: JMC at SU-AI, FB at SU-AI, JLH at SU-AI, LES at SU-AI,
wiederhold at SUMEX-AIM, WVC at SU-AI
CC: LLW at SU-AI, BS at SU-AI
Due both to the long latencies normally associated with processing largish
sub-contracts through the LLL and DoE machinery, and with the even greater
delays associated with Summer vacations and end-of-the-Federal-fiscal-year
panics, I have just been informed by the local Contracts shop that they
will not guarantee on-time execution of contracts scheduled for 1 October
commencement unless all the required paperwork is in their hands no later
than 15 July. Since I need a few days to write my portions of the
required paperwork and get it typed, I will greatly appreciate receiving
(in final, previously agreed-upon content-and-form) all proposals for
continuation of present S-1 Project-related work by 9 July. If I don't, I
obviously cannot begin to promise that there won't be funding gap(s)
starting 1 October.
Thanks for your consideration of local latency problems, and for
appropriate action thereupon.
Lowell
∂20-Apr-79 0455 REM via AMES-TIP Huffman in person
To: RWG, DEK, RWW, JMC
I assume the Huffman who will be lecturing on paper folding next week
is the inventor of the optimum discrete variable-length code?
I'll try to corral him before and/or after the talk and discuss my
new data-compression methods...
The very same Huffman.
∂20-Apr-79 1448 Sacerdoti at SRI-KL (Earl Sacerdoti) thanks
Date: 20 Apr 1979 1447-PST
From: Sacerdoti at SRI-KL (Earl Sacerdoti)
Subject: thanks
To: jmc at SAIL
... for the note on Wilkins. --Earl
-------
∂21-Apr-79 2317 LLW
To: JMC
CC: LLW
∂20-Apr-79 0931 JMC
We shall find out what Carlson had in mind - i.e. whether he was thinking
of Mark 2a or Mark 3 or was awaare of the distinction. What he had in mind,
however, was not giving Stanford 2 to 4 S-1s for Stanford's use, but rather
having Stanford operate an ARPAnet facility for general Interlisp work.
The questions would be the color of his money, whetherhis time scale
matches that of the Mark 3, and what is in it for Stanford and for the
S-1 project. I'll be back to you on the other problems.
[John: I would be very happy to support a Stanford facility of most any
type with any reasonable number of Mark IIIs, in addition to the Mark I
and Mark IIA single units already planned between us. If ARPA wants to
carry the cost of the Mark III set, it's fine with me--we anticipate that
a decently equipped Mark III system should come in between 50 and 75 K$
incremental cost in a reasonably sized production run, so even ARPA might
be able to afford a few (in reality, as well as in dreams)! Lowell]
∂22-Apr-79 0957 Feigenbaum at SUMEX-AIM Re: Shrinking CS206
Date: 22 Apr 1979 0953-PST
From: Feigenbaum at SUMEX-AIM
Subject: Re: Shrinking CS206
To: JMC at SU-AI, DPB at SU-AI
In response to the message sent 21 Apr 1979 2211-PST from JMC at SU-AI
I am surprised too. Who made that deccision?
Ed
-------
∂23-Apr-79 0017 LLW
To: JMC
CC: LLW
Personal Computing Cost-Effectiveness
Dear John:
As you can see from the internal discussion paper below, your thinking and
the consensus of the Project's regarding the relative utility of smart
personal computers and more amply endowed time-shared systems during the
near term are seemingly very similar. The former simply do not appear to
well endure the cost-effectiveness test, relative to what appear to us to
be the obvious extensions of the latter during at least the next
generation of underlying technology.
Lowell
FOR DISCUSSION: A SMART TERMINAL USING S-1 ARCHITECTURE/TECHNOLOGY
Could the S-1 Project create a smart terminal embodying significant
portions of the evolving S-1 technology, expressing at least a
well-characterized subset of the S-1 architecture, and
benefitting/deriving from the Project's main line of work for the Navy?
Could such a terminal be realized without seriously straining or
distorting the Project's structure, or disrupting its Plan? If such a
terminal were to be created, what existing/already planned S-1
technologies would it use, what would its capabilities be, what would its
costs and initial availability be for various levels of capability, etc.?
This is informal documentation of a discussion of these and related points
one late evening recently between TM, JBR, PMF and LLW, with participation
from DLW, OTA, JEF, and HWC. It is intended as an issue paper which may
lead to modification of the Project Plan, if subsequent discussion
suggests that we can produce such an item without major internal
distortions, that doing so would not unduly distract from the Project's
basic Navy work, and that there is sponsorship readily available for it.
It was noted at the outset that the cost and size of the Mark IIA
processor precludes its use in a smart terminal, and that the Mark III
(LSI ECL gate array) Processor is the first reasonable candidate for such
an embedded application (on both cost and size bases). The Mark IV (CMOS
LSI gate array) and Mark V (VLSI components) Processors are progressively
more reasonable candidates for an embedded smart terminal role, but are
sufficiently far downstream at present Project funding levels that they
probably shouldn't be considered for use during the next couple of years
for this class of applications.
It was also noted that the Mark III, with its 25-50 MIPS peak processing
rate, was at first glance a rather muscular processor for a smart terminal
(even by local standards), and that a special purpose, 1-2 MIPS S-1
processor (e.g., comparable in overall performance to a KL-10) was
probably more reasonable for single-user service. Unfortunately, there is
no such processor presently in the Project Plan, though there has been
considerable discussion within the Project of the utility of an S-1
'mini,' of which this might be a first instantiation. Such a processor
might be either a severely cut-down Mark III (e.g., no pipeline,
minimal/no cache), or an ECL-flavored-2901 bit-slice implementation. The
latter would be substantially more economical in mass production (e.g.,
5-10 K$ total cost), but would require significantly greater development
effort and funding. The former would presumably derive somewhat more
readily from the Mark III development (particularly the design effort, and
the use of LSI ECL gate arrays), and thus would be somewhat more powerful,
though possibly more expensive (10-15 K$ total cost) in mass production.
TM estimated that the design effort delta would be about 2 man-years for
the bit-slice 'mini' option, and 3 man-years for the cut-down Mark III
(including microcoding in both cases), and would be mostly incurred in
FY80 (thus involving burdened costs of roughly 200 and 300 K$,
respectively). The prototyping cost delta of the cut-down Mark III could
be several hundred K$, however, unless we were very lucky and managed to
use Mark III gate arrays almost exclusively in its implementation.
All this considered, it was agreed that the 'minimum distortion' approach
to the S-1 smart terminal would involve the use of the 'pure' Mark III
processor, and a smart terminal of this flavor was the subject of the
remainder of the discussion (and of this documentation). It was noted
that the Mark III, as presently planned, will be constituted of roughly
300 LSI ECL gate arrays of a flavor comparable to the present Motorola
Macrocell technology (with something like 50% gate utilization, in order
to secure high commonality), along with about 700 4 K-bit ECL RAMs for
cache and control stores. All of these will be mounted on a total of
20-25 stiffened PC cards of the order of 6" x 9" dimensions, for ready
militarization. The estimated production cost of these Mark III
Processors (in FY79 $, 2 years hence) is about 30 K$ in dozen quantities,
consisting of about 12 K$ for gate arrays, 9 K$ for RAMs, 2 K$ for PC
boards, 1 K$ for power supply and fans, 2 K$ for connectors, and 4 K$ for
assembly labor.
To this basic cost must be added the cost of display, network and
peripheral interfaces, keyboard, disc system, minimal memory, and memory
expansion interface.
PMF suggested that the minimal display should be a 512x512 dot resolution,
raster scan-type monitor. The display memory would be organized as four
independent memories of size 256K by 9 bits. The outputs of each memory
would each drive three intensity maps. There would be an four-input adder
and intensity map for each of three color channels. There would be DAC's
assocated with individual bits, the twelve first level intensity maps and
the three color intensity maps. This gives the capability of using the
system in any of three modes:
1. Thirty-six independent 512x512 one-bit displays.
2. Four independent nine-bits/pixel color displays. Each display
would have an intensity map for each color, thereby allowing full
specification of 512 colors for each display.
3. One color display with a full nine bits/pixel for each color
and an auxiliary nine-bit overlay channel.
It was noted that this format was compatible with the required one
megabyte of nMOS 16 K-bit RAM of 200 nsec access capability, properly
interleaved to meet bandwidth requirements. This minimal display could be
duplicated within an individual system to support as many displays as
needed. It was also noted that this display format facilitates re-loading
the VRAM section of memory from high-level language programs, as it meshes
naturally with the S-1 architecture.
The minimal terminal memory module (of 1 megabyte capacity, dedicated to
display maintenance) and controller for the display were estimated to cost
3 K$ and 2 K$, (based on S-1 and LISP Machine costs, by JBR and DLW,
respectively), while the 512x512 tricolor monitor was estimated to be
available for 1 K$. A suitably compact 200 megabyte hard-surface
disc-and-controller were estimated to be available for 5 K$ in the 1981
period (with JBR reciting details of the rapid cost-effectiveness gains of
such systems at the present time). DLW estimated the present cost of a
MIT ChaosNet interface to be 2 K$. Keyboard (with programmable
microprocessor for flexibility), cabinet and minimal controls/indicators
were estimated to sum to 2 K$, and assembly and fabrication to 5 K$. An
LSI-11/23 system tied into an I/O Data Store of the Mark III for
interfacing to the ARPANET and to standard peripherals (e.g., printers,
etc.) was estimated to be available for 1 K$ (similar to the 11/2 at
present). (All of these [1979 $] cost estimates are for second generation
systems produced in late 1981, 'after the book had been written' during
the pre-prototyping process in early 1981.)
Add-on memory costs were estimated to be 2 K$/megabyte, with at least 8
megabytes being packagable within the half-desk-type terminal cabinet
itself (in 64 K bit nMOS RAMs); memory capacity extension in a separate
cabinet would add negligibly to basic memory system cost.
The cost of the basic smart terminal is thus estimated to be 51 (FY79) K$;
equipping with a full 8 megabytes of in-terminal-embedded high speed
memory would increase the total terminal cost to 65 (FY79) K$. Centered
on a S-1 Mark III Processor, such a smart terminal system would have
roughly the computing power and memory capacity of a Cray-1. It would be
cooled with (forced) ambient air, and would apparently require 15-20 amps
of 117 VAC, single phase power (e.g., be serviced from an ordinary wall
receptacle). Its external appearance would be similar to that of a
Hewlett-Packard 300 smart terminal. It would be capable of simultaneously
providing half-KL-10 levels of computing throughput and quarter-KL-10
levels of memory capacity to each of 36 users, driving 512x512
dot-plotting B&W monitors for each of them, or of servicing a total of a
dozen 512x512 dot-plotting, 3-bits/pixel color displays each with KL-10
levels of computing and memory capabilities.
If the S-1 Project were to undertake the development of such a smart
terminal, we would probably need a funding delta in FY80 of about 325
(FY79) K$, to cover a (burdened) engineer and a senior mechanician doing
system design and packaging prototyping and 125 K$ of
consumables/long-lead-time items. We would probably run the first,
pre-prototype smart terminal early in FY81, using a non-embedded Mark IIA
to operate it, as an interim capability demonstration and performance
exploration operation, before prototype replication commenced.
The required FY81 funding delta for the smart terminal work would probably
be in the neighborhood of 925 (FY79) K$, to cover the same (burdened)
engineer and senior mechanician, plus 5 man-years of assembly
technician/mechanician effort (including supervision), plus a (burdened)
man-year of documentation effort, and 500 K$ of consumables. This FY81
effort would output 10 prototype basic S-1 smart terminals and a
comprehensive 'technology data base' package for use in subsequent
commercial procurement by anyone interested in mass production.
The Navy's donation of use of the entire Mark III technology (hardware and
software) was assumed in these cost estimates. LLW noted in this
connection that LLL management approval of extending the Project in the
direction of smart terminal development wouldn't be automatic, and
probably wouldn't be forthcoming unless the Navy remained the Project's
sole proposing/reporting point (as the Lab continues under DoE pressure to
cut back on DoD-sponsored work, and a perceived-as-new program has a poor
outlook for approval). Navy supervision of any Project smart terminal
work therefore appears required by two rather independent considerations.
LLW agreed to inquire as to possible Navy interest in such development
work by the Project.
∂23-Apr-79 0918 DPB Shrinking 206.
To: JMC at SU-AI, feigenbaum at SUMEX-AIM
What happened is that we (I thought all of us) figured that we
needed the advanced Lisp programming course. Since we have in the
past (not last year, but in years before) given two flavors of 206,
the JMC version and the CCG version, we could "add" the advanced
Lisp course to be taught by Doug Lenat without adding to teaching
loads by deleting the CCG version 206. See the course 222 in the
bulletin.
I still think that this is the correct decision. JMC will be "gone"
all next year. EAF will be gone part of it. That leaves us with
Doug and Terry (and Bruce if nec.) to cover all of the courses.
Just about everybody is interested in a new course, but nobody wants
to add it/them next year when we are short-handed.
...
Larry Butas, would-be programmer of the perfect game of chess.
phone call April 23, 1979
In that case, it would be better if you would list CS206 as ordinarily
given in both quarters, but not given in Spring next year. There is
another issue connected with my desire not to allow the theory to
be squeezed out by mere programming.
∂23-Apr-79 0924 CSL.FB at SU-SCORE Digital Dataphone Service
Date: 23 Apr 1979 0917-PST
From: CSL.FB at SU-SCORE
Subject: Digital Dataphone Service
To: jmc at SU-AI, admin.gorin at SU-SCORE
Ma Bell has something called Digital Dataphone Service in this area.
In comes in 2.4,4.8,9.6, and 56.4 Kb forms. Requires "no modems".
Do you know much about it, like what it costs and if it would be
useful in a dialnet environment?
Forest
-------
∂23-Apr-79 1116 Bmoore at SRI-KL (Bob Moore) S and P problem
Date: 23 Apr 1979 1115-PST
From: Bmoore at SRI-KL (Bob Moore)
Subject: S and P problem
To: jmc at SAIL, max at SAIL
I am having some trouble understanding the axiomatization of the S and P
problem. It would help to have an intuitive explanation of the predicates,
particularly the Q, B, R, and C predicates.
Bob
-------
∂23-Apr-79 1633 PAT a faculty meeting
has been scheduled for Wednesday, April 25 at 1pm in Serra house conference room
I have put it in your calendar.
∂23-Apr-79 1634 PAT telegram
was from Israel acknowledging your acceptance to go to symposium.
∂23-Apr-79 2153 Feigenbaum at SUMEX-AIM Re: Shrinking 206.
Date: 23 Apr 1979 2136-PST
From: Feigenbaum at SUMEX-AIM
Subject: Re: Shrinking 206.
To: DPB at SU-AI, JMC at SU-AI
In response to the message sent 23 Apr 1979 0918-PST from DPB at SU-AI
What Denny says sounds OK to me.
Ed
-------
∂24-Apr-79 1000 JMC*
Remember Mitman letter.
∂24-Apr-79 1038 ES Mark Thomas Clifton
He was hit by a car a couple of weeks ago, and is now recovering at home.
He would like you call him.
∂24-Apr-79 1112 REP Elephant
Your note reminded me that I have not gotten back to you on the topic of summer
employment. I would like to accept the support with the possibility of jumping
ship if Bob and J's NSF grant comes through. If these terms are acceptable to
you, mail me a quick note to close the deal. I'll keep you posted about the SRI
situatation as information becomes available.
Rich
(who is doing thess things with elephant?)
∂24-Apr-79 1509 MLB
∂23-Apr-79 2245 JMC
For usefulness, NEWDO writeup must be merged with the old.
This will be done when the program is made the "official" version. - Marc
∂24-Apr-79 1602 PAT reservations
are as follows:
May 2
SFO-Dallas AA leaving 7:40am arriving 12:50pm
Dallas-DC AA leaving 1:45pm arriving 5:36pm
May 4
DC-NYC AA506 leaving 12:05am arriving 1:09pm
May 8
NYC-SFO AA 15 leaving 2100 arriving 2356
Reservations at the Marriot Twin Bridge in Washington for 2 nights.
∂24-Apr-79 1947 ES Book for MTC
Could you bring in Hardy & Wright, "Nember Theory" so I can bring to Marc
to read while he is recovering.
∂25-Apr-79 0013 LLW Automatic Distribution Privileges Availability
To: tk at MIT-AI
CC: "@S1[DIS,S1]" at SU-AI
Dear Tom:
Since we all consider you to be a friend (as well as a colleague in
computational crime), we are quite willing to add you to any/all S-1
Project distribution lists of interest to you, thereby saving you the
present trouble of frequent audits and wholesale exports of our logs and
mail files. Mike Farmwald suggested this courtesy a while back, when he
noticed his mail file floating Cambridge-ward, and it seems a reasonable
idea to me this evening (on a not totally dissimilar occasion).
If you would be interested, I'll mail you a complete index of Project
distribution lists, so that you can add yourself on. I also expect that
most everyone on the Project would be willing to flush their old mail
files in your direction (instead of merely into the Infinite Bit Bucket),
when it comes time to clean them out. (Incidentally, JBR remarks that
reciprocity in the form of putting us all on the LISP Machine mailing
lists won't be necessary.)
Warm regards,
Lowell
∂25-Apr-79 0838 DPB
∂24-Apr-79 1636 JMC
To what have you changed the name of the department phone file?
[The department file and LES' file are the same: PEOPLE.DAT[PER,CSD].
-Denny]
∂25-Apr-79 0935 Bmoore at SRI-KL (Bob Moore)
Date: 25 Apr 1979 0934-PST
From: Bmoore at SRI-KL (Bob Moore)
To: jmc at SAIL
I get the general picture, although all the details are not clear. How about
meeting sometime tomorrow (I am on my bicycle today). I have to be at SRI for
lunch, but otherwise my calendar is clear.
Bob
-------
How about meeting with me and Ma at 2pm here tomorrow, e.g. Thursday? This
is tentative till he agrees, and he's not in.
∂25-Apr-79 1311 RWW ARPA proposal
I have done little actual writing, but much thought. What is your opinion
about it being similar to the NSF proposal?
richard
∂25-Apr-79 1715 Bmoore at SRI-KL (Bob Moore)
Date: 25 Apr 1979 1713-PST
From: Bmoore at SRI-KL (Bob Moore)
To: jmc at SAIL
I will plan on 2:00 unless I here otherwise.
Bob
-------
∂26-Apr-79 0916 CSL.OWICKI at SU-SCORE
Date: 26 Apr 1979 0909-PST
From: CSL.OWICKI at SU-SCORE
To: JMC at SU-AI, SSO at SU-AI
In-reply-to: Your message of 25-Apr-79 2156-PST
I feel unjustly accused. I returned one set of corrections; Takasu then
sent me another transcript I had not seen before. I also sent him corrections
for that, but he would probably be receiving them about now.
-------
I should explain that I am not really in this loop. I was talking to
Ed Blum about another matter, and he mentioned it, and I said that with
our great computer system, I could send you a message instantly. Perhaps
some regulator should suggest banning message systems as giving people
too much tendency to get involved in other people's business.
∂26-Apr-79 0934 Bmoore at SRI-KL (Bob Moore)
Date: 26 Apr 1979 0934-PST
From: Bmoore at SRI-KL (Bob Moore)
To: jmc at SAIL
I am still planning to put my thesis out as a joint SRI/SAIL report, if that
is still agreeable to you, but several people have promised to give me
comments on it in the next few weeks, so I am waiting on that. Meanwhile
I have been making batches of 5-10 copies to give to people who have an
immediate interest in it, so if you come across anyone who is really
anxious to read it, I will be happy to send them a copy.
Bob
-------
To tell the truth, I would like one hard bound copy, assuming that M.I.T.
theses are nicely bound like Stanford ones.
∂26-Apr-79 0951 RWW arpa proposal
Do you want me to write up the entire formal reasoning section.
Should I rework the ANALYST section. What should I do about the
other people. I have got stuff from chris and talked to carolyn.
What about Lew, I have not seen him lately. I want to write
something more detailed about data base stuff (a la Filman in
old proposal, who should we say can do it?) Gigina Aiello
mentioned that she was looking for money and she might be a
a good choice. She will be good value for the money. (She said
she only needs $700 a month: about as much as a student). I
should have a draft late tonight.
richard
∂26-Apr-79 1000 JMC*
Call Chandra.
∂26-Apr-79 1003 Mcgoveran at SUMEX-AIM FACULTY MEETING NEXT TUESDAY
Date: 26 Apr 1979 1002-PST
From: Mcgoveran at SUMEX-AIM
Subject: FACULTY MEETING NEXT TUESDAY
To: faculty.list:
Ed has asked that a faculty meeting be set up for next Tuesday, May 1.
The agenda for this meeting is below. I have arranged for this meeting
to be held in Polya 152 (the SCIP conference room) from 2pm until 4pm.
Please let me know whether you will be able to make it to this meeting.
Agenda:
1. Consideration of Asst. Prof. appointment in the Numerical
Analysis area (Golub action item)
2. Consideration of some Research Associate and courtesy appointment
items (Feigenbaum item).
3. Does the plan for allocation of square footage to Research
Associates and Research Assistants in M.J. Hall meet with general
approval? McCarthy has criticized it on some reasonable grounds, so I
need to know how widespread the feeling is.
4. Shall I explore a small "departmental overhead percentage" on
grants and contracts to help pay for common facilities and services?
It is not uncommon at Stanford to do this.
5. Shall we make a requirement that our grad students take a certain
number of courses in other departments (Golub item)?
Mary McGoveran
-------
∂26-Apr-79 1105 Bmoore at SRI-KL (Bob Moore)
Date: 26 Apr 1979 1105-PST
From: Bmoore at SRI-KL (Bob Moore)
To: jmc at SAIL
I don't know how Stanford theses are bound, but the only copies of my thesis
that I have or will have are those that I produce myself. MIT requires
the original copy of the thesis which goes into the archives and is reproduced
on microfiche, plus one additional copy (usually Xerox) which they bind in
a very crummy fashion and put in the library. The only way a student has
of getting copies of his thesis is to make them himself (except for
microfische).
Bob
-------
∂26-Apr-79 1128 CSL.OWICKI at SU-SCORE
Date: 26 Apr 1979 1118-PST
From: CSL.OWICKI at SU-SCORE
To: JMC at SU-AI
In-reply-to: Your message of 26-Apr-79 1048-PST
As an alternative to banning message systems, we could require everyone to use
them. Then Ed and I (or even Takasu and I) could communicate directly.
But banning might be better> I can't decide whether I save time by having
such an easy means of communication or lose time because I use it so much more
than necessary.
-------
∂26-Apr-79 1605 ARK S-1 Meeting
To: S1 at SU-AI
There will be an S-1 meeting in the Serra House conference room on Wed.,
2 May at 2:30 pm.
Arthur
∂26-Apr-79 1652 PAT
∂26-Apr-79 1453 JMC
Let me not forget newbor.le1.
----
You sent it on the 24th...
∂26-Apr-79 2050 MRC via SU-TIP Dialnet debugging
To: LES, JMC
I was attempting to find out some information about the problem
with the Dialnet modems, and have been hampered in this because
an inconsiderate asshole by the name of Hans Moravic decided to
hang up the phone on me. I called up the lab trying to find out
what the status lights on the modem were at the lab so I could
compare it with that at GSB. The person I called asked Hans to take
it and when he found out it was me the sonofabitch hung up.
I don't give a damn about what Hans thinks of me (or for that matter,
what obscure reason it is that he such a grudge against me).
I do not think that such behavior is tolerable.
Les, fyi, I suspect that the reason that Hans is pissed at me is
because his privilege grabbing hacks got broken at one time by
something I did to the system, so he couldn't read everybody's
protected files. For a good time, look at the [HAK,HPM] directory
for all sorts of pirate type programs. Don't expect them to do
anything under your account; they have PPN checks so only certain
users can run them.
∂26-Apr-79 2329 RWW arpa proposal
did you get my message?
Yes, let's talk tomorrow, but I suppose I would welcome a draft for the
whole group. Including Gigina seems reasonable, if you think she
will do good things.
∂27-Apr-79 0942 MAX Sato's paper
Do you have a copy of Sato's paper "A study of Kripke-type models for
some modal logics by Gentzen's sequential method"?
∂27-Apr-79 1221 DCO
To: "@GROUP[1,DCO]" at SU-AI
PROGRAM VERIFICATION AND ANALYSIS SEMINAR
PLACE: ERL 237
TIME: 2:30 p.m., May 1
TITLE: A Simple, Sample Operating System written in PASCAL PLUS
SPEAKER: Richard Karp, Stanford University
ABSTRACT:
We present a simple operating system written in the programming
language Pascal Plus. Pascal Plus is designed for writing verifiable
concurrent systems; in this talk we will deal mainly with describing the
implementation of an operating system written using the language. After a
brief introduction to the toplevel of the operating system (although
nothing with this audience), we discuss two alternatives for the
structuring of a disk file system. It is in this area that efficiency and
language construction seem to clash most; hopefully the audience will be
able to suggest solutions that resolve the clash.
∂27-Apr-79 1401 TOB visitor
John
I have a request from Prof Ulrich Rembold for
an invitation to one of his research associates
for one month here, during May. He wants the
invitation to get funding. He is in a big
hurry for it.
The intent of the visit is to get experience
with AL. They plan to get a copy of AL in
about a year.
I suggest that we send the invitation. We have
lots of room here.
Tom
∂27-Apr-79 1522 Morton at PARC-MAXC AI Review
Date: 27 Apr 1979 3:21 pm (Friday)
From: Morton at PARC-MAXC
Subject: AI Review
To: McCarthy@SU-AI
cc: Bobrow, Morton
Please let us know what your decision is on Derek Partridge's paper entitled "A
Philosophy of 'Wicked' Problem Implementation". A. Newell has rejected the
paper. You only need to comment briefly unless you think the paper should be
accepted.
Thanks
Dianne Morton
(D. Bobrow's secretary)
∂27-Apr-79 1627 MAILER failed mail returned
In processing the following command:
MAIL morton I go along with Newell.
The following message was unsent because of a command error:
∂27-Apr-79 1627 JMC
I go along with Newell.
∂27-Apr-79 1748 PAT restored essays
The following files were restored today to ESS,JMC
∂29-Apr-79 1338 PEG Compatible versions of PCPASC, SOPA, runtimes
To: sherman at CMU-10A
CC: S1 at SU-AI
The following combinations of versions of PCPASC, SOPA, and runtime
routine files for FSIM (the system version) seem to work together (at
least for small test cases):
PCPE04.DMP[DMP,S1], SOPZ11.DMP[DMP,S1], PASSIX.LDI[LIB,S1]
PCPE04.DMP[DMP,S1], SOPS09.DMP[DMP,S1], PASSIX.LDI[LIB,S1]
These combinations of PCPASC and SOPA also "work" (in the sense that
they run to completion) with the PASRUN.LDI[LIB,S1] runtimes, but the
output for characters and character strings looks very strange, due
to the offset of 40(octal) for all characters output by PCPASCs and
SOPAs which include the CHARDIF = 40B modification.
∂29-Apr-79 2153 RWW
To: "@FOL.DIS[FOL,RWW]"
Meeting of FOL group for this week canceled. Next meeting
MAY 10: THURS 2:45-4:00 Bldg 120 room 123; meeting of FOL group
∂29-Apr-79 2224 RWW proposal
do you want me to include the stuff cg is doing? It is actually closer in spirit
to the FOL work than Zohars stuff, but last proposal included it with Zohar.
richard
Yes, include Chris.
∂29-Apr-79 2227 RWW proposal
Let me correct that last message. I am including Chris as part of
formal reasoning. If you badly object let me know.
richard
∂29-Apr-79 2239 RWW proposal
how do you spell ketonen
richard
∂30-Apr-79 1113 PAT proofs for MI9
are on your desk
∂30-Apr-79 1834 Mark Crispin <Admin.MRC at SU-SCORE> Dialnet progress
Date: 30 Apr 1979 0848-PDT
From: Mark Crispin <Admin.MRC at SU-SCORE>
Subject: Dialnet progress
To: JMC at SU-AI
GSB can now dial and take Dialnet calls, and receive packets. There is
a bug (hardware or software) at SAIL's end; SAIL is sending parity which
is screwing me royally. Since it is either Marty's or the hardware
department I've asked ME to look into it. It isn't any of my stuff.
-------
∂30-Apr-79 1835 GUERNYLUSBY at USC-ISI May 3 visit to DARPA
Date: 30 Apr 1979 1111-PDT
From: GUERNYLUSBY at USC-ISI
Subject: May 3 visit to DARPA
To: McCarthy at SU-AI
cc: guernylusby, carlson
I have set up times for you to meet with the
Information Processing Techniques Office staff
on May 3 from 9:00am to l2:00 Noon, break for lunch,
and resume meetings from 1:00 pm to 3:30 pm.
When you arrive on Thursday, see me and I will give you
a more detailed schedule.
If the above times are inconvenient, I will be glad to
work around your schedule.
The IPTO staff are looking forward to your visit Thursday.
Judy Guerny-Lusby
Secretary to Wm. Carlson
-------
∂30-Apr-79 1015 PWM SAIL visit of Dr. Peter Claringbold
To: JMC, TOB, ALS, DCL, RAK, PWM
Dr. Peter Claringbold, Chief of the CSIRO Division of Computing Research,
will be visiting SAIL today.
∂30-Apr-79 0308 RWW
ps. biblo and section called The Formal Reasoning Group
has also not been looked at.
∂30-Apr-79 0306 RWW
John,
I have benn typing for hours. The section on goals I think is
quite good - over 600 lines long. There are 47 of them. I will leave a
copy of the entire thing in your mailbox. The only section NOT really ready
is the section on accomplishments. I have left yours to be filled
in with the one exception where I believe you might not shout enough
about your ideas. Lew's accomplishments also needs to be filled in.
I have not heard from him, I have included his previous section on goals
verbatim. The rest of the section on accomplishments is pretty
rough, I will write more but I'm tired and its late. Feel free to call
and wake me in the morning if you want to know something. I have
an appointment for lunch which should be over at 1, I should be here by 1:30
or so.
I don't know anything about the budget. I have made the presonel section as
follows
//pers John#McCarthy, Richard#Weyhrauch, Lewis#Creary, Lugia#Aiello, S.#Ketonen
sra Christopher#Goad, Carolyn#Talcott, David#Wilkins, Ben#Mozkowski
I have not included the new visiting chinese. We will also have
visitors: Ma Xiwen, Jim Ward, Gianfranco Prini, who are not mentioned
anywhere.
Probably Gigina should be considered as half time $800-$900/mo. Her italian
salery is very low I think. (ps. so is mine: my $24,000 is less than every
recent graduate who has gotten an academic job, and substantially less than
industry saleries (Both SRI(data from Peter Hart) and IBM research(data from
Jim King) would pay me 33-35). Is there anything we can do about that?)
Anyway, I will be in tomorrow afternoon.
richard
∂01-May-79 0009 WD dog
To: JMC, WD
Mut delivered safely. May I stay at your place while your are gone?
What did my poor beast do to offend? Yes, you are welcome to stay.
We will return next Tuesday night.
∂01-May-79 0727 MRC Dialnet progress
To: LES, JMC
GSB successfully received and transmitted Dialnet packets today. This
means that the interrupt-level code (the hard stuff) is basically in
good shape. There are still some JSYS-level problems where the job
can get hung and some loose ends (a lot lot of loose ends in fact),
but with any luck I should be able to push a file over to SAIL tomorrow
morning (with the debugging FTP, not the official one). Some strange
things are happening with the autodialer. We don't yet know whether
or not it is SAIL, our code here, or hardware. We'll keep you posted.
-- Mark
∂01-May-79 0900 morris at PARC-MAXC (repeat) Talks on Semantics, 6/4-6/8
Date: 1 May 1979 8:57 am (Tuesday)
From: morris at PARC-MAXC
Subject: (repeat) Talks on Semantics, 6/4-6/8
To: McCarthy@SAIL, Luckham@SAIL, Oppen @ SAIL, Pollock@SAIL
cc: Scott, Horning, Morris
We are planning to have a series of talks and discussions on Programming
languauge semantics during the week of June 4. You, and any of you interested
colleagues, are invited. We shall keep you posted regarding the schedule and
abstracts, etc..
Jim
∂01-May-79 1008 DCO
To: "@GROUP[1,DCO]" at SU-AI
PROGRAM VERIFICATION AND ANALYSIS SEMINAR
PLACE: ERL 237
TIME: 2:30 p.m., May 8
TITLE: Array Theory -- An Introduction
SPEAKER: Michael Jenkins, Queen's University and IBM Cambridge Scientific Center
ABSTRACT:
Array theory has been developed by More as a model of data. It is
based on a one-sorted universe of nested rectangular arrays with eight
primitive operations. The talk will present a brief summary of array
theory, discuss its potential use as the basis for a programming system
and suggest open areas of research in the relationship of data structures
to programming language design.
∂01-May-79 1348 DEW oral exam
Did you send me a message last night saying that you couldn't come to my ora
exam on May 16 and that I'd have to change it?
If so, could you please call me at 497-1617.
Thanks, Dave
∂01-May-79 2004 LGC Analysis of Feasibility Concept
Please take a look at FEAS.TXT[EP,LGC], to see whether you think the analysis
is worth mentioning in the ARPA proposal. I think it probably is.
∂01-May-79 2129 WD via SU-TIP
To: JMC, WD
Beast entirely inoffensive. Thanks. See you next week.
∂02-May-79 0042 BCM noch ein Witz
mnogouvazhayemyi tovarishch Slonimir Pamyatovski!
A Czech and Polish dog meet on their mutual border. The Polish dog says,
"I'm going to Czechoslovakia to gnaw on a bone." The Czech dog replies,
"And I'm going to Poland to bark."
Sabu
∂02-May-79 0034 BCM politicheskaya shutka dla vas
Czech boy: Daddy, is it true that the Russians have just landed on the moon?
Father: Yes, son.
Boy: And it is true that it's possible to live on the moon?
Father: Not anymore.
Sabu
∂02-May-79 0847 MCM Dialnet progress
To: JMC, LES
We transferred a part of a file today. Unfortunately, we didn't
get the close routines to work and were already grossly overtime,
so don't know if its been fixed yet or not. But we were able
to send essentially the file intact, minus the last buffer's worth.
We also found some WAITS dialing bugs and I (MRC) am going to
fix them soon...hopefully tonight.
∂02-May-79 0917 SSO via SU-SCORE Seminar
To: "@MODAL.DIS[1,SSO]" at SU-AI
"Sometimes" is sometimes "not never"
Lesley Lamport
Thursday, May 3 4:15
ERL 237
A discussion of two models for temporal logic: linear and branching time.
It will be shown that the models are essentially different, in that each can
express some properties that cannot be expressed in the other. The suitability
of each model for use in program verification will be considered.
∂02-May-79 0930 DEW oral
John,
Is 2:15 pm on Tuesday May 15 an ok time for my oral? Dave
∂02-May-79 1118 PAT III board meeting
III just called and are having trouble getting a quorum on the 16th.
The proposed dates for the board meeting are either the 23rd or the 30th
depending on which date they can get a quorum. A memo will be sent out
as soon as they know which date it will be.
∂02-May-79 1205 PAT III board meeting and Dave Wilkins orals
I talked again with III and it looks like the meeting will be on the
23rd. Therefore, Dave Wilkins orals will still be on the 16th.
∂02-May-79 1330 DAN
It took me a while to finally understand what you did last night. I must
congratulate you......I haven't laughed so hard in years. Thanks.....Bud
∂02-May-79 1347 DEW oral exam
To: "@ORAL.DIS[THE,DEW]" at SU-AI
John's meeting for May 16 has been cancelled so the oral will take place
as scheduled on May 16 at 2:15pm in ERL 401. Please mark it on your
calender. I spent a whole day on the phone trying to reschedule everything
before finding that the old day was all right, so I hope nobody changes
their mind again. Thanks, Dave
∂03-May-79 1029 REM at MIT-MC (Robert Elton Maas) Continued STOALL search, progress report
Date: 3 MAY 1979 1328-EDT
From: REM at MIT-MC (Robert Elton Maas)
Subject: Continued STOALL search, progress report
To: PRATT at MIT-AI, JMC at SU-AI, LES at SU-AI
CC: REM at SU-AI, RWG at SU-AI, RWW at SU-AI
Trying to find a programming language which has good storage allocation
and other more-usual programming features all working.... (needed for
data-compression, sort/merge program, several applications which
use sort/merge as subroutine) ....
MacLisp and MainSail come closest to a desirable language among all
the languages available at SU-AI (PASCAL, SAIL, UCI-LISP, FAIL all
moderately inferior; FORTRAN, BASIC very inferior). But both have
serious problems at present. I have been able to make storage
allocation in PLISP work except for array allocation which doesn't
do the right thing about GCMIN but which can be programmed-around,
but a gross bug in random-access file positionning prevents my
sort/merge program from working. Mainsail is more deficient in
storage allocation feedback to user program and user-control, so that
at present there isn't any way to know when memory is filling up;
one method that ought to work clobbers the internal workings of mainsail
so badly that it tries I/O on the wrong channel and gets a UUO error-trap
(need ENTER before OUTPUT). Thus I am caught between the devil and
the deep blue sea, between waiting months or years for a working
maclisp or mainsail (so far I've been waiting 5 months since I first
began encountering these types of problems) or programming things in
very crude ways using assembly language (I've done some of this and the
result is very very slow progress and considerable difficulty debugging
together with a poor user-interface with poor command scanner, plus
algorithms that are very limited in dynamic allocation thus can't do
some tasks at all).
All this information for your curiosity mostly. Some may be a repeat
in different words, some is new.
∂03-May-79 1052 PAT III board meeting
has been confirmed for May 30.
please call iii about board meeting, and if it doesn't conflict with
wilkins oral tell him he needn't change date
∂03-May-79 1226 PAT my absence
To: JMC, LES, HVA, JW, LJH, "@MUSIC.DIS[DOC,MUS]"
I plan to be away from the afternoon of May 9 thru May 15.
I will be on vacation from the 9th thru the 11th and will be
attending a seminar at the Hyatt Union Square in SF on
the 14th and 15th. I will be reachable at home in the evenings
on the 14th and 15th.
∂03-May-79 1517 MRC Dialnet alive!
To: LES at SU-AI, JMC at SU-AI, Yeager at SUMEX-AIM,
G.Ryland at SU-SCORE, REM at SU-AI
Today, Dialnet successfully transferred a non-trivial file (the
DIALNE.MAC source for Tops-20!) from the Stanford Graduate School
of Business to the Aritificial Intelligence Lab a few miles away.
This was over TelCo lines, using 1200/1200 baud full duplex VADIC
modems. The user mode programs used were CRUFT and CSVR, a primitive
FTP protocol suitable mainly for debugging Dialnet implementations.
This is the first Dialnet file transfer between two separate machines.
File transfers with SAIL talking to itself have, of course, been done
for a long while now.
I am hoping to have a Dialnet Tops-20 running in production use at
GSB in a few days (the current version isn't stable enough for
production use).
The current version of Tops-20 Dialnet can be installed in any release 3
or 3a Tops-20 system with or without sources. Besides linking in
DIALNE.REL along with the other files, a few minor edits to STG are
necessary to insert the device. There are four patches which need
to be inserted at POSTLD time (when you put in the BUGCHK/BUGHLT
breakpoints) to bind the necessary hooks between DIALNE and TTFEDV -
I choose this way for the benefit of people who didn't have sources.
There will probably be a few more patches in the release version of
Tops-20 Dialnet for better bulletproofing.
-- Mark
∂03-May-79 1527 PAT
∂03-May-79 1511 JMC
please call iii about board meeting, and if it doesn't conflict with
wilkins oral tell him he needn't change date
------
John, they changed the date because 3 other people couldn't make it on
that date NOT because of your conflict.
∂05-May-79 1009 MRC Tops-20 Dialnet documentation
To: IAZ, LES, JMC
... is in the file DIALNE.DOC[20X,MRC]. I guess you should start thinking
about programming FTP for Tops-20 now.
∂05-May-79 1302 Jrobinson at SRI-KL (Jane Robinson) Margaret Young
Date: 5 May 1979 1304-PDT
From: Jrobinson at SRI-KL (Jane Robinson)
Subject: Margaret Young
To: mccarthy at SU-AI
Hi, John. I've just come from Margaret's. She told me you had been
there and left Vera's tapes. When I left, she and a fellow mountaineer
were listening to them and talking. What this message is about is--
Margaret would like to have the topo of the climbs she and Vera did
in Alaska, around Gruening'(I may have the name wrong or misspelled)
glacier across from Homer. She wants her ashes to be put there. She would
also like to see the slides or any pictures from that trip. Martha Colton
told me all this, and that you had them. I gather there is an uncertain
amount of time, probably not much, that is left for her. The toxins
are growing and, while her spirits seem good, her speech is getting
slurred. Anyway, I volunteered to get this message to you, figuring that
while you might not be easily reachable by telephone, you are sure
to check your mail.
Jane
-------
∂05-May-79 2212 DD at MIT-AI (David Dyer)
Date: 6 MAY 1979 0106-EDT
From: DD at MIT-AI (David Dyer)
To: jmc at SU-AI
This message should be redirected to the systems programmer
responsible for your C-1 data chennel made by foonly inc.
We at III have the next iteration of the design for the C-1 channel.
Be warned that your channel will not work without
some hardware changes should you ever decide to use it to run magtapes.
It also will not run without some microcode changes if you ever connect
more than one controller of any kind to either channel.
To contact me, use DD@AI, which is my mail drop, or you can
call me at Triple-I.
∂06-May-79 1102 REM via AMES-TIP MACLISP TOTALLY BROKEN
To: JMC, RWW
PLISP, the only MacLisp on this system with dynamic allocation of
arrays, is now totally broken. I have complained to RPG and HIC just
a minute ago. I have almost but not quite gotten MainSail to do
what I want with storage allocation, am thrashing with Greg Jirak
about how to work with it. My likely course of action at this point
is to give up on both MacLisp and MainSail for the really tricky
storage allocation stuff, write the toplevel and most workings in
MainSail, and run FAIL-hand-coded routines or SSORT through a PTY
underneath MainSail for the stuff that MainSail can't do right.
Kludgery, but probably the only way to get moving on data compression...
∂06-May-79 1752 MAX GO TO JAPAN
I am trying to find travelling expense from Chinese sources, could you
help me by writing me a invitation letter?
Another question
Mr. Tung, Mr. Tang and I plan to invite you and some other professors
(and the wives) to a dinner party. Please declare which days will be
inappropriate to you between 16 and 22 this month. Thank you.
∂07-May-79 1551 CSL.SALLY at SU-SCORE Space Needs in the Center for Integrated Systems
Date: 7 May 1979 1550-PDT
From: CSL.SALLY at SU-SCORE
Subject: Space Needs in the Center for Integrated Systems
To: JMC at SU-AI
Cc: LES at SU-AI, csd.dbrown at SU-SCORE, csl.sally at SU-SCORE
Marty Hellman has been asked to prepare a document for
John Linvill outlining space needs in the new Center for
Integrated Systems building.
Essentially, this would be a five-year-from-now
projection and should include anticipated office and
laboratory ssquare footage and a statement about whether
the CIS requirements would free up space in another
location.
Marty also needs a projected number of personnel
to be involved in these activities: faculty, research
associates, students, etc.
I have suggested to Ed Feigenbaum that he coordinate
the CSD requirements.
Would you please send your estimates to him for
amalgamation into a general CSD picture.
A hard-copy version of Hellman's request is being
sent to you through the campus ID mail.
Marty would like our replies by 9 May.
Thanks very much.
Sally Burns
Z
-------
∂07-May-79 2335 CSL.OWICKI at SU-SCORE (Susan Owicki) Modal logic seminar
Date: 7 May 1979 2320-PDT
From: CSL.OWICKI at SU-SCORE (Susan Owicki)
Subject: Modal logic seminar
To: all at SU-AI, ark at SU-AI, bth at SU-AI, dcl at SU-AI, dco at SU-AI,
To: dww at SU-AI, jck at SU-AI, jmc at SU-AI, jrg at SU-AI, lcp at SU-AI,
To: mfb at SU-AI, nh at SU-AI, pls at SU-AI, rak at SU-AI, ref at SU-AI,
To: sjf at SU-AI, smg at SU-AI, stt at SU-AI, wop at SU-AI, ttm at SU-AI,
To: wp at SU-AI, zm at SU-AI, bmoore at SRI-KL, elspas at SRI-KL, feiertag at SRI-KL,
To: haas at SRI-KL, jrobinson at SRI-KL, melliar-smith at SRI-KL, moriconi at SRI-KL,
To: nilsson at SRI-KL, waldinger at SRI-KL, bennett at SUMEX-AIM, shoch at PARC-MAXC
From time to time I will send announcements of seminars relating to modal logic.
If you don't want junk mail, let me know and I'll take your name off the list.
-------
∂07-May-79 2336 CSL.OWICKI at SU-SCORE (Susan Owicki) [Shel Finkelstein <SJF at SU-AI>: Modal logic seminar]
Date: 7 May 1979 2322-PDT
From: CSL.OWICKI at SU-SCORE (Susan Owicki)
Subject: [Shel Finkelstein <SJF at SU-AI>: Modal logic seminar]
To: all at SU-AI, ark at SU-AI, bth at SU-AI, dcl at SU-AI, dco at SU-AI,
To: dww at SU-AI, jck at SU-AI, jmc at SU-AI, jrg at SU-AI, lcp at SU-AI,
To: mfb at SU-AI, nh at SU-AI, pls at SU-AI, rak at SU-AI, ref at SU-AI,
To: sjf at SU-AI, smg at SU-AI, stt at SU-AI, wop at SU-AI, ttm at SU-AI,
To: wp at SU-AI, zm at SU-AI, bmoore at SRI-KL, elspas at SRI-KL, feiertag at SRI-KL,
To: haas at SRI-KL, jrobinson at SRI-KL, melliar-smith at SRI-KL, moriconi at SRI-KL,
To: nilsson at SRI-KL, waldinger at SRI-KL, bennett at SUMEX-AIM, shoch at PARC-MAXC
---------------
Mail from SU-AI rcvd at 7-May-79 1424-PDT
Date: 7 May 1979 1422-PDT
From: Shel Finkelstein <SJF at SU-AI>
Subject: Modal logic seminar
To: "@EE385.DIS[1,SSO]" at SU-AI
Concurrent Programming-Modal Logic Seminar
Thursday, May 10, 1978, 4:15pm, ERL 237
Shel Finkelstein will speak on an application of modal logic based on:
ON THE PROBLEM OF INFORMATION INCOMPLETENESS IN DATA BASES
Prof. Witold Lipski, Jr., Polish Academy of Sci., Warsaw
The abstract Lipski used when he gave a talk last term (to a database
audience, so modal logic was soft-pedalled) follows:
"Various approaches to interpreting queries to a data base with
incomplete information are discussed. A simple model of a data
base is introduced, based on attributes which can take values in
specified attribute domains. Information incompleteness means
that instead of having a single value of an attribute, we have a
subset of the attribute domain, which represents our knowledge
that the actual value is one of the values in this subset. This
extends the idea of Codd's null value, corresponding to the case
when this subset is the whole attribute domain. We consider two
different interpretations of the same query. We present an
algorithm of computing the best lower and upper bounds on the
external interpretation of a query logically derivable from the
information contained in the data base."
As you might guess, the interpretations correspond to "necessarily"
and "possibly".
---------------
-------
∂08-May-79 0817 WD
To: JMC, WD
Sarah called last night. She would like you to call her back.
∂08-May-79 0936 Mcgoveran at SUMEX-AIM Faculty Meeting for 5/14
Date: 8 May 1979 0937-PDT
From: Mcgoveran at SUMEX-AIM
Subject: Faculty Meeting for 5/14
To: JGH at SAIL, RWF at SAIL, GHG at SAIL, Feigenbaum,
To: EJM at SAIL, DEK at SAIL, JMC at SAIL, BS at SAIL,
To: DPB at SAIL, CET at SAIL, McGoveran, FB at SAIL,
To: TAR at SAIL, JED at SAIL
Ed Feigenbaum has asked for another faculty meeting, this one to be set for next
Monday, May 14, at 2pm. I shall send another message about the location later
today.
The agenda of this meeting will include three appointments and promotions items:
a. consideration of promotion for Joseph Oliger,
b. reappointment of Gio Wiederhold, and
c. a new appointment in Systems.
Please let me know whether you will be able to attend.
Mary
-------
∂08-May-79 1505 RPG Proposal
To: JMC, CLT
CC: RPG
Last month we discussed briefly the possibility of me writing a
chapter in you Lisp book. Since then I have had several people express
interest in seeing some large document describing in detail the techniques
useful in writing large systems in Lisp, including pattern matching, game playing,
manipulating list structure with `bad' functions, multi-processing etc.
I was thinking that such a book would make a reasonable companion
to yours, one being practical the other theoretical. The two together would
represent an extremely comprehensive guide to Lisp programming.
Writing this second volumn of a set would eliminate the need to
include in an afterthought way any dirty, practical matters where they might
not be appropriate. And the failure of one would not effect the other much.
This summer I will be writing the S1 Lisp with Steele, and notes for
this volume could begin then.
-rpg-
∂08-May-79 1748 Zenon.Pylyshyn at CMU-10A (X320ZP51) ai-Phil conference
Date: 8 May 1979 2044-EDT
From: Zenon.Pylyshyn at CMU-10A (X320ZP51)
Subject: ai-Phil conference
To: rcm @ su-ai
CC: jmc @ su-ai
Message-ID: <08May79 204420 ZP51@CMU-10A>
Is the conference definitely on? I have another name I just thought of:
I have been looking at Shimon Ullman's thesis, and also at some other stuff
he has been writing of a more gemeral nature. I think he would be a good
addition (assuming that Marr can't come -- which at the moment seems a good
assumption since he's going back to England for more drastic tratments).
Also, do you know whether the center will have any provisions for terminals?
I very much would like to continue writing on line through an editor like
EMACS (I have gotten spoiled this year).
Our plans, by the way, are to arrive in Palo Alto in mid July -- about a
month and a half before anyone else is due at the Center. Is anything at all
happening around Stanford in the summer months?
-------
Cheers,
Zenon
∂08-May-79 2035 DCO
To: "@GROUP[1,DCO]" at SU-AI
PROGRAM VERIFICATION AND ANALYSIS SEMINAR
PLACE: ERL 237
TIME: 2:30 p.m., May 15
TITLE: Reliability in a Distributed Computer System for Air Traffic Control
SPEAKER: Dr. Greg Haynes, Texas Instruments
The Discrete Address Beacon System (DABS) is being implemented by
Texas Instruments for the FAA. This system will provide DABS-equipped
aircraft with highly reliable tracking and collision avoidance support.
Each DABS sensor is responsible for an area of radius approximately 60-200
miles, and consists of a network of TI990 computers interfaced to a beacon
and a skin track radar. The 33 computers in this network communicate
through a 528k word global memory. The software schedules interrogations
of aircraft by the radar beacon, computes tracking information for each
aircraft, and if a potential conflict is detected, transmits maneuver
adivsories to the pilots.
Capabilities are also provided for cooperation between adjacent
DABS sensors, to handle aircraft in overlapping areas of responsibility,
and to cover failure of an adjacent sensor.
Program verification has potential for increasing the reliability
of the software. A program is underway to evaluate present verification
technology by applying existing systems to a subset of DABS. This talk
will give an overview of the DABS architecture and describe some important
issues related to verification of this system.
∂08-May-79 2334 Feigenbaum at SUMEX-AIM getting together with John
Date: 8 May 1979 2329-PDT
From: Feigenbaum at SUMEX-AIM
Subject: getting together with John
To: admin.gorin at SCORE
cc: jmc at SAIL, mcgoveran
Ralph, your idea of getting together with John to discuss the plans for moving the
SAIL computer is a good one. John and I have a faculty meeting next Monday afternoon.
Perhaps meeting for 30 to 60 minutes before that would be good. Mary
Mcgoveran (7-4079) can help to arrange it.
Ed
-------
∂09-May-79 1056 PAT IBM San Jose
John Williams called yesterday morning regarding your not showing up
for the seminar. He said that it would be easy to arrange another
time.
∂09-May-79 1057 PAT III board meeting (again!)
III called and the day for the board meeting is still fluctuating. The
current plan is either May 31 or June 1. They will confirm as soon as
they know.
∂09-May-79 1115 PAT Macy's calculator
the new one has come in.
∂09-May-79 1643 PLS Potluck
To: "@POT.DIS[1,PLS]" at SU-AI
POTLUCK!!
Date: Sunday, May 20th.
Time: 4.00 p.m. (or earlier if you like)
Place: My new house - alias Richard Waldinger's - alias 1033 Bryant St.
(between Addison and Lincoln), Palo Alto.
Phone: 327-0457
Friends etc welcome. Try to let me know if you plan to come, and also how
many people you intend bringing.
If you know what you are bringing in the way of food, tell me, so I can steer
those who are foolish enough to ask what they can bring in the right direction.
Peter.
∂10-May-79 0700 JB THESIS MANUSCRIPT.
I finished it tonight and left it on PAT's desk,
with your old corrections also. Thank you for
all your support.
∂10-May-79 1042 Mcgoveran at SUMEX-AIM Meeting with Ed and Ralph
Date: 10 May 1979 1043-PDT
From: Mcgoveran at SUMEX-AIM
Subject: Meeting with Ed and Ralph
To: JMC at SAIL
A meeting has been tentatively set for Monday at 1 in Ed's conference room
to discuss the moving of the SAIL computer. can you attend at this time?
if not, please let me know so an alternate time can be arranged.
Mary
-------
∂10-May-79 1118 LES SAIL move
To: JMC at SU-AI
CC: McGoveran at SUMEX-AIM
Monday is OK with me. We must decide by then if we are not going to move
on June 15 in order to allow a month lead time to various interested parties
(e.g. ARPA, DCA, Ma Bell).
∂10-May-79 1403 MRC Arpanet directory
To: HVA at SU-AI, DEA at SU-AI, TOB at SU-AI, JB at SU-AI,
LES at SU-AI, REF at SU-AI, ME at SU-AI, RAK at SU-AI,
DEK at SU-AI, DCL at SU-AI, ZM at SU-AI, PAM at SU-AI,
JMC at SU-AI, DCO at SU-AI, SSO at SU-AI, ALS at SU-AI,
CJS at SU-AI, RWW at SU-AI, TW at SU-AI, JW at SU-AI, PAT at SU-AI
You are currently listed in the Arpanet directory. Please check your
entry in the file ARPA.DIR[NET,MRC] and verify that the data on you is
correct. If you wish the data to be modified, deleted, or moved to
another site's membership list, send me a message; in the case of SCORE
I can do it myself.
∂10-May-79 1528 Mcgoveran at SUMEX-AIM Visitng Scholars from China
Date: 10 May 1979 1529-PDT
From: Mcgoveran at SUMEX-AIM
Subject: Visitng Scholars from China
To: DEK at SAIL, JMC at SAIL
Dean Carnochan sent a memo around recently to publicize a meeting which will
be held from 1:30 to 3:30 on May 16 in Room 271 in Tresidder Union.
From the dealings I know you have had with the Chinese, I wondered whether you
had also been sent this memo. If you haven't and are interested, let me know,
and I'll send you a copy.
Mary
-------
∂10-May-79 1659 Mcgoveran at SUMEX-AIM Faculty Meeting on Monday, May 14
Date: 10 May 1979 1659-PDT
From: Mcgoveran at SUMEX-AIM
Subject: Faculty Meeting on Monday, May 14
To: Feigenbaum, Lenat, Wiederhold, Winograd at PARC, Liebes,
To: DPB at SAIL, RWF at SAIL, GHG at SAIL, JGH at SAIL,
To: DEK at SAIL, JMC at SAIL, EJM at SAIL, FB at SAIL,
To: TAR at SAIL, BGB at SAIL, CCG at SAIL, JEO at SAIL,
To: DBW at SAIL, ACY at SAIL, FFY at SAIL, LES at SAIL,
To: JLH at SAIL, KUM at SAIL, SSO at SAIL, FAT at SAIL,
To: WVC at SAIL, BS at SAIL, CET at SAIL, JRG at SAIL,
To: JED at SAIL
From: Ed Feigenbaum
Subject: Faculty meeting on Monday, May 14
There will be two faculty meetings (in quick succession) on Monday.
Beginning at 3:30pm in Room 252 (big second floor conference room),
the entire faculty and student representatives will meet to discuss
the filling of our appointment slot in the systems area. Other small
matters may be discussed.
Immediately thereafter, the Full,Associate, and Adjunct Professors will
meet to discuss an emerging tenure promotion case.
I aim to hold each meeting to 45 minutes if possible.
Please attend if you possibly can.
-------
∂11-May-79 1054 DEW Hans Berliner's visit
John,
Hans is flying out this weekend so he'll be here Monday and Tuesday before
my oral if you want to talk to him. He's leaving Wednesday night after my oral.
Dave
∂11-May-79 1141 REP Summer Job
Bob and J's NSF grant just came in so I will be working for them this
Summer. Thank you for setting up a Summer job for me at the lab with no
strings attached, it took quite a bit of the tension off. I am interested in
what you and Corky will be doing and will keep in touch. Thanks again.....
Rich
∂11-May-79 1549 DEW
I'll have Hans mail you a message on sunday setting up co-ordinates.
∂11-May-79 1550 DEW room change for oral exam
To: lenat at SUMEX-AIM, buchanan at SUMEX-AIM, JMC at SU-AI
To make use of the new facilities, Carolyn has changed my oral exam from
ERL 401 to Margaret Jacks 252. (It is still this coming Wednesday at 2:15.)
Please don't go to the wrong room. Dave
∂11-May-79 1800 Raphael at SRI-KL (Bert Raphael) OPEN HICE
Date: 11 May 1979 1802-PDT
From: Raphael at SRI-KL (Bert Raphael)
Subject: OPEN HICE
To: Friends:
(That's two days of open house!)
We're 90% unpacked, but the rest may take all year, so we're not waiting.
Please come share some wine, snacks, and conversation at our new address
12820 Viscaino Road
Los Altos Hills, CA 94022
on Saturday or Sunday, May 19 & 20 (come either day).
Bert & Anne Raphael
DIRECTIONS: From Arastradero Road go South on Fremont (between 280 and
Foothill Expwy.), at second stop sign keep right on Concepcion, turn right
on Viscaino Drive, right on Viscaino Road, watch the mailbox numbers on
the right, and beware of driveway congestion!
-------
∂11-May-79 1804 Raphael at SRI-KL (Bert Raphael) OPEN HICE Times
Date: 11 May 1979 1805-PDT
From: Raphael at SRI-KL (Bert Raphael)
Subject: OPEN HICE Times
To: Friends:
3:00 to 7:00PM both days.
-------
∂11-May-79 1951 RWW prolegamena paper
To: CLT, JMC
thought you might be interested that it has been officially accepted
for publication in the AI journal.
richard
∂12-May-79 1613 DEW message from Hans Berliner to arrange meeting
To: JMC at SU-AI, buchanan at SUMEX-AIM
I understand from Dave that you would like to see me. I do have a couple
of appointments, but basically all day Monday is schedulable and late
Tuesday would be OK too. Please contact me on this account. I will be in
at the AI lab early Monday morning too.
Hans Berliner
How about 11am Monday at Lab + Lunch. If you don't have a car, meeting
somewhere else is ok too.
∂13-May-79 1503 DEW
John,
I'll meet you here at the lab at 11 on Monday. Hans
∂13-May-79 1707 RWW dinner tonight
I forgot if you said 6:00 or 7:00
richard
∂14-May-79 1025 Bmoore at SRI-KL (Bob Moore) terminals at CASBS
Date: 14 May 1979 1018-PDT
From: Bmoore at SRI-KL (Bob Moore)
Subject: terminals at CASBS
To: jmc at SAIL
The last message we got from Zenon reminded me about the question of terminals
at the Center. Have you done any looking into this? Could we get the
Center to pay for phone lines, perhaps in exchange for less secretarial
support? Could we borrow some terminals from SAIL? If you have any ideas
about this I would like to know.
Bob
-------
I think we could borrow a few from SAIL, and presumably others could be
rented. I called Gardner Lindzey about it, and he said that you should
talk to Carol Trainer about technical matters such as phone lines. He
pleaded poverty, and I forgot about the secretarial saving argument. I
said that the AI Lab could provide the computer time. Those individuals
who can borrow terminals from their organizations should do so. It would
be good if you could work up a cost estimate for doing things right including
providing terminals for the philosophers.
∂14-May-79 1318 DPB Jacks Hall third floor
To: JMC, DEK
My current model of the space situation on the third floor is that
there are three "groups" of people and offices: NA, DEK, and JMC.
Should I instead consider that the DEK and JMC groups will combine
for space allocation purposes?
Current general model:
JMC offices are 302, 362, 360, 358, 356, 353, 351, 336, and about half
of 352.
DEK offices are 322, 324, 325, 326, 328, 330, 332, 334, 338, 340,
341, 342, and about half of 352.
I assigned Gacs to 351 (DEK's visitor in JMC space) for convenience.
The three PRC visitors are assigned (almost randomly) as follows:
Ma - 353 JMC, Tung - 351 JMC, and Tang - 322 DEK.
Don recently sent me John Gill, who would like a spot in Jacks as an
affiliated faculty member. Makes sense. It was suggested that the
Gacs office is (or soon will be) available. I hesitate to use
what is (sort of) JMC's spot for a "permanent" person. (I didn't feel
so bad about Gacs since he was temporary.)
If DEK and JMC agree to pool their space resources, we will probably
have more flexibility. On the other hand, maybe each of you want to
keep tight control over group space.
Comments?
-Denny
I'll get back to you after a bit, but I can't agree to anything now.
∂14-May-79 1447 ARK S-1 Meeting
To: S1 at SU-AI
I think there will be an S-1 meeting on Wednesday, 16 May at 2:30 pm in
Margaret Jacks Hall room 301.
Arthur
∂14-May-79 1456 100 : ljh
Your daughter called. Please send her a check. (Call her if you don't know what
this message means).
∂14-May-79 1531 Engelmore at SUMEX-AIM Discussion of your proposal
Date: 14 May 1979 1532-PDT
From: Engelmore at SUMEX-AIM
Subject: Discussion of your proposal
To: JMC at SU-AI
I have numerous questions and comments that have arisen from reading
your draft proposal. I think it would be of benefit to me if I spent a day
or so with you and your colleagues and discussed the various program areas
in some detail. I could arrange to visit on the 25th, 29th or 30th. Are
any of those days acceptable to you, Weyhrauch, Creary, et al.? Please
let me know as soon as possible. Thanks,
Bob
-------
∂14-May-79 1751 LGC Engelmore's Visit
May 25, 29 or 30 would be fine with me. Would you like me to prepare a
presentation of some sort?
∂14-May-79 1900 LGC CBCL.PRO
The basic idea seems sound and interesting. I haven't seen Appendix B yet
(does it now exist?), but on the basis of what I have seen, I think I would
definitely be interested in contributing to the research effort on such a
project. What percentage of my time do you think it would be appropriate
and/or wise for me to consider spending on it, as compared with the
representation and reasoning work connected with the problem-solver
project?
One aspect of the proposal that appears to need strengthening is
the general characterization of the semantics to be provided for CBCL and
its role in the construction of production software. It isn't clear how a
commercial software house, for example, would use such semantics in the
construction of programs based on CBCL. How would the proposed semantics
relate to the methods of writing program specifications that are currently
in everyday use by software developers? One very down-to-earth form that
such a semantics might take (well short of a fully automatic programming
system) would be a compiler that would take high-level specifications in
CBCL and automatically produce program specifications in a form that would
be immediately intelligible to and usable by a typical commercial software
development group. (I assume that such people would not find a highly
theoretical formulation of the semantics very useful.)
Do you see the CBCL project as a possible addition to the ARPA Proposal?
P.S. Is there some easy way to read CBCL.XGP online? I thought I saw
mentioned somewhere a program for such a purpose, but I can't now
find any such reference.
To answer only your last query, the program is XGPSYN, but it can onlyy
be used on data disks and only when at least three channels are available.
The files CBCL[F75,JMC] is intended as the appendix in question. See also
files.pro[s79,jmc] about which comments are solicited. I am requesting
a research associate for each project.
∂15-May-79 1150 Elschlager at SUMEX-AIM CONCEPTS AS OBJECTS
Date: 15 May 1979 1151-PDT
From: Elschlager at SUMEX-AIM
Subject: CONCEPTS AS OBJECTS
To: JMC at SAIL
A referee for a paer that I recently
wrote suggested that I find out about your development of
"concepts as objects". Is there something written up on your development?
If there is, could I look at it or zerox it sometime, or if there
isn't, perhaps you could explaine some of it to me ? If you did want to talk,
if you suggest a time during Wednesday afternoon or any
time duringThursday or Friday, I can drop by then.
-------
Look at Stanford AI Memo 325 after which we can talk if you want.
∂15-May-79 1537 CET Help -- Colloquia Speakers Needed
To: "@FAC.DIS[1,DPB]" at SU-AI
Colloquia speakers needed for 5/29 and 6/5. Please send suggestions to
DPB@sail. Any volunteers???
∂15-May-79 1654 CLT concert
There is a baroque flute concert tonight at 8pm
∂15-May-79 1802 DEW Remember my oral!
To: JMC at SU-AI, lenat at SUMEX-AIM
Wednesday, 2:15, Jacks 252. Thanks
∂16-May-79 0942 DCO
To: "@GROUP[1,DCO]" at SU-AI
PROGRAM VERIFICATION AND ANALYSIS SEMINAR
PLACE: ERL 237
TIME: 2:30 p.m., May 22
TITLE: Reasoning in Equational Theories
SPEAKER: Derek C. Oppen, Stanford University
We consider the problem of determining whether a boolean
combination of equalities and disequalities between terms in a first order
logic follows from a given set of equations between terms with free
variables. The problem is undecidable in general, but is of sufficient
importance in practice that much research has been devoted to finding ways
that "work well in practice" to handle the problem. Most research has
concentrated on "rewrite rules", in which the set of identities are
treated as one-way replacement rules -- they directionally determine
substitution of equals for equals. Rewrite rule systems suffer from two
major drawbacks in practice -- they depend on the rules having the
Church-Rosser property and they cannot directly handle identities, such as
the commutative rule x * y = y * x, which are non-terminating. We present
an alternative approach which manipulates equivalence classes rather than
terms; it has the advantage that it does not assume the Church-Rosser
property and handles identities such as x * y = y * x without diverging.
The procedure will diverge on certain identities; a simple heuristic is
described that appears to work well. Some (relatively negative) results
on the complexity of the algorithms used will be given.
We will then discuss the problem of handling conditional equations
of the form f(x) = if p(x) then g(x) else h(x), of handling free existential
variables as in f(x) = if p(a,x) then g(a,x) else h(a,x), and of handling
more general styles of axioms.
∂16-May-79 1110 PAT
∂02-May-79 0837 PW
Please don't send out any Elephant papers until you hear from John.
(this was a message by phone from CLT at 8:30am on wed.)
--------
John, is this message still true?
∂16-May-79 1115 PAT
∂10-May-79 0653 JB THESIS MANUSCRIPT.
I placed my manuscript on your desk. I have
already talked to John and he wants to sign me
off this quarter. Would you please see that h
he gets the manuscript today?
------
John, did this get to you???
∂16-May-79 1311 MRC reload this morning
To: ROB, JMC
Was Marty called before you reloaded? I certainly wasn't. From the
CTY log, it looked as though the problem was perfectly fixable. Granted
that it did make the all-time uptime record, but I am distressed that it
died without a definite death certificate signed by ME or myself. It was
almost 11 in the morning, a perfectly reasonable time to call either of
us at home.
Sorry about that. I supposed ROB knew there was nothing to do but reload.
∂16-May-79 1420 TW via SU-SCORE Summer guest
To: LES at SU-AI, JMC at SU-AI
CC: joloboff at PARC-MAXC2
Vania Joloboff is a French natural language researcher who has
been spending this year working with Martin Kay here at Xerox. He
would like to spend some time at the AI lab before he goes back, so he
can get to know what is going on there. Is it possible for him to have an office
from the middle of June until the move? I am glad to vouch for him,
put him under my NL category, etc.
Thanks --terry
∂16-May-79 1519 Engelmore at SUMEX-AIM Visit to SAIL
Date: 16 May 1979 1520-PDT
From: Engelmore at SUMEX-AIM
Subject: Visit to SAIL
To: JMC at SU-AI
If you have no objections I will spend Wednesday the 30th at SAIL.
Bob
-------
Wednesday the 30th will be fine.
Bob Engelmore from ARPA will be here on Wednesday the 30th all day.
∂17-May-79 1119 CET MTC Qual
To: JMC, PAT, CHT
John,
An MTC Qual hasn't been given in 18 months. Carolyn Talcott is due to
take it and there are probably others. Have any plans been made? Do
you want me to start a sign up list?
-Carolyn
Please start a sign-up list. I have been postponing it until Manna
returns. Therefore, we will give it in the Fall. Most likely it
will be oral in the late Fall, so that we can ask for some additional
reading.
∂17-May-79 1220 TW
Sloman, Aaron, %eThe Computer Revolution in Philosophy,%* ??, ??, 1979
Can you give me more details on this? Are any copies actually available
here now? Also, do you have a complete list of who will be at the center
when next year? Thanks --terry
Sloman's book is published by Humanities Press. I have a copy and
so does the Stanford Library (Meyer Basement). I can lend you my
copy if theirs is out. The seven present for the year include
myself, Pat Hayes, Bob Moore, Marvin Minsky, Daniel Dennett of Tufts,
John Haugeland of U. Pittsburgh, and Zenon Pylyshyn. We plan to
hold a two day meeting with a long list of invitees but not supporting
travel, so what we'll get is random. You are one of the invitees.
I hope to get Sloman for some time, but that isn't certain.
∂17-May-79 1348 PAT
To: JMC
CC: LES, HVA
17 May 1979
Dear John:
I have talked with Les regarding his thoughts on the following, but I
would like, at this point, to have your view of the world. I realize that
you prefer to remain uninvolved, but it would help to have some input from
you so that some sort of framework for the future can be formed as a basis
for making decisions. I am only asking for your opinions, not concrete
statements.
Aside from my duties as your secretary at the lab, and because you are
fairly self-sufficient, my job has evolved over the last few years as to
its scope.
This has included having an integral part in the decision making process
of the lab in regard to policies of publications (also CS dept) and
administrative and personnel procedures in the front office. I have had a
lot of freedom in handling things my own way and have generally done what
I could, with occasional consultation with Les, in making the lab run
smoothly in areas I felt needed attention.
I am now, as the move approaches, and in view of some past events,
becoming slightly apprehensive as to what my role at Margaret Jacks, and
that of the other administrative staff of the lab, will be. I understand
that compromises and changes will have to be made. I also realize that
there will be overlaps in positions and differences in policies. But I
fear that assumptions regarding the configuration of the merged department
(both within the lab and within the department) are being made without
consideration and consultation of the people involved and without
information or regard as to what people's current positions are and what
the needs of a merged department will be in regard to clerical support.
With regards to people and responsibilities, I cite two events
illustrating my point:
1. A meeting occurred several months ago, at my suggestion, to facilitate
the merging of office related supplies and procedures (i.e. general
supplies, Xerox, coffee room, mailing, etc.). The people attending were
Hersche, Les, Betty Scott and myself. These things were resolved fairly
easily and satisfactorily. Also on the agenda was the position of
receptionist, which we discussed. The suggestion was made, by Betty, that
the department was thinking that Leslie should fill that position on the
basis of that being her position at the lab. I brought out the fact that
the receptionist will basically greet people, answer phones, take care of
mail, etc. (usually classified Secretary I level) and that Leslie
(classified Secretary II Specialist) had many more responsibilities at
lab. In particular, running the publications program for me and editing
and pubbing papers on the system. To put Leslie in the slot of
receptionist would be a demotion and a waste of the training and skills
she has. (I do feel that the receptionist will make use of the system, but
in a more limited way.) The system work that Leslie does for various
people at the lab serves a valuable function in the productivity of the
lab and should continue to be a full time position (the department does
not currently support this kind of position). A point in case - Zohar's
requirements keep both of us busy on the system with he is around. Les
agreed with my thinking.
Apparently our words fell on deaf ears. The department has repeatedly
refused to assign Leslie an office in Margaret Jacks. I have pointed this
out to Les and he has called Denny Brown twice that I know of, without any
results. In addition, the department has decided not to hire a
receptionist until the end of the summer. One of their secretaries is
"filling" in until then "because the summer is slow". Leslie is listed on
the current map under unplaced ?CSD, not ?AI. There are secretarial
offices that are currently unassigned on the present map configuration.
I fear the result of all this will be that Leslie will, by default, become
the "receptionist" when we move to campus. I also fear that this is game
playing on the part of the department and an injustice to Leslie in
regards to her position at the lab. It has also been demoralizing to her
to continually be unrecognized by the department when she is an integral
part of our staff.
2. We also discussed the fact that certain responsibilities would have to
be realigned to avoid duplication of effort. In particular, Betty felt
that the merged CS publications would become a one person job (currently
shared by Connie Stanley and myself) with the responsibility falling on
Connie. I mentioned that would take away part of my current
responsibilities (which I am not opposed, there is plenty of work to go
around) and Betty offhandedly said "then you will have time to be a
secretary for four or five more faculty!!!"
I realize that the lab as a separate entity will cease to exist once we
move to campus, but I resist the implication that we will fill in the
"holes" in the department staffing situation and have to conform to THEIR
perception of what our responsibilities and administrative policies should
be. (Who decides what my job is? Les, Betty, Denny, Ed, Hersche or you?)
With regard to policy that will directly affect us I cite a few examples:
1. We don't have a particularly hierarchical world here at the
lab. I fear the opposite is true in the department.
2. The department has administrative policies regarding support
staff that are basically different in philosophy from ours.
(i.e. Betty Scott is in charge of ALL administrative support
in the department regardless of which group they work in, the
AI lab is the only exception. She also maintains that work
be performed by these people between the hours of 8-5 only.)
3. Up until now, basic decisions regarding each group have been
made autonomously within that group, due to location.
The decision making process regarding different groups within
the merged department appears arbitrary and not at all clearly
or procedurally defined.
I feel some thought should be put into merging the administrative staffs
of the lab and the department in a fair and efficient way. At the moment
it is unclear to me just what the configuration of the merged department
will be and I don't have any indication as to what anyone is thinking
about it. (Will the lab administration remain autonomous, or will we
slide into roles in the department administrative structure, under their
control?)
I am aware that some things can't be resolved and won't be apparent until
we move. But I think we can reasonably anticipate some of the problems
and deal with them intelligently instead of arbitrarily or by default.
With regards, Patte
Possible major conflicts and questions
personnel replacement, policies, and supervising
administration of budgets/clearing accounts/petty cash funds
will we still authorize or control AI computer access/computer equipment
purchasing/supplies/maintenance?
distribution of clerical work/priorities
BASICALLY - who does work for whom and who authorizes what?
i.e. lab people we currently support, Zohar, Richard Weyhrauch,
Tom Binford, Derek Oppen, David Luckham, will be scattered
throughout the building - what happens to the current structure
of performing work for these people?
∂17-May-79 1414 TW AI qual
To: "@AIFAC.DIS[1,TW]" at SU-AI
The AI qual reading list has been made official and details given to the
students. It is on the AI lab system as AIQUAL.TW[UP,DOC] and there will
soon be hardcopy and teletype versions available as well. Thanks to all
of you who sent in additions and corrections. I don't know how many
students will be taking the exam yet. I plan to set up oral exams with
three examiners per student, which means I will be asking each of you to
be in on 2 or 3 exams. The date is set for the first two weeks of June.
Would you like to try doing them all on one day like we did last year or
to spread them out according to people's individual schedules? In either
case, please let me know when you are available during the first two weeks
of June (including Saturdays). Thanks -terry
∂17-May-79 1431 HVA Patte's Memo to JMC
To: PAT
CC: JMC, LES, HVA
Patte: Thank you for copying me--I've'only read through once, and very quickly,
but will read again at more leisurely pace. It's good to have your thoughtful
questions and impressions set forth in this fashion so we can all keep them in
mind. Thank you.
∂17-May-79 1544 MAX Information about dinner
When: May 20, Sunday. 6:00 p.m.
Where: 10H, Quillen. Mr.Tang's room.
Welcome!
∂17-May-79 1612 PAT
∂13-May-79 2149 JMC Tang
I have heard that Tang expects to be able to come, so you might tickle
the I-Center for a DSP-66 for him.
------
John, The DSP asks for the dates of his visit? Do you happen to know? Patte
Well, it's still indefinite, but I suppose it will be for about a year,
but it's hard to specify a starting date till he gets through the
Chinese bureaucracy.
∂18-May-79 1637 Mcgoveran at SUMEX-AIM Status of PRC Scholars: Summer Quarter and Beyond
Date: 18 May 1979 1638-PDT
From: Mcgoveran at SUMEX-AIM
Subject: Status of PRC Scholars: Summer Quarter and Beyond
To: JMC at SAIL
cc: McGoveran
John,
A two-page memo has arrived from the US-China Relations Program with the
above title.
1. Would you like to have a copy?
2. Would you like to be kept apprised of the China situation?
Mary
-------
Alas, yes to both questions.
∂18-May-79 1647 CET Bill Miller
To: JMC, PAT
John,
Ed suggested that I ask you to introduce Bill Miller at the CS Colloq. on
Tuesday. Will you be available?
Carolyn
Yes, I will introduce him.
∂18-May-79 1656 CET
∂18-May-79 1654 JMC
Yes, I will introduce him.
//
Thanks, John.
∂18-May-79 2332 ME
∂18-May-79 2330 JMC
Do we need the second disk controller when musicians separate?
ME - Yes, also the second channel.
∂21-May-79 1106 PN Help for Users
To: S1 at SU-AI
Due to popular demand, I have created a system of DO files on [DO,S1] to
make it easier to compile and run Pascal and Fortran programs at the AI
lab using the Pcode/Ucode systems (this will also incorporate the
optimizer and interpreter). It it also useful for doing part of the
translation (say, Fortran to Pcode) and looking at the result.
For more details, see page 4 of INDEX[DOC,S1].
∂21-May-79 1302 DCL Delivery of XGP output to Jacks Hall
To: JMC, LES
This idea is not working too well. It is encouraging remote use of the XGP
and jumbling of unattended output. Consequently at the moment it is more
a waste of our resources than an service.
-David
∂21-May-79 1917 MRC Dialnet
Well, I seem to have fixed dialing up SAIL, but transfers of SAIL → GSB
still aren't right. GSB → SAIL works just fine. I think it's a problem
at the GSB end. Will look at it Saturday when I can get downtime there.
∂21-May-79 2026 Zenon.Pylyshyn at CMU-10A (X320ZP51) Access to conputer
Date: 21 May 1979 2114-EDT
From: Zenon.Pylyshyn at CMU-10A (X320ZP51)
Subject: Access to conputer
To: JMC @ SU-AI
Message-ID: <21May79 211456 ZP51@CMU-10A>
Since I have gotten into the comfortable habit of doing all my writing
on the machine using the very nice EMACS editor, I've been wondering
whether I shall be able to continue this mode of operating when I get to
the Center. Will I be able to get access to the SAIL machine? Does it
support EMACS or a comparable editor? Do you think it will be possible
to get the Center to supply me with a 1200 baud video terminal and a
Vadic coupler?
I guess I should be addressing some of these questions to Preston, but
the issue of access to conpute power and high speed line is one that you
are in a position to answer for the Lab. It would really make a big
difference to my productivity if I could get to a good text system.
By the way I shall be arriving in Palo Alto in mid July. Is there
anything interesting happening during the summer. Mrs. Duignan informs
me that most fellows at the Center will not be arriving until Sept.
Cheers,
Zenon
We will make SAIL available. It supports an editor called E that is
comparable to but not compatible with EMACS. I have spoken to Gardner
Lindzey about terminals, and he doesn't seem inclined to supply them.
However, we may be able to scrounge something. I don't about anything
special going on this summer and will be away from mid July till the
beginning of September. - John
∂22-May-79 0119 MRC
∂21-May-79 2351 JMC
I thought you could work at GSB late any night.
MRC - They got a bit pissed off that week when I was taking it down
every night. They have 24-hour users these days.
∂22-May-79 0128 LLW
To: FB at SU-AI
CC: LLW at SU-AI, LES at SU-AI, JLH at SU-AI, JMC at SU-AI
∂21-May-79 2227 FB via SU-SCORE personnel
To: LCW, LLW, TM
We have two RA's on our S-1 contract that make us run out of money with the
current load. I have proposed that we flush MLB so that we can pay these
more useful people (Lam and Eicelburger). I am asking Les to figure out if
there is then enough left over for Brooks (about whom I have feelings whatever).
It does seem that it would be better if Farmwald didn't carry on as though he
were in charge of this budget (like asking Les to put Brooks on). Comments?
Forest
[Dear Forest:
There seems to considerable unanimity about flushing MLB, and I presume
that this has been done. PMF spoke to both MLB and LES about this after
securing my concurrence, as MLB was originally put on the sub-contract for
the present funding period to do ABOX-related work for him, but was doing
so at an unacceptably slow pace. Independently, GLS had asked me for
ROD's help on the S-1 LISP effort this Summer and, after I had agreed to
this, PMF agreed to ask LES to make the necessary arrangments re the SAIL
payroll. To the best of my knowledge, PMF took no independent actions in
this area; the net effect of the actions he has requested of LES has been
to terminate an unacceptably unproductive worker originally assigned to
support him, and to substitute in this slot someone needed to work with
GLS on a relatively high-priority project.
I've never heard of Lam or Eicelburger before, or what they're scoped to
do, to supplement the scheduled Pascal* development effort. I'm perfectly
amenable to their joining the effort upon your recommendation, once I
understand what they're going to be doing in addition to the proposed
effort, and once you, LES and I figure out if there's sufficient funds to
pay them (or get supplemental funding added to take care of their costs).
Please let me know what you would like to arrange for in this case.
On a somewhat related matter, when during the next couple of weeks (e.g.,
toward the end of next week, or the latter part of the following week)
would it be convenient for you and yours to get together with a few of us
for an afternoon somewhere at/around Stanford to review what has been
accomplished during the Spring, what is planned for the Summer, and what
the proposal for next academic year should look like?
Lowell
∂22-May-79 2137 Zenon.Pylyshyn at CMU-10A (X320ZP51) Terminals etc
Date: 23 May 1979 0034-EDT
From: Zenon.Pylyshyn at CMU-10A (X320ZP51)
Subject: Terminals etc
To: John McCarthy <JMC at SU-AI>
Message-ID: <23May79 003408 ZP51@CMU-10A>
In-Reply-To: John McCarthy's message of 22 May 79 01:49
Thanks for the note. I will press Lindzey myself, it case it does any good.
I'm sorry you will be away until June. Would you authorise an account for
me on SAIL before you depart so I can begin to familiarize myself with the
software (maybe I could even start while I'm out here).
I hope there's enough fuel around this summer so we are not confined to
our house!
------
Zenon
∂22-May-79 2248 DCL Lab ARPA budget
To: LES
CC: JMC
Les, We need some projection of what to budget for lab computer use
for then next two years.
I think our deadline for getting everything to ARPA is June 1st.?
-David
∂23-May-79 0038 HPM
you must have realized by now the science cover would be titled
"The General and the Professor" rather than "The Professor and the
Graduate Student"
∂23-May-79 0042 HPM
the general provides the azimuth, the professor the elevation and sensing.
∂23-May-79 0926 FB VAX OS
To: EJG, JMC
I have been following the CMU file on VOX OS discussions. I could put in some
comments on VAX DEMOS as Erik suggested but I have been somewhat depressed by
much of the discussion. The technical criticisms are mostly correct but some
people are mostly upset about not having their favorite whisle rather than
about basic structural problems. UNIX is a simple minded system, lovely for
a small number of friendly users. VMS is an ad hoc nightmare (except for the
scheduling and paging, which works quite well provided you are not trying to
run you system with too little main memory, like most of those people are).
[On VMS, one should have 2 megabytes; these half megabyte systems are ridulous.]
I am interested in any thoughts or recommendations you two might have to me on this
subject.
Forest
∂23-May-79 0934 FB VAXes
I would like to talk with you soon about your thoughts and plans on VAXes and
design systems. Since I will be out of town for the rest of this week, what
about this weekend or next week? I will probably be out at the Lab Saturday.
Saturday is fine. How about 11am at the Lab. Yes, please on display.
∂23-May-79 0937 FB displays
The first board of our high speed, high resolution display went out for
wire-wrap this week. (It is three boards with substantial local processing
power.) Maybe you would like more information on that?
∂23-May-79 1335 Bennett at SUMEX-AIM China
Date: 23 May 1979 1328-PDT
From: Bennett at SUMEX-AIM
Subject: China
To: jmc at SAIL
Prof. McCarthy,
Avron Barr and I are planning to attend IJCAI and then continue
around-the-world for the following 80 days or so. A large portion of this
trip will be spent in Southeast Asia; approximately 9-10 days will be
spent in Hong Kong immediately after the conference. We had heard that
entry into the People's Republic was possible but that arrangements
had to be made on arrival to visit, say, Canton. We were wondering if
you had more information about entry and especially Canton (i.e., universities,
museums, etc.). Thank you,
James Bennett
-------
∂23-May-79 1453 DCO
To: "@GROUP[1,DCO]" at SU-AI
PROGRAM VERIFICATION AND ANALYSIS SEMINAR
PLACE: ERL 237
TIME: 2:30 p.m., May 29
TITLE: On the Development of Correct Programs with Documentation
SPEAKER: Andrzej Blikle, UC Berkeley and the Polish Academy of Sciences
The paper presents a method for the systematic development
of correct programs. A program is called correct if it is partially correct
with respect to given pre- and post- conditions and if it neither loops
indefinitely nor aborts. The requirement of non-abortion makes our correctness
stronger than the so-called total correctness which is usually understood as
partial correctness plus non-looping. In the described method programs are
developed and transformed together with their documentation. The
documentation consists of a precondition, a postcondition and a set of
assertions. The assertions are chosen in such a way that they may be used
in the correctness proof of the program. This provides an adequate
description of the algorithm and may also be useful in program testing. The
rules of program derivation are sound, that is, if applied to correct programs
they yield correct programs.
∂23-May-79 1640 REM via SU-TIP Status of data-compaction R+D
To: RWW, JMC
The bug in PLISP whereby FILEPOS returns a result module 200, thus making
disk random-access useless, remains broken even now, but RPG thinks he may
have finally found the bad code so maybe soon it'll be fixed and I can
return to a choice between MacLisp and MainSail...
Meanwhile I have written a brand-new adaptive data-compression program,
similar to IC2.FAI but much more flexible because it's written in
MainSail instead of FAIL so it's not only easier to do scanning/parsing
and complicated storage allocaton (providing I carefully avoid the known
bugs in MainSail) but much faster to develop new code so within a few days
I have written much more than I did all summer '78 in FAIL. I haven't
included any actual compaction/coding, it just computes the entropy that
would result if an optimum code were used, assuming IRSM will produce
virtually identical compression when installed in future. At present it
is merely a research tool to develop the methodology that was planned
for IC3 but never implemented (purging of histograms when storage full,
instead of simply freezing adaptivity like IC2 does) and some later ideas
I came up with (like using reduced histograms, i.e. with escape node for
all rare tokens in given context, to fit the same effectiveness in less
memory). It is working nicely so far. Contact me for a demo. It has
lots of nice trace facilities: compression ratio and memory statistics,
purging a histogram, printing whole model at various breakoints -- all
with facility to automatically suspend trace for a specified number of
tracepoints and then resume user-control at that point, thus you can
tell it to wait until 3000 characters have been compressed then give you
statistics on compresson+storage. Or you can tell it to automatically
give you the statistics of the next 50 characters compressed without
requiring confirmation from user on each one. Want a demo? It sort of
shows nicely how my left-context reduced-histogram adaptive-histogram
model works. The only thing (in adaptive mode) it doesn't demonstrate
is the actual interval-refinement code which generates the compactified
output file.
∂23-May-79 2247 Buchanan at SUMEX-AIM IJCAI invitation for Ma Xiwen
Date: 23 May 1979 2246-PDT
From: Buchanan at SUMEX-AIM
Subject: IJCAI invitation for Ma Xiwen
To: jmc at SU-AI
cc: ijcai79
will be glad to issue an invitation on letterhead. Will send it up
as soon as Bethanne is back this week. It will be to Dr. Xiwen, unless
you tell me to use a different form of address.
bgb
-------
I dunno about Dr. I think they were bourgeois during the reign of Mao.
However, his family name is Ma and personal name is Xiwen - pronounced
"shiwen", so it should be addressed to Mr. or Dr. Ma. Chinese put the
family name first, but it can be either way.
∂23-May-79 2302 DCO
To: "@GROUP[1,DCO]" at SU-AI
PROGRAM VERIFICATION AND ANALYSIS SEMINAR
There will be no seminar on Tuesday, June 5. (That is in two weeks -
Andrzej Blikle will be speaking next tuesday, May 29. )
∂23-May-79 2348 JED the picnic wants YOU!
To: "@FAC.DIS[1,DPB]" at SU-AI
Hi. This is just a reminder that the '79 version of the CS Department
picnic will be happening this Friday, the 25th. We'd like to see as
many of the faculty as possible attend.
The picnic is in Mitchell Park, in south P.A., starting at about 2:30.
Full details can be found in the notices posted on bulletin boards at
Jacks and SAIL, and in PICNIC.INF[1,JED] @ SAIL.
Questions? See Carolyn Tajnai, Jim Davidson, or Randy LeVeque.
Hope to see you Friday.
R.S.V.P.
∂24-May-79 0007 REM via AMES-TIP New demo/R+D program for compression
I am still working out the storage allocation methodology. At present it
continues to grow new contexts until it runs out of memory and start thrashing
(swapping program modules in and out of core). Purging has been installed for
histograms, so more will fit in core than would last year with full histograms
(the current ones average 10 to 20 explicit characters plus an escape token,
whereas last year's have about 136 explicit tokens consisting of all 128
ascii characters plus about 8 special tokens), but no purging or limiting
of the context tree has yet been programmed.
At present it can get about 37620 bits into the file before the context
tree and histograms fill up available memory. Using 79511A.WRU as input
text, the c.r. over the entire file up to that point is 1.803 to 1.
Note that it starts out with no compression at all, rapidly adapts to
a sufficient c.r. that overall it has reached 1.5 to 1, and gradually
creeps up and up as it adapts. When it reaches the present limit ov
37620 bits input (1045 pdp-10 words) it has grown 226 different contexts,
has sprouted an additional 1218 pre-contexts, and the 226 contexts each have
a histogram such that the total number of tokens (characters mostly) in
all the histograms equals 3843 (thus averaging 17 entries per histogram,
16 characters or other tokens plus the escape token). -- Note that
1.8 to 1 isn't too bad for an adaptive algorithm applied to a 1k file.
1k is really too small for the program to adapt properly. Usually I
require 5k words to expect 2 to 1 c.r. and 50k or 100k for a really good
estimate of what the program really is capable of.
∂24-May-79 0057 LES Manna segment of ARPA proposal
Zohar will dispatch the handwritten version of his proposal tomorrow.
I'll send someone (probably me) to the airport to pick it up when it appears.
He seems quite reluctant to take the acting director job and urges you to
consider other alternatives. I gather that he doesn't like the idea of
interacting with ARPA. He also left the door open a crack, saying he is
willing to discuss it further if we are in a real bind.
∂24-May-79 0837 Feigenbaum at SUMEX-AIM Re: Rosse memo
Date: 24 May 1979 0836-PDT
From: Feigenbaum at SUMEX-AIM
Subject: Re: Rosse memo
To: JMC at SU-AI
cc: LES at SU-AI
In response to your message sent 23 May 1979 1639-PDT
From connecticut en route:
John, I have not seen the Rosse memo to which you refer, but your
line of reasoning is identical to mine. I will have to deal with
this when I return (to orrow). If he's trying to get more money
out of us, then I may reduce our offer from 25k to zero.
Ed
-------
∂24-May-79 1127 MAX
Can you take me to Faculty club and back? Please mail very soon!
∂24-May-79 1328 TOB
To: JMC, LES, RWW
∂24-May-79 0627 Engelmore at SUMEX-AIM Re: proposal
Date: 24 May 1979 0626-PDT
From: Engelmore at SUMEX-AIM
Subject: Re: proposal
To: TOB at SU-AI
In response to your message sent 23 May 1979 1453-PDT
Dear Tom,
Larry D. is right, and your proposal should be in by June 1. In
case that does not happen, as I'm almost positive it won't (given
past history), let me set a realistic ABSOLUTE deadline of having the
proposal in the ARPA office by Monday, June 11th. I'm glad to know that your
part is nearly all done. Perhaps you can push Les and JMC to get the
whole thing together as quickly as possible (it needn't wait for my visit).
See you,
Bob
-------
∂24-May-79 1335 HVA Martin Davis
To: JMC, HVA
Somehow (I don't know how), your msg. to me re Martin Davis has disappeared
from my file. If you still have it in your files, would you pls. send to me
again? Many thanks.
∂25-May-79 0146 LES computer time for music
In response to your inquiry, the computer time consumed by the music
project and music classes, measured in percentage of total doubloons,
has been as follows:
'71 '72 '73 '74 '75 '76 '77 '78 '79 (to date)
8% 13 8 10 15 19 28 25 29
←----- BUREAU -----------→ ←--- SOB --→
Note that the figures reported by BUREAU and SOB are not directly
comparable in that BUREAU reported computer time for system services (e.g.
spooling and news service) as separate items, while SOB subtracts these
times from the pool before computing percentages. This has the effect of
making the SOB figures larger but more accurate.
Ed:
The following figures provide ammunition for reminding
the dean and the musicians that their expenditures should be
considered as paying for the use of the facility and not merely
for expansion. This was the understanding and "expectation".
Also they should be reminded that the musicians in fact contributed
an additional approximately $75K for the KA-10 but, as Chowning
put it at the time, "mortgaged" the KA-10 when they were short
of salary money. Moreover, if we hadn't had them as the heaviest
users we would not have needed the second disk channel. Moreover,
the Lab never got any use out of the disk they bought, because,
contrary to the original understanding, they insisted that it be
used as a second user drive, since they have many user packs, rather
than as a system pack that would expand the file system.
∂25-May-79 0752 Feigenbaum at SUMEX-AIM (Response to message)
Date: 25 May 1979 0751-PDT
From: Feigenbaum at SUMEX-AIM
Subject: (Response to message)
To: JMC at SU-AI
In response to your message sent 25 May 1979 0204-PDT
Thanks, John. That was extremely helpful!
Ed
-------
∂25-May-79 1536 DEW
John,
Yes, I plan to be around next Wednesday. I'm also considering working
at SRI so the details of the offer may be important to help me make
a descision. (I hope to decide by June 15 although I won't be done by then.)
One thing I'm concerned about is the Jacks environment.
It seems like it will be even more of a zoo there when the lab moves,
and many students are not very happy already.
Perhaps I worry to much. It would be nice to continue work on my program.
Thanks, Dave
∂25-May-79 2018 FB meeting
11am Saturday at the lab, ok.
∂26-May-79 0917 REM via AMES-TIP Addenda on c.r.
IC2 (the program that has been working since September) gets 1.85 to 1
compression ratio with 79511A.WRU (a very small text file, about 1 k).
Note that it uses a full 210000 words of PDP-10 memory for constructing
tables, or at least it has that available (it doesn't fill it so I don't
actually know how much memory it uses).
IC7 (the new program I'm working on) claims entropy of about 1.803 to 1 c.r.
not counting overhead of header that would have to be added (about 5 words).
But IC7 has only 128k to work in, which includes not only data but mainsail
kernel (or maybe it's 64k, I forget), and the data structures of mainsail
take more space than those of hand-coded FAIL, so considering the memory to
work with I'll be pleased if I can make do as well. Of course, the idea isn't
to do as well, but rather to do almost as well in much less memory. Working...
∂26-May-79 1750 Mark Crispin <Admin.MRC at SU-SCORE> [PBARAN at USC-ISI: PAN - a tiny electronic mail system]
Date: 26 May 1979 1743-PDT
From: Mark Crispin <Admin.MRC at SU-SCORE>
Subject: [PBARAN at USC-ISI: PAN - a tiny electronic mail system]
To: JMC at SU-AI, Les at SU-AI
---------------
Mail from MIT-MC rcvd at 26-May-79 1546-PDT
Date: 26 May 1979 1509-PDT
From: PBARAN at USC-ISI
Subject: PAN - a tiny electronic mail system
To: msggroup at MIT-MC
cc: gage at ISIE
Up until March of '79 The Personal Computer NETwork (PCNET) Committee has taken
a 'bottom up' approach to personal computer telecommunication; to provide
general and powerful mechanisms for reliable data transmission, and when these
were thoroughly debugged to supply the electronic mail and file
transfer systems based on them.
We've changed our strategy. We're now going to take a 'top down'
approach, providing an immediately usable electronic mail system. The
sophisticated bells and whistles can be added later. We intend to
continue work on the full PCNET protocol, with all its power and
complexity. However, priority will be given to immediately usable
systems.
Our new electronic mail system is called PAN; it has been written by
Doug Gage in collaboration with The PCNET Committee. PAN is presently
implemented for an 8K PET and TNW modem; we plan to write compatible
PAN software for the Apple, the TRS-80, etc. Any PAN should be able
to communicate with any other PAN, regardless of the hardware used. PAN
is designed to use the dial telephone system and is fully distributed; no
equipment external to the personal computers and modems is required.
To make complete use of PAN, the modem used must be under full software
control. The software must be able to control the phone line status
(on or off hook), dial numbers, select originate or answer mode, and
detect incoming ring signals. It is desirable that the modem be
hardwired to the phone line; the modem should be FCC certified for
direct connection or be used with a Data Access Arrangement.
Acoustically coupled modems could be used, but dialing, answering the
phone on incoming calls, changing between answer and originate calls,
etc., would all have to be done manually. Present PAN software would
have to be modified to support manual operation.
PAN User Interface
PAN is similar to Community Bulletin Board Systems (CBBS) in general
functioning and command
structure; see the Nov '78 Byte for a description of CBBS. It is different in
that PAN is a distributed system; there
is no single phone number and set of equipment through which all
messages must pass. Upon first starting up PAN, the user is prompted
for the time and asked to rewind the message logging tape.
The PAN electronic mail system accepts single character commands from
the user. They are as follows:
E Enter a message
The user enters the message text with the screen editor. The
first ten (seven if a local call) digits encountered by PAN will
be interpreted by PAN as the phone number of the recipient PAN.
Entry is terminated with the '←' character and aborted with the
RVS key. When entry is terminated PAN assigns a message number
and the user is given a menu of message modes:
N No mode assigned;
I Immediate. Transmit this message now;
D Deferred. Transmit the message later; user is prompted for
the time of transmission.
M Assign a new mode to the message whose number is given by the user.
As above, the user selects from the mode menu. Two modes are
assigned by PAN:
S for messages successfully sent, and
R for messages received.
K Kill the message whose number is given.
The user is asked for confirmation, and the message is deleted
from the buffer. The space freed is available for new messages.
R Retrieve the message whose number is given and display it on the screen.
S Summarize the messages in the buffer.
Display on the screen the number, mode, size and first 26
characters of each message in the buffer.
? List the PAN commands.
U Unload and terminate PAN.
Write an EOF to the message logging tape and exit to BASIC.
Note that the PCNET version of PAN has commands which do not conflict
with and in many cases are the same as those of the CBBS.
PAN is designed for unattended operation; complete automatic dialling
and call management are provided.
PAN has undergone about 8 weeks of operational tests. The PCNET
Committee plans to start distribution of PAN software as soon as we arrange for
cassette tape duplication, write a user's manual and arrange
distribution channels.
Dave Caulkins
-------
---------------
-------
Well, it appears that after taking two years to come around to the
Dialnet philosophy, they have about caught up in implementation.
However, I don't quite understand what they have done about
making the mail facility compatible with simultaneously using
the computer for other applications.
∂26-May-79 2317 MRC PCNET mail
To: JMC
CC: LES
Well, I don't see how they can use the computer for other stuff
simultaneously. You are talking about PETs, TRSs, etc. That ain't
much computer to work with.
∂27-May-79 1835 ARR Unused files in [PAS,S1]
To: S1 at SU-AI
The next files in [pas,s1] are not being used by their creators. If you are
using them, please let me know. If nobody claims them by wednesday, 30-jun-79,
I am going to delete them.
PASRUN.LDI
Numer1.ldi
Pas6.do
Pascal.do
SOPAW5.DMP
FASM.DMP
FSIM.DMP
PS: ME is complaining about our disk space. Probably it would be good if we
all did something like this on our space in S1.
Armando.
∂27-May-79 1840 ARR my previous message
To: S1 at SU-AI
OOPs! I goofed. I meant wednesday, 30-may-79, or next wednesday.
Thanks to EJG.
Armando.
∂28-May-79 0848 CSL.FB at SU-SCORE The Dover Printer
Date: 28 May 1979 0839-PDT
From: CSL.FB at SU-SCORE
Subject: The Dover Printer
To: jmc at SU-AI
Xerox people say that the Dover is supposed to be able to run at half speed.
This would make random graphics possible. However, the machine would have to
be readjusted to run at that speed all the time. That, I think, is why Xerox
Parc has never tried it. They are really attached to the high speed.
-------
If half speed would be a page every two seconds that is three times
the speed of the XGP. There would undoubtedly be some delays when
the Dover was supporting the whole department, but I think they would
be tolerable and could be mitigated by putting something like a
Printronix on each floor. Therefore, I think we should consider
running the Dover at half speed, because it will be very expensive
in manpower to keep the XGP running. Moreover, it will allow programs
that produce good text and reasonable graphics on the same page.
∂28-May-79 1430 PAM
To: "@AIQUAL.DIS[1,TW]" at SU-AI
Hear Ye! Hear Ye! The AI Qual will be held Thursday June 7
All those concerned should send a note to TW@SAIL specifying times that work
for getting together.....
∂29-May-79 0759 Feigenbaum at SUMEX-AIM (Response to message)
Date: 29 May 1979 0748-PDT
From: Feigenbaum at SUMEX-AIM
Subject: (Response to message)
To: JMC at SU-AI
cc: fb at SAIL
In response to your message sent 26 May 1979 1246-PDT
I'll be glad to meet with you and Forest friday afternoon. I have a class
until 3:05. How about 3:30 or 4? What's it about?
Ed
-------
∂29-May-79 0755 FB Dover and XGP
To: DEK
CC: JMC
John mentioned scraping the XGP as soon as the Dover came in and I pointed
out that the Dover will not support arbitrary graphics. (You probably know
about the reasons why.) I reported that the Dover was designed by the
xerography people to run at half speed and that it could probably be made to
do random graphics at that speed. No one at Parc has ever tried it, I think,
because they prize the speed more than the extra power. It would have to be
completely readjusted to run at half speed so it is not a switchable option.
I just sent you a copy of John's response to these comments. Your comments?
Forest
∂29-May-79 0804 FB Subject of Friday's meeting
To: EAF at SU-AI
CC: JMC at SU-AI
VAXes, design systems, and terminals & terminal systems.
∂29-May-79 0832 Feigenbaum at SUMEX-AIM meeting Friday
Date: 29 May 1979 0823-PDT
From: Feigenbaum at SUMEX-AIM
Subject: meeting Friday
To: jmc at SAIL, fb at SAIL
Looks like I already have a meeting at 4pm. Better make out meeting at 3:20
to 4, or else real late in the afternoon, like 5...Ed
-------
5 is best for me, but 3:20 is also ok.
∂29-May-79 1319 Feigenbaum at SUMEX-AIM meeting Friday
Date: 29 May 1979 1317-PDT
From: Feigenbaum at SUMEX-AIM
Subject: meeting Friday
To: jmc at SAIL, fb at SAIL
It looks like 5pm is the right time for our meeting. See you here in MaJaHa
in the Chairman's Conference Room. I'll assume 5 unless I here negatively
in the Chairman's Conf Room at 5, unless I hear otherwise from Forest.
That's Friday at 5.
Ed
-------
∂29-May-79 1358 ARK via SU-TIP S-1 Meeting Reminder
To: S1 at SU-AI
CC: pickering at SUMEX-AIM
There will be an S-1 meeting tomorrow, Wednesday, 30 May in room 301
Margaret Jacks Hall.
Arthur
∂29-May-79 1537 MAX
I will foward your message to Hung very soon.
∂29-May-79 1558 100 : ljh Message
Please call Mr. Lindsay in the morning (5/30). The number is 321-2052 and the
message is "the sooner in the morning, the better".
∂29-May-79 2336 LGC via AMES-TIP Engelmore Visit
Should I be making any special plans or preparations for tomorrow?
I haven't received any information on this subject, and have been
too busy with my IJCAI paper to notice.
∂30-May-79 0258 ME
We were once again unable to load the system with only one controller.
I don't know why this time. It should have worked.
∂30-May-79 1055 PAT
∂30-May-79 0200 JMC reservations
Please get me psa reservations from San Jose to LAX and back Thursday
for III board meeting. Meeting starts at 10am, and I can take plane back
after 4:30.
-----
You are confirmed on:
PSA 162 leave SJ 8:30am arrive LAX 9:25am
PSA 565 leave LAX 5:30pm arrive SJ 6:25pm
You can pick up and pay for the tickets at the airport 35 minutes ahead of
departure.
∂30-May-79 1136 ARR S-1 meeting today
To: S1 at SU-AI
∂29-May-79 1358 ARK via SU-TIP S-1 Meeting Reminder
To: S1 at SU-AI
CC: pickering at SUMEX-AIM
There will be an S-1 meeting tomorrow, Wednesday, 30 May in room 301
Margaret Jacks Hall.
Arthur
[I asume that it will happen at 2:30PM, as usual. Armando.]
∂30-May-79 1151 JMC call
call bob kirkman at Bank of West 998-6608 (returning your call of last week)
∂30-May-79 1203 Morris at PARC-MAXC Testing, 1, 2, 3, ...
Date: 30 May 1979 12:01 pm (Wednesday)
From: Morris at PARC-MAXC
To: sw.dl↑
Subject: Testing, 1, 2, 3, ...
Stand by for a message of interest
∂30-May-79 1209 Morris at PARC-MAXC Xerox PARC Semantics Week Lectures
Date: 30 May 1979 12:07 pm (Wednesday)
From: Morris at PARC-MAXC
To: sw.dl↑
Subject: Xerox PARC Semantics Week Lectures
We have arranged a series of lectures on semantics of programming languages for
the week of June 4. You, and any interested colleagues, are cordially invited.
Besides the lectures there will probably be some informal discussions. Feel free
to call me or Dana Scott if you have any questions.
The lectures are rated according to the expected orientation of the audience.
Uses of Mathematical Semantics of Programming Languages (G)
James Morris, Xerox PARC, 10:30 Monday June 4, CSL Commons
I shall describe the approach and role of MSPL by some analogies with locally
familiar ideas like spline fonts and finite state machines. An application of
Scott's theorem to reasoning about programming abstractions will be presented.
Structures for a Mathematical Theory of Computation (G)
Dana Scott, Oxford University, 1:30 Monday June 4, CSL Commons
There has been much confusion in the literature about the use of things like
lattices in Scott/Strachey semantics. Starting from simple examples I will outline
how the concepts were forced on me. The emphasis will be on motivation and
there will be no attempt to give proofs. I will try to provide some bibliography
for further reading.
Understanding "Data Type" (PG)
James Donahue, Cornell University, 1:30 Tuesday June 5, CSL Commons
"Data Type" is a much written-about phrase in the recent programming
language literature. In this talk I discuss a new theory of data types in
programming languages that casts light on several issues of language design.
Surprisingly, this theory is inspired by Scott's construction of typeless "universal
spaces." I sketch the mathematical properties of these typeless spaces and show
how they lead to a new way of thinking about types.
On the Development of Correct Programs with Documentation (PG)
Andrej Blikle, Polish Academy of Sciences, 10:30 Wednesday June 6, CSL
Commons
The documentation of a program in our system consists of a precondition, a
postcondition (global documentation) and a set of assertions (local
documentation). Such a program is called correct if: (1) It is partially correct
with respect to the pre and postcondition. (2) The precondition guarantees that
the program neither loops indefinitely nor aborts. (3) The assertions of the
program are sufficient for the proof of (1) and (2). Programs in our system are
derived by such composition and refinement rules which always preserve the
correctness. The requirement of nonabortion in our definition of correctness
leads to the necessity of using a three-valued (or partial) logic. We are using
McCarthy's logic in that place.
Models for Type Theory (X)
Dana Scott, 1:30 Thursday June 7, CSL Commons
From my point of view "typeless" lambda calculus is the theory of infinite
types. This is only possible within the realm of partial functions. The talk will
concentrate on the mathematical background for building mathematical models of
such a function theory and how a variety of different types (including
recursively defined types) can be modelled. A discussion of what makes the
models "computable" will also be necessary.
Type-Checking in a Typeless World (R)
James Donahue, 10:30 Friday June 8, CSL Commons
What's type-checking all about? In this talk, I explore this question in detail,
using a polymorphic lambda calculus to show: (1) how polymorphic procedures
can be type-checked and (2) what semantic properties are gained by doing
syntactic type-checking. I give a semantic characterization of "strong typing"
and show the lambda calculus semantics to be strongly typed.
G: General audience, PG: Programming Language Buffs, R: Semantics
Aficionados, X: Hardcore Mathematicians
XEROX CORPORATION, PALO ALTO RESEARCH CENTER, 3333 COYOTE HILL
RD., PALO ALTO, CA. 94304 . (415) 494-4440
∂30-May-79 1417 TW via SUMEX-AIM Assignments for Qual
To: "@AIQUAL.DIS[1,TW]" at SU-AI
The AI qual will be next Thursday, June 7. Specific times for individual
exams will be set tomorrow. If you have not told me your preferences,
please do.
Below are the assignment of articles to read for your examination next
Thursday. Please obtain your article from Marion Hazen (229 Margaret
Jacks Hall), anytime after 12 noon, Thursday, May 31. Write a brief (2-3
page) critique of its content (NOT form), and turn it in by 11am Tuesday,
June 5. Your critique should address some of the following points:
A very brief summary of the paper
What are the new ideas, the "contribution to knowledge"?
What are the new techniques, the new engineering, the new accomplishments?
What is the real significance of this; why did he choose it to work on?
How has the author's work built upon, or at least learned from, that of others?
What might the "geneology" of this work be?
How does this work fit in with that from other subfields of AI (or CS)? Are there
analogies which could be employed here to good effect?
What related work has the author failed to mention? Do some of his ideas overlap
others'?
Where are the blind spots in his argument, the overgeneralizations, the errors?
What are some suggestions for extending this work, that you might pass along to
him if you were advising him? What are future directions to go in,
experiments to perform on the existing system, etc? How difficult is
each of these; in particular, which might be appropriate thesis-sized
projects?
All these articles are extremely current (they are drafts accepted for
IJCAI-79); note that you have intentionally been assigned an article far
outside your avowed subspeciality in AI. At your oral exam, your critique
will be used to aid in sparking discussion. Rest assured that there will
be ample opportunity at that exam to demonstrate your mastery of other
subareas.
KJK Kevin Karplus (systems)
John McDermott: Learning to use Analogies
ROD Rod Brooks (vision)
Jaime Carbonell: Computer Models of Human Personality Traits
JED Jim Davidson (natural language/representation)
John Gaschnig (SRI): A Problem Similarity Approach to Devising Heuristics
SJW Steve Westfold (automatic programming)
M. Stefik: An Example of a Frame-Structured Representation System
DLO David Lowe (natural language)
P. Langley: Rediscovering Physics with Bacon III
RDG Russ Greiner (vision/discovery)
D. Barstow: The Roles of Knowledge and Deduction in Program Synthesis
JSB Jim Bennett (knowledge acquisition/expert systems)
Roger Schank: Parsing Direcly into Knowledge Structures
∂30-May-79 1546 DCL Lab computer costs
To: LES
CC: JMC
Les, do we have figures for the LAb. computer costs yet?
-David
∂30-May-79 1549 HVA MARTIN DAVIS
To: JMC
CC: LES, HVA
Out of which pocket are we going to pay Martin Davis?
∂30-May-79 1811 MRC
To: JMC, LES
∂30-May-79 1235 Yeager at SUMEX-AIM TOPS20 implimentation
To: MRC at SU-AI
Date: 30 May 1979 1225-PDT
From: Yeager at SUMEX-AIM
Subject: TOPS20 implimentation
To: crispin at SU-AI
Mark,
I can now send an retrieve files from TOPS20 sites. I use TYMNET
and have a minimal number of retransmissions, and often have none.
I'll be starting my PDP11 implimentations early next week, and
will let you know how things are going with the VADICs when
ever I hear.
Bill
-------
∂31-May-79 1926 ME 1-controller system
To: JMC, LES
We are now running a system that uses only one disk controller.
∂01-Jun-79 0742 PHY
To: FB, JMC
I discussed your notes with students and with PARC people. At PARC the
machine stands idle about half the time, so half speed wouldn't be too much
bottleneck. But Leo felt one could get as arbitrary graphics as you would
want (considering that Dover doesn't print solid black areas) by lowering
the resolution and improving the software. I think he may be optimistic
about going at full speed and I guess we won't be doing too many multiple
runs on Dover so the half speed is safer.
- Don Knuth
∂01-Jun-79 1028 REM at MIT-MC (Robert Elton Maas)
Date: 1 JUN 1979 1327-EDT
From: REM at MIT-MC (Robert Elton Maas)
To: REM at SU-AI
CC: RIVEST at MIT-MC, JMC at SU-AI, RWW at SU-AI
This morning I came up with another "obvious" idea. One problem with the
Rivest-code has been the one-bit-per-block expansion of data for each
block of 200 bits encrypted. If more than one stage of encoding is done,
for example to both encrypt and sign a message, then in most cases the
odd bits added to each block will cause a phase-shift of one bit per
block requiring not only repacking of blocks between the two codes but
one extra block at the end to hold the overflow bits and a lot of zero
padding. For example with 200-bit modulus, and a message of 1024 bits,
the first Rivest-encoding breaks the message up into blocks of 199 bits
each, thus 199+199+199+199+199+29 bits going into the code, and
200+200+200+200+200+200 bits coming out. Then this is reblocked into
199+199+199+199+199+199+6 to feed into the second Rivest-code, which
comes out as 200+200+200+200+200+200+200=1400 bits. This is a
considerable waste of bandwidth when transmitting such a message.
My solution is to treat the modular residues (the integers from 0 to
m-1 which go into and come out of the individual Rivest-codes) as a
Markovian stream and to optimally compress and decompress this stream
using my IRSM method. There is virtually no expansion of the data during
stream mode, then there is at most about 16 bits of slop at the very end
when the stream is closed and resynchronized. Thus 1024 bits of message
would require 1040 bits instead of 1400 bits of transmission.
Note that since the Rivest-code itself requires 200-bit multiplications,
the Markovian stream could be processed as-is using bignum multiply/divide,
but then the slop at resync would be 200 bits. Better would be to break up
the stream into 8-bit bytes, complicating the model a little (it becomes
a multi-state model instead of a trivial one-state memoryless random stream)
and very slightly degrading the entropy match, but reducing the slop to
the 16 bits advertised above.
(Slight correction, there would still be Rivest-slop going into the very
first code or coming out of the very last code depending on how you analyze
it. I'll have to compute the correct amount of total slop sometime.
I expect that for short messages there will be 200 bits of slop instead of
n*200 bits where n is the number of signatures etc., whereas for long
messages there will be 200+x*q (x<<1 q=length of message) bits instead of
n*200+q. In any case, my method will be significantly better than the
ordinary obvious method.)
∂01-Jun-79 1125 Feigenbaum at SUMEX-AIM divorce settlement
Date: 1 Jun 1979 1124-PDT
From: Feigenbaum at SUMEX-AIM
Subject: divorce settlement
To: jmc at SAIL
Talked to Rosse this morning. The divorce settlement is final at $25K.
The memo re $46K was just to let you and me know that he thought we were
not being overly generous. However, I let him know that I thought we were
being overly generous. But $25K is acceptable and final.
See you at 5pm today.
Ed
-------
∂01-Jun-79 1133 Feigenbaum at SUMEX-AIM reminder:meting
Date: 1 Jun 1979 1125-PDT
From: Feigenbaum at SUMEX-AIM
Subject: reminder:meting
To: fb at SAIL, jmc at SAIL
I'll be here in MaJaHa to meet with you guys at 5pm today.
Ed
-------
∂01-Jun-79 1203 TW via SU-TIP Schedule for exams next Thursday
To: "@AIQUAL.DIS[1,TW]" at SU-AI
Here is the schedule for the AI qualifying exams. All of the exams will
be in Jacks hall (we'll find appropriate conference rooms or use offices).
Committee 1: Winograd, Lenat, Nilsson
9:00 Rod Brooks 10:30 Steve Westfold 1:00 David Lowe
Committee 2: Feigenbaum, Green, Weyrauch
10:30 Jim Bennett 1:00 Jim Davidson
Committee 3: McCarthy, Buchanan, Goldstein
10:30 Russ Greiner 1:00 Kevin Karplus
The exams can go up to 1-1/2 hours, shorter at the option of the committee.
There will be a meeting of all the examiners at 2:30 to decide on the
results, and they will be available following that meeting (hopefully long
before Earl Sacerdoti's talk at 4:15).
I assume you will let us know what papers our examinees have chosen to comment
on and that we will be able to get copies.
∂01-Jun-79 1343 Morris at PARC-MAXC Semantics Week schedule change
Date: 1 Jun 1979 1:38 pm (Friday)
From: Morris at PARC-MAXC
Subject: Semantics Week schedule change
To: AllXerox↑, SW↑
The first talk by Donahue and the talk by Blikle have been swapped, so that
Donahue's is on Wednesday at 10:30 and Blikle's is on Tuesday at 1:30. (This
message corrects the previous electronic message. The posted notices are correct.)
Jim
∂02-Jun-79 0015 LGC IJCAI Passage
I would very much appreciate your comments on the following passage
from my IJCAI paper. I do not have great confidence in the final
paragraph, though it seems to be onto something. Do you think it
merits inclusion? Sometime tomorrow I may be in a position to give
you a copy of the almost final version of the entire paper.
Thanks.
-----------------------------------
When it comes to reasoning about propositional attitudes, our
approach is quite different from the one currently taken by McCarthy,
which relies heavily on principles relating statements about propositional
attitudes to the truth values of the propositional objects of the
attitudes in various possible worlds. While this "possible worlds"
approach provides a powerful tool for systematizing reasoning about
knowledge <][McCarthy 1975; Moore 1977, 1979]@, in its simplest form it is
committed to some quite strong idealizations that would preclude
its applicability in many practical situations, and that violate the
second desideratum adopted above in {yonsec desiderata}. For example, the
following principle plays a fundamental role in this approach:
.bnp
PW1: If a proposition P is true in every possible world compatible with what
a given person knows, then that person knows P.
.ec
This entails (most implausibly) that a person knows
all the logical consequences of his knowledge. Moore <][1977, 1979]@
acknowledges this difficulty in principle, and proposes to solve it by
regarding the conclusions sanctioned by PW1 as only presumptive in nature,
to be overridden by better supported information to the contrary.
However, his technical developments thus far take no account of this
caveat, and instead assume PW1 in its simple, unqualified form.
The implementation of PW1 as a merely presumptive rule of
inference would not seem to fit very well into a system designed to take
account of the reasoning abilities of people. For example, suppose that a
certain proposition C is deducible from Jones's knowledge, but only by
arguments that are too subtle and complicated for Jones himself to follow.
Imagine further a system of the sort in question in which the conclusion
(via PW1) that Jones knows C is to be overridden by a contrary conclusion
(via some other rule R), based partly on the information that Jones can't
follow the deduction of C from what he knows. It would seem that any
system of this sort could be improved by replacing PW1 with a rule to the
effect that a person knows any proposition that he can easily determine
(correctly) to be a consequence of his knowledge. This would avoid such
unproductive conclusions as the one formerly sanctioned by PW1, without
any serious loss. Alternatively, the antecedent of PW1 might be
strengthened by adding the condition that a person be capable of correctly
inferring P from knowledge that is clearly his. But then the original
antecedent condition (that P be true in all possible worlds compatible
with the person's knowledge) would seem to become superfluous, and thus
eliminable. Either way, the result would be the same. Thus, when realistic
considerations concerning inferential abilities are introduced into a
system for reasoning about knowledge, principles such as PW1 appear to
lose their value. This suggests that in the long run, reasoning systems
originally based on the possible worlds approach may find themselves under
internal pressure to evolve in the general direction of the approach to
reasoning proposed here.
Well, I agree with you in the long run, and I'd bet I said as much
already in the 1950s. However, I can solve the S and P
problem with my formalism and hope to be able to solve more realistic
AI problems with it. I have no confidence in any current proposal for
limiting the amount of deduction, and I am not optimistic that you or
I or anyone else will soon find one. For this reason, the paragraph
seems like wishful thinking, and it seems to call for abandoning approaches
that solve some problems before finding anything that seems to be
better.
∂02-Jun-79 1139 DCL ARPA proposal
John, I would like to discuss the present proposal with you.
There are some details cncerning submitting my portion of it
through CSL which we should discuss. LES knows about this.
Also, there are some questions of the cost of AI Lab machine
which are required for the budget and which LES has been pretty
slow in working out. This is now an urgent matter since the deadline
was June 1st.
-David
∂02-Jun-79 1435 YYY via SU-TIP IBM visit
I understand that you are going to visit Watson Res. for some time in the
summer. Can you tell me when and for how long. I will not be back at the
lab until Aug. Would like to have you look at our LISP370 efforts and
future. Also like to talk to you about your views on symbolic and
algebraic computation, since I will be starting a group there.
I will be there from July 16 to approximately August 20 and would be
glad to discuss the questions you mentioned, although I haven't been
active in algebraic computation. We could also talk here sometime.
∂02-Jun-79 1548 Mark Crispin <Admin.MRC at SU-SCORE> G.xxx people
Date: 2 Jun 1979 1538-PDT
From: Mark Crispin <Admin.MRC at SU-SCORE>
Subject: G.xxx people
To: JMC at SU-AI
They are authorized guest users.
-------
∂02-Jun-79 1604 MRC via SU-SCORE
To: JMC, LES
BH says he'll work on adding Dialnet to MAIL in July
∂04-Jun-79 0158 MINSKY at MIT-AI (Marvin Minsky)
Date: 3 JUN 1979 2133-EDT
From: MINSKY at MIT-AI (Marvin Minsky)
To: jmc at SU-AI
CC: EF at MIT-AI, MINSKY at MIT-AI, selfridge at BBN-TENEX
Agree. Note that cost in Europe has been twice or so for a long time.
I would like to combine my teleoperator proposal, but perhaps not at this stage
because it would confuse people.
Am going to consult at Schlumberger some, and will probably be very
knowledgeable about this stuff soon.
∂04-Jun-79 0204 Feigenbaum at SUMEX-AIM mccarthy's views on terminal system
Date: 3 Jun 1979 0834-PDT
From: Feigenbaum at SUMEX-AIM
Subject: mccarthy's views on terminal system
To: jmc at SAIL
John, in case you're interested this is what you said a year ago re terminals.
What do you think now?
Ed
15-May-78 09:34:46-PDT,1654;000000000001
Mail from SU-AI rcvd at 15-May-78 0934-PDT
Date: 15 May 1978 0933-PDT
From: JMC at SU-AI (John McCarthy)
To: FXB
CC: LES, JJK, feigenbaum at SUMEX-AIM
∂15-May-78 0649 FXB via AMES-TIP#14 terminals
What is the status of departmental terminal projects, if any? I guess
I would like to get involved and do something although it would likely
be something that didn't quite fit your view (as I have understood it
in the past). Are you thinking about this subject these days?
As you know, I have wanted bit map terminals but found your old
terminal too expensive; a centralized memory seems to be cheaper.
I have delayed the project in the AI Lab, at LOTS and in the
Department for two reasons. Mainly, the new 16 bit large address
microprocessors forthcoming from Intel and Zilog has made me
want a larger address, because squeezing all the fonts, etc. into
the 12K address space that remains in a PDP-11 after 16K is used
to address the bit map seems marginal. A larger address space
makes the size and shape of the bit map more flexible, and we can
contemplate a system that will grow to high resolution and even color
if we have enough address space. A secondary reason is that the
price of 16K RAMs hasn't declined as rapidly as was expected.
In short the project is delayed but not forgotten. Johnathan
King has been pursuing money, and he has different ideas. I
propose to argue for my ideas when money arrives or the technology
is ready - whichever is first.
It seems to me that agreement will be easier, when we are ready to
go financially, because some of the options will be clearer.
-------
-------
∂04-Jun-79 1555 Zenon.Pylyshyn at CMU-10A (X320ZP51) Center
Date: 4 June 1979 1853-EDT
From: Zenon.Pylyshyn at CMU-10A (X320ZP51)
Subject: Center
To: John McCarthy <JMC at SU-AI>
Message-ID: <04Jun79 185317 ZP51@CMU-10A>
In-Reply-To: John McCarthy's message of 2 Jun 79 14:22
Too bad about Marvin, though Pat Winston had warned me some time ago that
this was likely, in view of Marvin's past record.
I also got a letter today from someone at the Center saying that you and
Gardner Linsey and Preston and Bob Moore had a meeting and were trying
to work out something for the AI group people to have terminals. I hope
something works out (in fact I'm depending on it). I don't suppose that
Haugeland will need one -- but propbably Hayes would appreciate one as well.
As for me all my work is on disk.
∂04-Jun-79 2000 DCL Cost of operating AI lab computer
To: JMC, LES
CC: DCL
John, FOR THE RECORD:
Les has given me an overall figure of $70K for 15% of the AI lab. machine
for two years.
This includes Systems programming and administration of computer, direct costs,
staff benefits and travel, 58% university overhead and amortization.
At the moment the estimated cost of SCORE is $1.5 k per 1% pie slice (including
disc storage etc.) per year.
Thus SCORE works out at $45k before university overhead for 15% for two years;
Final comparison: AI Lab 70K, SCORE $71.1 .
I am now preparing the CSL budget for my share of this and I hope LES
will help Sally Burns to do this if she needs more details.
-David
∂04-Jun-79 2040 LGC Proposal Revision
I have finished my revisions of two paragraphs in the ARPA proposal
dealing with the S and P problem. You'll probably want to change
a few things. The paragraphs start on line 65 and line 81 of page 8,
in the file FR[PRO,RWW] (note especially line 82).
Richard plans to PUB the file tonight.
∂04-Jun-79 2210 LES SAIL computer cost
To: DCL
CC: JMC
David, FOR THE RECORD:
I did not give you firm figures of any kind -- that was a hand wave based on
a 15 second calculation. Since you saw me do it, you know the depth of
thought that went into it. You also heard me hedge a minute later when
I thought of some things that I hadn't included.
In summary, please don't use that number. We should have a usable one tomorrow.
∂05-Jun-79 0130 JB
I have finished reviewing your corrections, and
I believe that I have accounted for all of them
satisfactorily. The one which I understand least
is your question mark about my observation that
"as one goes on adding new heuristic rules to a
theorem prover in order to make it capable of proving
a larger class of theorems, by necessity a point
of diminishing returns must eventually be reached.
What is then gained in terms of power will not
pay off in terms of the increased complexity and
costs of computing resources". Did you object just
to the formulation, or is my statement false?
Aboout Artificial Intelligence, I wanted to see whether
RWW's Prolegomena has already appeared, so I could quote
it correctly.
It hasn't yet appeared, and I don't think he knows what issue yet.
The statement seemed like an excuse for not having done more,
and I think we are very far from having exhausted the possibilities
of goal seeking proof strategies - probably even within
the present FOL.
∂05-Jun-79 0141 JB YES.
Yes, I fully agree. We have barely scratched the surface of
theorem proving strategies in GOAL. However I did expend
a large amount of effort in LOGIC and it was in general
true that the additions did not pay off. I will reformulate
the statement to avoid giving the impression of an excuse.
Your comments are very valuable. Thank you very much.
∂05-Jun-79 0853 SELFRIDGE at BBN-TENEXD tech comittee notes
Date: 5 Jun 1979 1152-EDT
From: SELFRIDGE at BBN-TENEXD
Subject: tech comittee notes
To: jmc at SU-AI, ef at MIT-AI, minsky at MIT-AI
cc: SELFRIDGE
I have put together the notes that John and I made last
Thursday; perhaps you could go over what is below and
message back to me corrections etc. I would like to RUNOFF
the whole thing and send it to Al by the
end of next week or so??
Oliver
.na
TECHNICAL COMMITTEE MEETING, 5/31/79
.sp 2
Present: Ed Fredkin
Al Fenaughty
John McCarthy
Oliver Selfridge
.sp 2
Agenda: several related topics had been brought up at the
Board meeting in the morning that were deemed to be suitable
for an ad hoc technical committee meeting. They included
the current state of OCR, the pricing of laser printers,
contract book reading or journal reading, the competition
for OCR and laser printers, software and hardware to replace
the III dependence on the III 15, and how to avoid the
fire extinction syndrome that makes it hard to assign high
level people to planning and development jobs. Some of
these were actually discussed.
.sp 2
1330:
.sp
John McCarthy pointed out that new technological developments made
storage of printed material about as cheap on disc as on the shelf.
There is apparently a CDC quote of $24K for 1.2 gigabytes, and
even if this is off by 3db it is directly comparable to book
and shelf costs. There will be increasing need on the part of
many businesses and agencies to computerize all incoming mail
and documents, and we should explore the use of OCR to do that.
.sp
ARPA seems to have more money available for new ideas, and it
might be receptive to a proposal to redo OCR better and cheaper.
A proposal would, J. McC. asserts, have a better chance if it
also emphasized a VLSI approach.
Various aspects of the low cost OCR problem were listed and
discussed:
.sp 2
LOW COST OCR
.sp
.in 3
.un
1. The Technological Situation
.br
Falling costs of disc storage
.br
Vast amounts of material that need to be computerized
.br
The increasing pace of office automation
.sp
.un
2. Applications
.br
library of Congress
.br
CIA
.br
Front end for office automation
.sp
.un
3. Costs
.br
System and optics
.br
Contract reading
.sp 2
.un
ACTIONS
.sp
.un
1. Research Proposal
.br
ARPA - VLSI with military applications
.br
CIA - journal reading, newspapers, etc.
.sp
.un
2. Company funded
.sp
.un
3. Market Study
.sp
.un
4. Post a price for }}simple}} book/journal reading and
plunge in
.sp 2
.un
OTHER CONSIDERATIONS
.sp
.un
1. Such a project would need a manager with no other higher
priorities.
.sp
.un
2. Competition?
.sp
.un
3. How much redesign in equipment, software, etc.?
.sp 2
.in
DISCUSSION
.sp
At this point three people including Arnie Griffiths joined
the discussion; they are currently working on AIDS and the
laser printer project. The initial topics concerned the costs
of the basic equipment and the possibilities of lowering them
by an order of magnitude.
.sp
The current input device is a scanner at about $.5M for one;
that sold with AIDS is about half that because of quantity.
It apparently costs about $750 to input a book with key strokes,
rough calculations suggested, and it was inferred that a cost
of about half that would be suitably competitive. John said
that we ought to announce that as a price, set up an office,
possibly near the CIA, and wait for business to flood in. That
price would hold for simple text of high quality, where we could
easily be sure of low error rates.
.sp
It was pointed out that some other companies are in the business,
Hendrix, REI, for example,and that we ought to find out their ecologies --
prices, performance, profits, etc. Such questions ought in any
event to be included in the market study, at which a serious attempt
has yet to be made.
.sp
Our current technique (and apparently that of the competitors) is to
copy the book on 35mm film; it might be possible to dump the picture
directly into memory, at, say, 20Mbytes/page. In fact (A.G.) you
don't need a whole page, but merely 3-4 lines. It's hard to get
good CRT's or flying spot scanners.
It was pointed out that memory/random access to a page is
irrelevant to pricing really, because the other costs would be
still the same, that is, overwhelming. After redesign, the 10th
machine would cost III about $350K, and would sell for about $1M,
provided that there were no large extra software costs.
.sp
Such a machine (A.F.) would also need about ten people to run it,
broken down approximately:
1-5 reject operators
2 machine operators
1 film operator
2-4 administrative/manager/secretarial
.sp
There is always some tricky balancing between reject rate and substitution
rate; the point would be to sell acceptable quality without proofing.
But it seems clear from these not terribly favorable cost
estimates that it would hard to find a company or agency willing to
take a gamble on buying a system with these uncertainties and costs.
.sp
(A.F.) Also, every time II has looked at setting up a service bureau, it
doesn't quite seem to pay. This is not to say that II wouldn't
accept service jobs -- they just have to be big enough, like $100K,
about 10**8 chars.
.sp
(J.McC.) Why not set up a service bureau at III?
.sp
(A.F.) To set up a responsible and responsive service would need
a committed machine and people, a new microfilm facility and service
people; the cost would be about $.5M to start up for a year.
.sp
Other problems touched on included copyright and foreign alphabets.
.sp
The general conclusion was that it was not obviously the right thing
to do to go forward: the big need was for more precise understanding
of the costs and the market.
.sp
O.G.S.
-------
Your memo seems fine to me. Unless you see some reason why
tact might be productive, you might add the following paragraph.
After hearing so many convincing arguments as to why nothing
can or should be done by III, McCarthy became convinced that nothing
is likely to be done at III and will try to set up a small
project on book reading at Stanford. Perhaps a completely different
approach will lead to success.
∂05-Jun-79 0910 Feigenbaum at SUMEX-AIM Acting Assistant Professor appointment
Date: 5 Jun 1979 0909-PDT
From: Feigenbaum at SUMEX-AIM
Subject: Acting Assistant Professor appointment
To: genesereth at MIT-MC
cc: buchanan, lenat, jmc at SAIL
Dean Rosse approved the appointment. I will move ahead to process it formally.
Details as discussed over the phone. Please acknowledge receipt of this
message. Let me know aws soon as you can what the probability is that you
will accept (hopefully, 1.0!).
Ed Feigenbaum
-------
∂05-Jun-79 1106 DEW Comments on my thesis
John,
I'm going to start rewriting sections of my thesis to include comments of
my readers. I would like to get whatever comments you have so far right
away. Thanks, Dave
∂05-Jun-79 1341 DCL
To: LES
CC: JMC
∂04-Jun-79 2210 LES SAIL computer cost
To: DCL
CC: JMC
David, FOR THE RECORD:
I did not give you firm figures of any kind -- that was a hand wave based on
a 15 second calculation. Since you saw me do it, you know the depth of
thought that went into it. You also heard me hedge a minute later when
I thought of some things that I hadn't included.
In summary, please don't use that number. We should have a usable one tomorrow.
REPLY: LES, as I told you yesterday, we have a deadline to meet.
I explained that I would have to go with your guestimate in order to meet that
deadline.
I requested these figures over two months ago, and again on May 1st.
I dont see what else I can do now.
If we do have time to use your revised figures, we will.
-David
∂05-Jun-79 1642 DPB End of quarter department meetings
To: "@FAC.DIS[1,DPB]" at SU-AI
To: CSD faculty, staff, students, friends
From: Denny Brown
Subject: End of quarter department meetings
Tuesday, June 12, 3 p.m. in Jacks 252 -- Faculty Meeting. Agenda below.
Thursday, June 14, 2:30 p.m. in Jacks 252 -- Black Friday Faculty Meeting.
Friday, June 15, 12 noon - End of Year full department meeting.
Agenda for faculty meeting - Tuesday, June 12, 3 pm. in Jacks 252.
1. Approval of degree candidates.
2. Report on support of new entering students. Please send your "claims"
to Denny Brown this week. Recall that Gene Golub circulated a
list of new students and a summary of their admissions info at a
previous faculty meeting. Only a few PIs have responded.
3. Old business, new business, announcements.
4. Discussion with students concerning the role of the comprehensive
examination in department policy. The students met recently and
requested such a discussion. See the memo from John Gilbert, one
of the current student bureaucrats, to EAF. This memo is enclosed
in the hard-copy version of this message. Find it on-line
on your favorite bulletin board(s).
The students intend to make some suggestions regarding the
Black Friday meeting. This portion of the faculty meeting will
be open to as many students as wish to participate (within the
constraints of the conference room.)
The Black Friday faculty meeting will be devoted entirely to the
annual evaluation of PhD student progress. Attendance at this meeting
will be restricted to CSD faculty members, non-CSD members of reading
committees, and student bureaucrats. Please mark your calendars and
plan to attend this important meeting.
The end of year full department meeting will be highlighted by presentation
of the Forsythe Awards. Annual recap of the state of the department,
general announcements, and open discussion will also be included.
Everyone associated with the department is invited to attend. This
includes faculty, staff, students, and friends.
Please send additions to the agenda of any of these meetings to Denny
Brown. Reminders will be sent out again next week.
∂06-Jun-79 0215 LLW Visiting Appointment for John Reiser
To: LES
CC: JMC, LLW
Dear Les:
I haven't heard a peep about the Visiting/Adjunct faculty appointment for
John Reiser which I've discussed with you and JMC, either from the BTL or
the SAIL directions. Is this matter moving along satisfactorily on your
end? When can I expect to start twisting Reiser's arm to show up?
Thanks,
Lowell
∂06-Jun-79 0925 LES Reiser appointment
To: LLW
CC: JMC
Lowell:
I checked into the appointment mechanism for John Reiser some time ago
but apparently didn't transmit the results --
There appear to be no problems. Formalities on visiting faculty
appointments are negligible. It can be dealt with at any time, including
after his arrival. If you want a letter saying this, I can have it
generated.
Les
∂06-Jun-79 1347 Feldman at SUMEX-AIM concept reification
Date: 6 Jun 1979 1346-PDT
From: Feldman at SUMEX-AIM
Subject: concept reification
To: jmc at SU-AI
I am eager to know if you have further pursued the work begun
in your 1977 paper "First order theories of individual concepts
and propositions", especially regarding formal details.
I am a graduate student at Rochester, studying with Dr. James Allen.
Sincerely,
Donald Perlis
-------
I have pursued it but have no further publication beyond that paper which
will appear in MI9. I also have an application of such a formalism to
a couple of puzzles involving knowledge. Lew Creary of this Lab has
modified my formalism to discuss concepts of concepts and has a paper
that will be given at the next IJCAI. He is called LGC.
∂06-Jun-79 1419 PERLMAN at BBN-TENEXA Any new ideas on internet congestion?
Date: 6 Jun 1979 1722-EDT
From: PERLMAN at BBN-TENEXA
Subject: Any new ideas on internet congestion?
To: PLUMMER, jmc at AI, clark at MIT-MULTICS, cerf at ISI
cc: PERLMAN
We left the internet meeting intending to have a small meeting on
source notification, and other congestion control items if they
came up. As far as I know, the only interested people are Bill
Plummer, Noel Chiappa, Dave Clark, and me. If anyone of you can
think of anyone else, tell me, and I'll schedule a meeting
shortly. Also, I wrote a brief description of my understanding
of the current state of things to refresh people's memories and
provide a basis for discussion.
This is my understanding of the state of congestion and flow
control.
Source notification, for the purpose of limiting the total amount
of internet traffic, is definitely desirable, so we will assume
it will be implemented and should be designed.
Other schemes for spreading what traffic there is more
efficiently throughout the internet are not as crucial, and would
not necessarily improve things (since they would add overhead and
could not react quickly to changing conditions since information
travels slowly in the internet). Unlike source notification,
which requires little work by the gateways, (although it does
require a reasonable amount at a source host), schemes for
efficient spreading of traffic would place significant demands on
gateways. Thus we will concentrate now on implementing source
notification, and leave other schemes in the thinking stage.
Source notification by the gateways is easy. All it requires is
that a gateway, when it is forced to drop a packet (or when it
notices its queues are building up before it is forced into
dropping packets), send a message back to the source that it
should cut back on the rate of traffic to that destination. We
can rely on probability to make sure the true culprit gets
notified eventually. In other words, if one source is generating
90 percent of the packets to a destination, then 90 percent of
the packets that will need to be dropped will be from that
source, and 90 percent of the notifications generated will be to
that source.
The problem is at the source. It will not be able to keep a
table of all internet destinations, so it will keep a table of
all destination nets. Packet generation rates will be based on
some time interval, say 10 seconds. The source will keep in the
table the number of packets sent so far during the current time
period, and the maximum desired number of packets to be sent
during a time period. Higher level protocols (such as TCP, FTP,
etc.) can look into that table and adjust their rates to
destinations accordingly. Either they should keep track of their
own rate or the internet layer can keep those figures. There are
two reasons for keeping the rate sent per protocol. One is so
that those modules can judge whether they are the major source of
packets for that destination, and the other is so that they can
adjust their rate (they have to know how many they sent in the
past time period in order to generate fewer packets in the next
time period).
!
-------
∂06-Jun-79 1608 ARK S-1 Meeting Notice
To: S1 at SU-AI
CC: pickering at SUMEX-AIM
There will be an S-1 meeting on Friday, 15 June at 11:00 pm in Margaret Jacks
Hall room 301. Note the change in day and time. If you cannot attend this
meeting, send me a message suggesting alternate times. Among other things,
at this meeting we will decide future meeting times.
Arthur
∂06-Jun-79 1700 JMC*
APPOINTMENT AT 6
∂06-Jun-79 2038 RWW arpa proposal
To: JMC, CLT, LES
it done
richard
∂06-Jun-79 2206 LGC ARPA Proposal
Enclosed below is an excerpt from a message I sent you last summer,
objecting to a passage (by you, I think) in last year's ARPA proposal.
In reading over the current ARPA proposal this afternoon, I rediscovered
the very same passage, and it still bothers me. I think it might bother
you also, so I'm raising the matter again (but this time, BEFORE the proposal
goes out). If you don't find a flaw in the objection, you may wish to fix
the paragraph objected to before LES PUBs the final version tomorrow.
--------------------
∂17-Jul-78 1550 LGC via AMES-TIP#20 Problem Solver
To: JMC
...
By the way, I am puzzled by something you say in column 1 of
page 5 of the ARPA proposal: "... mathematicians have proved the
completeness of the rules of inference of first-order logic; admitting
any general laws that generate conclusions not attainable by the
old laws makes the system inconsistent." This remark would be
appropriate if the predicate calculus were complete in the syntactic
sense that the addition of any non-theorem as an axiom leads to
inconsistency. However, the predicate calculus provably is NOT
complete in this syntactic sense (though the propositional calculus
provably is); rather the predicate calculus is complete in the semantic
sense that every valid wff is a theorem. Thus, your remark seems wrong.
Have I missed something here?
...
--------------------
One way of revising the offending paragraph (which starts at line 145 of
page 4 of FR[P,LES]) is as follows:
Circumscription is a fully formal mode of reasoning and can be programmed
for a computer. On the other hand, it is not %2valid%1, i.e. it can yield
false conclusions from true premises. This is to be expected, because
mathematicians have proved the completeness of the rules of inference of
first order logic; hence all the infallible inferences that do not depend
on the interpretation of the premises and/or conclusion have already been
provided for. So, programs that use circumscription cannot be certain
that their conclusions are as good as their premises, and must be made
capable of withdrawing conclusions arrived at by circumscription that turn
out to be unacceptable, without necessarily also renouncing the premises.
This will make them more like humans - getting increased power at the
price of fallibility.
A simpler version might go like this:
Circumscription is a fully formal mode of reasoning and can be programmed
for a computer. On the other hand, it is not %2valid%1, i.e. it can
yield false conclusions from true premises. Therefore, programs that use
circumscription cannot be certain that their conclusions are as good as
their premises, and must be made capable of withdrawing conclusions
arrived at by circumscription that turn out to be unacceptable, without
necessarily also renouncing the premises. This will make them more like
humans - getting increased power at the price of fallibility.
Let me know what you think.
I'm sure it won't matter much to ARPA, but I would be grateful if you
would undertake to replace the offending paragraph by either of your
amended versions. I have a slight preference for the first version.
∂06-Jun-79 2217 ARK Corrected S-1 Meeting Notice
To: S1 at SU-AI, pickering at SUMEX-AIM
∂06-Jun-79 1608 ARK S-1 Meeting Notice
To: S1 at SU-AI
CC: pickering at SUMEX-AIM
There will be an S-1 meeting on Friday, 15 June at 11:00 pm in Margaret Jacks
Hall room 301. Note the change in day and time. If you cannot attend this
meeting, send me a message suggesting alternate times. Among other things,
at this meeting we will decide future meeting times.
Arthur
Correction: The S-1 meeting will be at 11:00 AM not 11:00 PM. Having an
S-1 meeting is not my idea of a Friday night party!
Arthur
∂07-Jun-79 0652 DCO
To: "@GROUP[1,DCO]" at SU-AI
PROGRAM VERIFICATION AND ANALYSIS SEMINAR
PLACE: ERL 237
TIME: 2:30 p.m., June 12
TITLE: Design of Approximate Methods for Analyzing Programs
SPEAKER: Patrick and Radhia Cousot, Universite de Grenoble
Patrick and Radhia Cousot are slated to talk next tuesday, but
they have not yet confirmed their visit nor have I received an abstract of
their talk. I will send another note around as soon as I know their plans.
∂07-Jun-79 1417 PLS
I unexpectedly have to go out to South Africa for a few weeks
as my father is not well. I have some comments on Elephant concerning
the use of time periods rather than instants, as well as some of my own work
that I'd like you to comment on, so I hope we can have a talk on my return.
Peter.
∂07-Jun-79 1500 LES
∂07-Jun-79 1435 ENGELMORE at USC-ISI Re: Proposed budget
Date: 7 Jun 1979 1433-PDT
Sender: ENGELMORE at USC-ISI
Subject: Re: Proposed budget
From: ENGELMORE at USC-ISI
To: LES at SU-AI
Message-ID: <[USC-ISI] 7-Jun-79 14:33:28.ENGELMORE>
In-Reply-To: Your message of 7 Jun 1979 1333-PDT
Got your budget. The image understanding portion looks fine, but the
rest is so far out of line that I better let Bill Carlson discuss it with you.
He will call ASAP.
Bob
∂07-Jun-79 1626 ARR Give a subject to your mail.
To: S1 at SU-AI
It is difficult to go through a LOG file because a lot of our messages don't
have a subject, so the directory page is no good. Please make sure that the
stuff you mail does have a subject. If you mail from inside E, you can use
the /sub switch. For example, to send this message, I will use the E command:
αXMAIL/SUB S1 Give a subject to your mail.
Armando.
∂07-Jun-79 1810 REM at MIT-MC (Robert Elton Maas) Better than Huffman-code in yet another way
Date: 7 JUN 1979 2056-EDT
From: REM at MIT-MC (Robert Elton Maas)
Subject: Better than Huffman-code in yet another way
To: REM at SU-AI
CC: JMC at SU-AI, RWW at SU-AI, DEK at SU-AI
A few minutes ago I realized another advantage of IRSM over Huffman-code.
It is insensitive to sequence, i.e. whatever order the tokens are in
it gives essentially the same (= Shannon entropy) compression. Thus
you can maintain collating sequence by putting your tokens in alphabetical
(collate) order in each histogram providing you don't have ESCAPE nodes
or you find a way to interleave things to maintain collate sequence.
Thus you don't have to resort to the Hu-Tucker algorithm to maintain
collating sequence at the expense of compression efficiency (where you
lose twice, once for insisting on discrete bits and once for insisting
further on collating sequence).
Note that to permit sort/merge of compressed records, data must be
re-synced to a discrete bit boundary at the end of each record.
(Alternative, compress several records into one block, syncing only at the
end of a block, then decompress and compress during sort/merge to
properly sequence the individual records. But during random-access lookup,
only the single block containing the desired record need be decompressed.)
∂08-Jun-79 0059 LLW Offer of Appointment to Reiser
To: LES, JMC
CC: LLW
Dear Les:
I will certainly appreciate SAIL's sending a brief letter to John Reiser
offering a Visiting Faculty appointment, as you have offered to do.
Having such an appointment formally offered apparently makes a large
difference to BTL management as to whether they'll give him what is
tantamount to half-time sabbatical leave for a year's time. If the letter
could say that Stanford CS will extend such an appointment for a 12 month
period, starting anytime between now and the commencement of Fall Quarter,
at his option, it will probably be well-tuned to his management's needs.
Thanks,
Lowell
∂08-Jun-79 1105 LES Reiser appointment
To: LLW
CC: JMC
Since my last message, I have learned that pseudo-faculty appointments
have gotten slightly more complicated. Given that he is not currently
a faculty member elsewhere, the appropriate title for John is Consulting
Assistant Professor. In order to set it up, we need a current biography.
I will be happy to push on this as soon as I get the bio.
Reiser might even have one in our computer that he could update.
∂08-Jun-79 1513 LGC Perlis
Donald Perlis may be associated somehow with Jerry Feldman; Perlis says
that you told him about my extension of your work on "reification of
concepts". He apparently is interested in foundations of set theory.
Does any of this ring a bell?
I think I remember talking to someone whose name I didn't get or forgot.
∂08-Jun-79 1636 REM at MIT-MC (Robert Elton Maas)
Date: 8 JUN 1979 1938-EDT
From: REM at MIT-MC (Robert Elton Maas)
Sent-by: ←←←103 at MIT-MC
To: REM at SU-AI
CC: RWW at SU-AI, JMC at SU-AI, DEK at SU-AI
I have a simple representation for reduced-histograms which maintain
collate-sequence. Thus my IRSM method can be used to obtain optimal
data-compression and the compressed records can be sorted into collate
sequence without having to decompress and recompress afterward.
The reduced histogram preserving collate-sequence takes about twice
the space as a standard reduced histogram (about 20-40 cells instead of
10-20 cells; compare with 130 cells for full histogram for ASCII text),
thus won't permit as many histograms (left-contexts) as would the
current desequenced Zipf-order reduced histograms, but contain more info
than the Zipf-order histograms thus might achieve better compression
ratio. ANYONE WANT MORE INFO ABOUT THIS?
∂08-Jun-79 1810 REM at MIT-MC (Robert Elton Maas)
Date: 8 JUN 1979 2112-EDT
From: REM at MIT-MC (Robert Elton Maas)
To: FFM at MIT-MC, JMC at SU-AI, LJH at SU-AI, PBARAN at USC-ISI
Moaaaaaaaaaaannnnnnnnnnnnnnnnnnnnn.
I modified my ad for a girlfriend to emphasize that I want a
longlasting close relationahip. I placed the ad in the Bay Guardian
and in the Times-Tribune (formerly Palo Alto Times and Redwood City
Tribune). The Bay Guardian has a policy against including phone numbers,
so I had to leave that out. This afternoon I have been napping, avoiding
leaving the apt and avoiding using the computer terminal, leaving the
phone available for calls in reply to the Times-Tribune ad and to the
one at the library that so far has gotten one almost-nice reply (I met
her and she seemed to like me, but then she changed her mind drastically)
and several weird high-school-type replies+hangups. I have scheduled
my evening to go to folk dance (to see a girl I met two weeks ago who let
me hug and nustle her) for just a little while and get right back to
be by phone if anyone calls. Suddenly a few minutes ago I realized that
I had forgotten to include my phone number in the Times-Tribune ad!!!!!!!!
Hell oh hell oh dammit!
F--k S--t P--s C--t C--------r Mother------ T--. (Come over and hear
the full seven words I can't say in this message due to FCC regulations.)
∂08-Jun-79 1816 PBARAN at USC-ISI (Response to message)
Date: 8 Jun 1979 1815-PDT
From: PBARAN at USC-ISI
Subject: (Response to message)
To: REM at MIT-MC, FFM at MIT-MC, JMC at SU-AI, LJH at SU-AI
cc: PBARAN
In response to the message sent 8 JUN 1979 2112-EDT from REM@MIT-MC
Indeed a sad story. Suggest you call the
paper first thing tomorrow and append
the phone #.
Dave C
-------
∂08-Jun-79 1820 REM at MIT-MC (Robert Elton Maas)
Date: 8 JUN 1979 2123-EDT
From: REM at MIT-MC (Robert Elton Maas)
To: FFM at MIT-MC, LJH at SU-AI, JMC at SU-AI, PBARAN at USC-ISI
P.s. the words f--king and mother------ have been broadcast on both
KBHK (44) and KGSC (36) within the past two weeks, on a "documentary"
called "Scared Straight", about a program for scaring juvinile offenders
into keeping out of prison, by exposing them (and the viewers of the
program) to the vilest and most abusive language I've ever heard.
So either the FCC doesn't know about it yet, or they have started to
make exceptions for profanity like they did for nudity about 3 years ago.
(Breast examinations, beautiful nakid pygmy teenage women, nakid east
indies women, even some white women portraying native-americans on
"Ten Who Dared" barebreasted, and of course Roots nakid beautiful sexy
young black women -- even frontal male nudity in several National
Geographic episodes about east indies, and birthing showing nude babies
genitals and all and mother vagina and all).
∂08-Jun-79 2234 DCO
To: "@GROUP[1,DCO]" at SU-AI
PROGRAM VERIFICATION AND ANALYSIS SEMINAR
PLACE: ERL 237
TIME: 2:30 p.m., Wednesday, June 13
*** Note change in day for this week only ***
TITLE: Design of Approximate Methods for Analyzing Programs
SPEAKER: Patrick Cousot, Universite de Grenoble
Program analysis is essential in oprimizing compilers and program
verification systems. The lattice theoretic approach to semantic analysis
of programs make use of a complete lattice representing information which
are considered to be meaningful, of isotone maps representing the action
of passing information forward or backward through elementary statements
and of program-wide analysis algorithms such as the merge over all paths
and fixpoint algorithms.
We first show that various semantics of programming languages can
be expressed using such program analysis frameworks. We next study
various equivalent formalizations of the approximation process which
underlies the design of automatizable methods for analyzing programs. In
particular we justify the use of complete lattices and isotone maps. We
next propose and illustrate a design method which guarantees the
correctness and optimality of the analysis algorithms with respect to a
given semantics. The question of non-distributivity and merge over paths
versus fixpoint algorithms is revisited. Extrapolation techniques are
proposed in order to speed up the convergence of iterative methods
whenever chain conditions are not satisfied. Finally we propose
systematic methods for combining a set of program analysis algorithms.
∂09-Jun-79 0245 CERF at USC-ISI Re: Any new ideas on internet congestion?
Date: 9 Jun 1979 0243-PDT
Sender: CERF at USC-ISI
Subject: Re: Any new ideas on internet congestion?
From: CERF at USC-ISI
To: PERLMAN at BBN-TENEXA
Cc: PLUMMER at BBN-TENEXA, jmc at AI, clark at MIT-MULTICS,
Cc: cerf
Message-ID: <[USC-ISI] 9-Jun-79 02:43:00.CERF>
In-Reply-To: Your message of 6 Jun 1979 1722-EDT
RADIA,
YOUR SUMMARY MATCHES MY MODEL. HOWEVER, PLEASE THINK ABOUT
THE PROBLEM OF FAIRNESS AND UNDESIRABLE INTERACTIONS AMONG
TRAFFIC SOURCES. WE HAVE SEEN CASES WHERE LUMPED RESOURCES
(AS IN THE CASE OF IMP BUFFERS) CAN BE DOMINATED BY ONE HOST
TO THE DETRIMENT OF ALL OTHERS. CAN YOU IDENTIFY ANY SIMILAR
PROBLEMS WITH THE SOURCENOTIFICATION SCHEME YOU HAVE IN MIND?
PLEASE DO GO AHEAD WITH THE MEETING!
LET ME KNOW WHEN ITIS - I MAY BE LUCKY ENOUGH TO FIND
TIME TO JOIN YOU.
VINT
∂09-Jun-79 1629 JMC
Roy Adler 914 941-6039
∂10-Jun-79 1709 ARR nd MSG files
To: S1 at SU-AI
CC: 100 at SU-AI
All files with extension LOG, MSG or OMS were moved to the new area [LOG,S1]
files that used to have a .LOG extension have no extension now.
(E.g. SOPA.LOG[DOC,S1] is now SOPA[LOG,S1])
Files with extensions MSG or OMS remain the same.
ALL distribution lists in [DIS,S1] and in *.DIS[*,S1] were modified accor-
dingly. People who have private distribution lists are encouraged to update
them before the next usage.
File AGENDA.MSG (proposed topics for next S-1 biweekly meeting) was left in
[DOC,S1]
Armando.
∂11-Jun-79 1241 Feigenbaum at SUMEX-AIM moving the TIP
Date: 11 Jun 1979 1223-PDT
From: Feigenbaum at SUMEX-AIM
Subject: moving the TIP
To: les at SAIL
cc: admin.gorin at SCORE, jmc at SAIL
Les, have you contacted DCA people about moving the tip to MJH?
Assuming the sail computer will begin to be disassembled on July 30,
i.e. will be down and not needing arpanet access, the TIP movers
can do their job in the period Aug. 1 plus or minus one day.
How is it set up with them?
Ed
-------
∂11-Jun-79 1750 JMC remaining on estate
july 17 return due
1. jewelry appraisal
2. property tax statement 78-78
3. check register (did I pay any bills right after death?)
4. plane fare
∂11-Jun-79 1834 ADMIN.GORIN at SU-SCORE Re: moving the TIP
Date: 11 Jun 1979 1825-PDT
From: ADMIN.GORIN at SU-SCORE
Subject: Re: moving the TIP
To: Feigenbaum at SUMEX-AIM, les at SU-AI
Cc: jmc at SU-AI
In-reply-to: Your message of 11-Jun-79 1223-PDT
I saw a note from DCA today confirming the move of the TIP for July 30.
-------
∂11-Jun-79 2240 LES Moving the TIP
To: admin.gorin at SU-SCORE
CC: JMC at SU-AI, Feigenbaum at SUMEX-AIM
Yes, I received that paperwork too. It isn't what I told them.
By the way, when are you going to fix that crummy Mail program
at Score so that I can address things to your human name?
∂12-Jun-79 1439 PAT lecture
Dertouzos secty called and would like a response to the invitation you
received to give a lecture as part of their distinguished lectures series.
I think that the letter is in the pile on your table.
∂12-Jun-79 1443 PAT King paper
To: JMC
CC: LES
Jim King gave me a copy of a paper entitled "Program Correctness: On Inductive
Assertion Methods" which he claims you agreed to publish as an AI memo.
There is a problem regarding the printing in that, because of the agreement
with IBM that he would not be paid from any government contracts, we cannot
charge the printing of the memo to ARPA. Are you willing to pay for the
printing from one of your accounts?
∂12-Jun-79 1600 JMC*
strawberries
∂13-Jun-79 0243 LLW Reiser Bio
To: LES
CC: JMC, LLW
Dear Les:
Is this Vita suitable for your purposes re the visiting faculty appointment
for John Reiser?
Lowell
VITA
John F. Reiser June 1979
Bell Laboratories 4F-635
Holmdel, NJ 07733
(201) 949-3942
Member of Technical Staff 2/77 -
Bell Telephone Laboratories
As a member of the Interactive Computer Systems Research
Department, I am primarily concerned with increasing the
capabilities and usefulness of digital computers. I con-
centrate on operating systems, computer architecture, and
software which supports the programming environment. My
colleague Tom London and I recently finished the implementa-
tion of the UNIX/32V operating system and complete user
environment on the VAX-11/780 computer.
Graduate student 9/73 - 2/77
Computer Science Department, Stanford University
My theoretical studies were mainly in analysis of algorithms.
I received a Ph.D. in March 1977, with the thesis "Analysis
of Additive Random Number Generators," advised by D. E. Knuth.
On the practical side, I was one of the chief maintainers of
the SAIL programming language, and developed the BAIL program
for source-language debugging of SAIL code.
Programmer summer 1975
Computer Science Department, Carnegie-Mellon University
I designed and implemented an overlay facility for SAIL
code which enabled the Hearsay II speech understanding
system fo fit into the address space of a PDP-10.
Undergraduate student 9/69 - 6/73
Michigan State University
B.S. Mathematics, June 1969, Honors College.
PUBLICATIONS
UNIX/32V Time-Sharing System: UNIX Programmer's Manual, Bell
Telephone Laboartories, Holmdel, NJ, February 1979.
SAIL (ed.), Stanford Artificial Intelligence Laboratory Memo
AIM-289, August 1976.
BAIL, A debugger for SAIL, Stanford Artificial Intelligence
Laboratory Memo AIM-270, October 1975.
Evading the Drift in Floating-Point Addition (with D. E. Knuth),
Information Processing Letters 3 (January 1975), 84-87.
∂13-Jun-79 1139 CLT meeting of the minds
The initial "bull session" for trading ideas and problems relevant to
logic and cs will be Thursday June 21st at noon in the math building
room 383P (next to the faculty lounge). Please let me know asop if
this time is not OK with you.
Carolyn
∂13-Jun-79 1400 JMC*
todorovich
∂13-Jun-79 1720 LES Discussion of move to Jacks Hall
To: JMC at SU-AI, MUZ at SU-AI, Feigenbaum at SUMEX-AIM
Confirmed: Friday, June 15, 10:30am in Ed Feigenbaum's conference room.
∂13-Jun-79 2258 DCO
To: "@GROUP[1,DCO]" at SU-AI
PROGRAM VERIFICATION AND ANALYSIS SEMINAR
PLACE: ERL 237
TIME: 2:30 p.m., September
Our seminar series will continue in the fall. Have a good summer.
∂14-Jun-79 0600 PMF Leave of absence
To: DEK, JMC, DPB
CC: PMF
Since Black Thursday is today, I thought a note reminding people of my
plans would be in order. I hope to take the next year off to work on the
S-1 project at Livermore. I would be back the next fall (with one year
left of Hertz fellowship) to finish my PhD. I plan to do a thesis on
mapping networks. I am going to continue working some small amount of time
on my thesis, however I see my obligations at Livermore taking most of my
time. Since these obligations (finishing the design and implementation of
the S-1 Mark IIA) are fairly important to me and to my career, I feel that
another years leave of absence from Stanford is justified.
I did not file the reading committee or topic forms yet, since I would
prefer to wait until I return. However I will file the forms if this is
deemed important (by Stanford or by you).
∂14-Jun-79 0845 CET Rich Pattis grade
Rich Pattis is taking 9 units of CS390 from you.
I need a grade.
Thanks, Carolyn
Grade is N.
∂14-Jun-79 1353 DEW
John,
I've decided to work for SRI starting this fall, so you needn't create a
research associateship for me. I would have liked
to continue work on PARADISE, but I also think it would be good for me
to develop other areas of expertise. I plan to both finish my thesis and
submit a paper on PARADISE to the AI Journal this summer and start at SRI
in September. I feel SRI offers most of the advantages of Stanford as well
as more money and better career opportunities.
I also have some doubts about the Jacks environment in the short term.
Thanks very much for the opportunity.
Cheers, Dave
∂14-Jun-79 1427 PERLMAN at BBN-TENEXA source notification meeting
Date: 14 Jun 1979 1727-EDT
From: PERLMAN at BBN-TENEXA
Subject: source notification meeting
To: PLUMMER, jmc at MIT-AI, clark at MIT-MULTICS, cerf at ISI,
To: PERLMAN
Would 2PM on Wednesday, June 27, be OK? Please tell me whether
you plan to attend.
radia
-------
∂14-Jun-79 1658 Ijcai79 at SUMEX-AIM IJCAI-79 Travel
Date: 14 Jun 1979 1656-PDT
From: Ijcai79 at SUMEX-AIM
Subject: IJCAI-79 Travel
To: IJCAI-79 Attendees:
TRAVEL TO IJCAI-79
It is still possible to make airline and hotel reservations
for IJCAI. Contact
Ms. Anu Reddy
Peoples Travel Agency
201 South Craig Street
Pittsburgh, PA 15213
Phone: 412-621-0799
Please send $150 deposit plus $125. pre-registration fee to
Ms. Reddy at this address. Even if you make your own travel
arrangements, send the registration fee to her.
OPTIONS:
A. The original Option I, as described in the travel brochure,
has been cancelled. Persons who signed up for this 9-day
group inclusive tour have coach space reserved, but
should confirm this with Ms. Reddy.
B. Group travel 14 days (Option II) $1611.00
(RT air fare Seattle-Tokyo)
Includes: $835.00 RT air fare, Seattle-Tokyo
776.00 Land package (dble. occupancy)
INCLUDES: 13 nights hotel ; Full
Amesican breakfasts daily. Itinerary -
8/17 Seattle to Tokyo
8/18 Arrive Tokyo, transfer to hotel
(New Otani Tower)
8/19 Optional excursion to Nikko,
motorcoach.
8/20 Sightseeing tour of Ginza
8/21 City tour of Tokyo, am.
Afternoon- Conference
8/22 Conference activities
8/23 conference activities
8/24 Shinjuku Gyoen Garden tour, am.
Afternoon for Conference activities.
8/25 Free day.
8/26 Tokyo/Kamakura/Hakone
8/27 Hakone/Atami/Kyoto
8/28 Kyoto
8/29 Kyoto
8/30 Kyoto/Osaka
8/31 Osaka to Seattle
Dinner Plan Option Available with this package ......$300.00
Single Room Supplement .........../...>.>............$350.00
C. APEX fare Stay of 16-31 days $715.air fare, RT from W. Coast
D. Coach fare No restricitons $1004. air fare, RT from W. Coast
E. Option III in travel brochure has also been cancelled.
F. Hotel rates, as described in the travel brochure, still apply
even if you make your own airline reservations. Reserve
rooms though Ms. Reddy.
-------
∂15-Jun-79 0315 CERF at USC-ISI Re: source notification meeting
Date: 15 Jun 1979 0311-PDT
Sender: CERF at USC-ISI
Subject: Re: source notification meeting
From: CERF at USC-ISI
To: PERLMAN at BBN-TENEXA
Cc: PLUMMER at BBN-TENEXA, jmc at MIT-AI, clark at MIT-MULTICS,
Cc: cerf
Message-ID: <[USC-ISI]15-Jun-79 03:11:20.CERF>
In-Reply-To: Your message of 14 Jun 1979 1728-EDT
RADIA -
I'LL BE ON MY WAY TO FT. GORDON AT THE TIME YOU MENTION, BUT
PLEASE PROCEED WITHOUT ME.
VINT
∂15-Jun-79 0824 Ijcai79 at SUMEX-AIM IJCAI-79 Invited Speakers &Panels
Date: 15 Jun 1979 0824-PDT
From: Ijcai79 at SUMEX-AIM
Subject: IJCAI-79 Invited Speakers &Panels
To: IJCAI-79 Attendees:
I N V I T E D S P E A K E R S
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
COMPUTERS AND THOUGHT AWARD:
Dr. David Marr, Massachusetts Inst. of Technology,
"Visual Information Processing: the Structure and Creation
of Visual Representations"
IJCAI-79 DISTINGUISHED SERVICE AWARD
Prof. Bernard Meltzer, formerly Editor-in-Chief of Artificial
Intelligence, Dept. of Artificial Intelligence, Univ. of
Edinburgh, "Some Speculations"
-------------
Dr. Alan Bundy, Department of Artificial Intelligence, Univ. of
Edinburgh, "Solving Mechanics Problems using Meta-Level
Inference"
Prof. Harold Cohen, Department of Art, Univ. of California at
San Diego, "What is an Imageα
Prof. Eiichi Goto, University of Tokyo, "FLATS, a Machine for
Numerical, Symbolic, and Associative Computing"
Dr. Barbara J. Grosz, SRI International, "Utterance and Objective:
Issues in Natural Language Communication"
Dr. Earl Sacerdoti, SRI International, "Problem Solving Tactics"
Prof. Herbert A. Simon, Carnegie Mellon University, "Artificial
Intelligence Research Strategies Strategies in the Light of
AI Models of Scientific Discovery"
Dr. K. Takase, Electrotechnical Laboratory, Tokyo, "Skills of
the Intelligent Robot"
I N V I T E D P A N E L S
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
Title Chairman
Reasoning with Uncertainty Randall Davis, MIT
History of AI:1956-1960 E.A. Feigenbaum, Stanford
Commercial Application Y. Shirai, ETL, Tokyo
of Robotics
Application of Language M. Nagao,
Processing Systems Kyoto Univ.
-------
∂15-Jun-79 1032 Mcgoveran at SUMEX-AIM NSF Proposal Snag
Date: 15 Jun 1979 1032-PDT
From: Mcgoveran at SUMEX-AIM
Subject: NSF Proposal Snag
To: JMC at SAIL
cc: McGoveran
Eamon Barrett just called in reference to your proposal #MCS7405998.
Apparently, there is a final report for another grant, ATM7403400,
dated 9/78, which was not filed. This grant went to the University of
Oklahoma. Until that final report has been filed, all action has
been stopped on your proposal.
If the final report has not been filed within 2 weeks, the application
process will be aborted, and you'll have to start all over again.
Eamon said he did not know what your connection was with the outstanding
report, or with U. of Oklahoma. but you are apparently responsible
for that final report.
Eamon's number is (202)637-5743.
Mary
-------
∂15-Jun-79 1031 DEW arpanet loss loses
John,
If the arpanet left before September I would be greatly inconvenienced.
I spend a lot of time communicating with Hans Berliner and use the net
for this almost everyday. I think my thesis will be much better because
of this communication and I would not like to lose it as I make all the final
changes.
David
∂15-Jun-79 1118 DEK tip first, lab afterwards
At the meeting yesterday I mentioned that I anticipated problems since I
would need to be sending lots of stuff to SCORE. However, on further reflection,
I guess I could live without the tip for a month, as a sacrifice to the
greater good.
∂15-Jun-79 1137 RPG Move
To: JMC
CC: PMF, LES, GLS, LLW, LCW
As you know, Guy Steele and I are writing the S1 LISP this summer.
The proposed move on July 30 falls in the last 2.5 weeks of his stay
here. The S1 LISP is intimately related to the VAX LISP project at
MIT, headed by MRG and JONL. The timing on producing this LISP
is critical - that is, every day is crucial. The project depends on
close communication with the MIT people in order to ensure the
success of both LISPs - they will probably use a lot of our code,
especially the compiler and interpreter (written in LISP and
only some modules machine dependent). I would say that BOTH
LISPs would be seriously threatened by the move occurring during
Steele's last weeks here. Perhaps we would both have to return to
MIT for this period, which would be an inconvenience in terms
of LLL money and personal hardship (my wife is pregnant and would not
be unable to accompany me).
Since I assume that the VAX LISP is of interest to you, I hope you
will consider this situation.
-rpg-
∂15-Jun-79 1139 PERLMAN at BBN-TENEXA mailing list
Date: 15 Jun 1979 1436-EDT
From: PERLMAN at BBN-TENEXA
Subject: mailing list
To: jmc at MIT-AI
You are right. I wanted JNC. Sorry about the messages.
Thank you for telling me--I was wondering about Noel's lack
of response.
radia
-------
∂15-Jun-79 1138 RPG Previous message
Also, I hope it doesn't need to be mentioned that the interaction
with MIT requires the TIP.
-rpg-
∂15-Jun-79 1212 REM via AMES-TIP Moving TIP before SU-AI
Due to moderately small disk allocation I use disk as cache for active files,
using datacomputer as major storage medium. If TIP is moved before SU-AI
(or vice versa) then during that interval I won't have convenient access
to files, so I'll have to use DART backups instead. If magtape drives
are as flakey as they seem to be (see \GRIPES discussion) it'll be a pain.
Of course I'm not working on supported research, so nothing important will
be delayed except data-compression work. If there is enough advance warning
I could pre-fetch everything I plan to be using and DART-fetch whatever
I forget to pre-fetch. -- Is TIP move planned for September, a month before
planned move one month after new music computer? -- I have generally used
TIP rather than dialup because of (1) access to MC without net-hop thru
su-ai and without redialing (2) positive signal if su-ai down rather than
just dead line (3) pressure to not tie up dialup lines. If everybody on
TIP switched to dialup, would dialups be very busy? -- Note that pumpkin
is unreliable, sometimes taking 2 weeks only to find out the file wasn't
restored because of random trouble earlier on tape which must be manually
skipped-over in person. Thus use of dart might require physical trips
to the lab, using gasoline... (moving SU-AI before TIP would cause less
gasoline usage, but I presume that is very unlikely since you want the TIP
moved to SCORE in Jacks)
∂15-Jun-79 1237 REM
To: JMC
CC: REM
P.s. re IRSM -- Rivest keeps saying that each of my ideas isn't new,
that Shannon's original paper on encryption had the same ideas
already. Do you know which paper this would be specifically and
which of my ideas I should look-for in it? I rather doubt that
Shannon invented IRSM 30 years ago, but he may have half-invented
it or some related ideas (like I did around 1971-73 when I invented
the mathematically-pure IRSM which until 1978 was impractical), and
I would like to give credit to co-discoverers of such things, if for
no other reason than to say that one of the most brilliant minds in
computer-science / information-theory was unable in 30 years to
find the numerical trick that made it practical, thus my patent isn't
an obvious idea that anyone working in the field would have found.
I think that Shannon wrote only one paper on cryptography and that it
appeared in Bell System Technical Journal in 1948. I don't recall that
it presented any detailed compression schemes. After you have looked
at it you might ask Rivest if that is all he has in mind.
∂15-Jun-79 1358 ES ARPA net
I would be inconvenient not to be able to use MACSYMA. However, I would rather
bike to Jacks to use MACSYMA than to be permanently in Jacks.
∂15-Jun-79 1412 PAM
I am using the arpanet to maintain and develop the nano-KRL language at
SCORE and Rochester; this is an ongoing project which will be much harder
without the arpanet connection. Perhaps we should move score to the AI
lab building, instead of moving the mountain to Muhammed...Paul
∂15-Jun-79 1411 DBG TIP
As I understand it, the lack of the TIP would make home DM's almost
unusable because of insufficient phone lines. This would be the
main inconvenience to me, since I plan to have a DM by then. However,
the lack of access to MACSYMA might also be a problem.
∂15-Jun-79 1431 BPM ARPAnet access
To: JMC
CC: LES
Having SAIL off the net at any time sounds like a very bad idea, based on the
normal amount of network communication which takes place with MIT, CMU, ISI,
DARPA, etc.
The PSI group would be hurt because we are relying on SAIL for local files
and listings from ISI until our F2 is up. When the F2 gets up stably
(sometime in July or August?) and on the ARPAnet, we will need to access
SAIL over the net to begin transferring files over to the F2.
If the TIP is moved without SAIL, a number of dialup lines have to be moved.
The 4 300-baud lines and 1 1200/1200 line are no problem. The 3 150/1200
lines require modems that live in the same modem rack as SAIL's direct
dialup lines. So who gets that rack, SAIL or the TIP?
All in all, moving SAIL and the TIP together, either in July or later, sounds
preferable to splitting them up.
∂15-Jun-79 1541 SMG via HARV-10 arpanet
To: JMC
CC: DCL
i am writing my thesis on SAIL. since the parts i have already written
use pub and the full stanford character set, it would be very difficult
to transfer my work to another machine for a short time.
until i return to stanford, it will be necessary to use the network.
steve
∂15-Jun-79 1545 DCL ARPAnet availability
To: JMC
CC: LES, FWH, PWM, SMG, DCO
John,
In response to your request for reactions to your proposal to move the Tip
before SAIL moves:
The whole verification group would be seriously hampered by lack of the NET.
Part of our current ARPA proposal (the fiscal 79 one) is to develop
verifiers for export to other research labs.
We are doing this. Currently export versions are being used at
SRI (by Philoco Ford Aerospace), HARV-10, NSA, SCORE, Utah,and
Arlington Texas (by TI).
We depend on the Net for interaction with these people on their problems,
for sending updated versions, for exchanging examples, and for demos. to
ARPA- security workshop at MIT, June 28-29th, and elsewhere.
Export versions are being maintained here using the Net by PWM (who is now in
Australia) and by SMG (who is at HARV-10).
SMGs work, on reports, the RUNCHECK system and his thesis would be cutoff.
Also, I presume you had a plan for alternate accounts at SCORE for NET mail,
which has become the normal mode of communication with us and many outside
sources.
-David
∂15-Jun-79 1617 ARR Inconvenienced by not having ARPA
To: JMC
CC: ARR
I would be inconvenienced: I am using the most recent version of PASCAL at
SCORE, for measuring and tunning up fof a PASCAL compiler. If it is going
to happen, please announce it in advance, so that I can bring and install
it here.
Armando.
∂15-Jun-79 2219 HPM TIP usage
I use MACSYMA almost daily, both interactively via TELNET, and batchwise,
writing and editing Macsyma programs here and FTPing the results
there, FTPing answers back to be inserted into programs or
documents. Some of the work is even on the critical path to my thesis.
I would be severely hampered without frequent access to this resource.
∂15-Jun-79 2227 MRC TIP going away
I'm unhappy about the missing SAIL Arpanet access, but I guess I support it
if it expedites the move to Jacks. SCORE is being screwed daily by not
being in Jacks and not having a local IMP to talk to. One thing that is
worrysome to me are the people who require regular Arpanet access. I
suggest that some temporary phone lines to the TIP (ala my TIP port) be
set up to provide such users uninterrupted, if inconvenient, Arpanet access.
It IS possible to hook up a TIP port to a SAIL TTY line - I have such a line
already for my network debugging purposes. I suggest at least four such
lines.
∂16-Jun-79 0624 BIS Losing NET Connection
I brought with me a number of 9-track tapes. These can't be read at SAIL,
so I'm reading them at SUMEX and moving the data over the NET to SAIL.
Without a NET connection, I can't do this.
∂16-Jun-79 0806 BH via MIT-AI moving the TIP
Naturally it would be a hassle for those of us who use SAIL from
elsewhere! Just how much of a problem it would be for me specifically
depends on which month, which the message didn't say.
∂16-Jun-79 0812 FB Tip move inconvenience
I would be seriously inconvenienced by a month without ARPAnet at SAIL
unless I had adequate warning. With warning I can move all my multipurpose
files to SCORE and be OK.
Seems like the biggest complainer is likely to be DEK. He wants to finish
his Metafont work at Sail this summer and he is working at Sail and moving
things (fonts especially) to Score. That's why some of us wanted to know
if Dialnet could be brought into service.
∂16-Jun-79 1259 WP Moving the TIP before the lab.
I do not use the Net too much usually, however, I see one
problem. Given that the system will be down for quite some time
after the move I probably want to move some of my files to other
net sites and continue working on them. In my case I can both use
the verifier and do text editing on other sites and the move
will coincide with the finalization of my thesis.
I presume that other people woulde like to do similar things.
But without TIP and 9-track tapes we are pretty much isolated
and moving files from here to there becomes painful.
Wolf
∂16-Jun-79 1334 DEA Loss of ARPA net
John,
Since much of my work is done at SRI, loss of the TIP would be
a big inconvenience, but not crippling, since I do have access to other
terminals.
- Doug
∂16-Jun-79 1423 RPG TIP Move and MacLisp maintainance
Should the early move of the tip be occasioned, the maintainance of
MacLisp could be affected. In particular, the sources for all of the
parts of the interpreter, compiler, and assorted other packages are kept at
MIT in a central location. In order to be able to handle any bugs that
come up during the TIP downage would require that the sources for
the interpreter and compiler be kept online here for quick
reference, a set of files in excess of 500k. In addition, we would
not be able to do any re-assemblies here due to the ITS-flavor
of the sources - we could not re-compile the compiler easily either.
These files, if compiled/assembled at MIT take about 10 minutes each
to FTP as it is now, and would be painful with any solution I
can think of. Thus we would need to fix any bugs with FRAID,
which is generally unpleasant.
Finally, though it might be tolerable to live without the
TIP for MacLisp for 1 month, the possibility of this stretching to
more than a month would be bad for us.
I should point out, too, that the verifier crowd
has spread their verifier around, which means that I sometimes
need to debug those versions by net.
-rpg-
∂16-Jun-79 2111 DON
∂16-Jun-79 1900 JMC
Words like co-operate and pre-exist are completely missing from unabridged.
--------------------
Don't blame me!
∂16-Jun-79 2113 DON
I note that "postexist" is included!
∂16-Jun-79 2308 DON dictionary
∂16-Jun-79 2256 JMC pre-exist
The reason I asked you was that I conjecture that these words are missing
because of some way the word list has been computer processed. Do you
know who did that?
--------------------
No, I don't know. But I believe BPM was responsible for the list turning
up here at SAIL, so you might try asking him.
∂17-Jun-79 0937 MINSKY at MIT-AI (Marvin Minsky)
Date: 17 JUN 1979 1239-EDT
From: MINSKY at MIT-AI (Marvin Minsky)
To: jmc at SU-AI
can you call 617-734-3363 about margaret young?
∂17-Jun-79 1023 BIS Manipulator safety in JAX Hall
To: ARG, MSM, JKS
CC: TOB, LES, JMC, DPB
After seeing the Hand/Eye room in JAX and seeing MSM's proposed layout for
it, I wonder if it is wise to install four manipulators in that room.
There will be no way to step back quickly if one of the arms makes a grab
for somebody. During experiments one should always be careful to do
things very slowly, but accidents can happen, and there needs to be a
decent margin for error.
What is your feeling about this?
∂18-Jun-79 1038 JKS tip move
I am currently using the arpa net about 2 times a week to contact macsyma. If
there is some alternate way I could contact macsyma (from some campus location?)
then the early TIP move would be ok. Otherwise, not having access to macsyma
for a month would be quite an inconvenience for me. Ken Salisbury
∂18-Jun-79 1100 JMC*
call kirkman
∂18-Jun-79 1154 RSC LISP Verifier meeting
Derek and I have to leave the Lab for the rest of the day. Should we try to meet
tomorrow morning? I'll call you this evening. -- Corky
∂18-Jun-79 2054 MXB via SRI-KL account
NO, I've been very grateful its been left on this long, since
its handy for news service, etc, but I can understand you shutting
it off. Thanks
MO
∂18-Jun-79 2334 CLT NIHIL EX NIHIL
.r fol
Welcome to the wonderful world of FOL.
New HELP command; try it!
*****DECLARE INDVAR X;
*****DECLARE INDCONST NIL;
*****ASSUME NIL=X;
1 (CALL (NIL . #341456) SUBR -21407 PNAME (#345315 #345314))=X (1)
*****↑C
long live qNIL!!!
∂19-Jun-79 0329 DON via Easy Street
To: BPM
CC: JMC, LES
UNABRD.DIC now updated to include words from POCKET.DIC[TEX,FML]
∂19-Jun-79 1344 Bmoore at SRI-KL (Bob Moore) Why don't we get together...
Date: 19 Jun 1979 1344-PDT
From: Bmoore at SRI-KL (Bob Moore)
Subject: Why don't we get together...
To: jmc at SAIL
to discuss plans for the AI/Phil workshop. Late afternoons other than
Wednesdays are gererally good for me. I'll be happy to meet you at the
Lab at some convenient time.
Bob
P.S. When I was in Boston I heard that Dennett wants to learn LISP while
he is out here, so presumably he would like to have a terminal also.
-------
∂19-Jun-79 1408 Feigenbaum at SUMEX-AIM confusion on the 20 vs VAX front
Date: 19 Jun 1979 1356-PDT
From: Feigenbaum at SUMEX-AIM
Subject: confusion on the 20 vs VAX front
To: fb at SAIL, jmc at SAIL, carlson at ISI, engelmore at ISI,
To: balzer at ISIB, rick at RAND-UNIX
Date: 19 Jun 1979 1345-PDT
From: Rindfleisch
Subject: FYI - HEDRICK (RUTGERS) VIEWS ON PDP-10/VAX
To: FEIGENBAUM, SWEER, GILMURRAY, ADMIN.GORIN at SU-SCORE
8-Jun-79 22:39:00-PST,1669;000000000000
DATE: 8-Jun-79 22:39
FROM: HEDRICK
Subject: Re: future of Tops20
To: RINDFLEISCH
Cc: levy
In-reply-to: Your message of 6-Jun-79 1319-EDT
I have just come back from DECUS. The story I get is that the rumors
you have been hearing are an unintended sideeffect of a little plot
by Gordon Bell. Apparently he was concerned that VAX software was
a disaster area, and was unlikely to get better unless someone from
the 10/20 group got involved. So he tried to stir up interest in
VAX among LCG personnel by saying that this was the system of the
future. This seems to have backfired, in that his comments got a bit
wider circulation than he intended. The latest rumor from him is
that the next machine will be essentially the same hardware for the
20 and VAX, and that the 20 one will be built first. At DECUS they
announced that this machine is now committed, but declined to give
details.
I continue to think that the reports about the demise of the 20 are
fallacious. It is true that software development on the 10/20 may
be slowed slightly be dilution, in that there is now one more system
for DEC's efforts to be spead out over. But Norma Abel claims that
the 10, 20 and VAX have equal software development budgets, and that
for VAX the developers have to do maintenance, while there are
separate maintenance staffs for 10 and 20, so effectively 10 and 20
have more than VAX. The 10 monitor is undergoing major development
work. So is the 20. And if you really believe them about the
new 20 being committed, hardware development is also continuing.
Time will tell, but I think CMU will end up with egg on their face.
-------
-------
-------
∂19-Jun-79 1708 Bmoore at SRI-KL (Bob Moore) Terminals at CASBS
Date: 19 Jun 1979 1707-PDT
From: Bmoore at SRI-KL (Bob Moore)
Subject: Terminals at CASBS
To: jmc at SAIL
I just talked to Peter Hart about borrowing a Datamedia terminal to
use at the Center next year. Peter's position is that currently it
appears that the AI Center has enough terminals that I could borrow
one, but if for some reason a shortage developed I would have to
return it. The situation with respect to 1200 baud couplers is less
clear, but it appears that SRI will probably be able to supply me
with one.
Bob
-------
∂19-Jun-79 1728 PHW at MIT-AI (Patrick H. Winston) Ma Xiwen
Date: 19 JUN 1979 2027-EDT
From: PHW at MIT-AI (Patrick H. Winston)
Subject: Ma Xiwen
To: JMC at SU-AI
Thanks for your recommendation. Without it I would have been
inclined to ignore him on the ground of too many people wanting
to come. He has not actually applied yet, saying only that he
plans to visit sometime this summer. We will try to accomodate
him.
Patrick
∂19-Jun-79 2332 LLW SAIL Moving Date
To: JMC
CC: LLW, LES
Dear John:
I am writing with respect to the posted schedule for moving SAIL down onto
campus, and specifically to plead for a 1-2 month postponement of this
move.
The announced date of July 30 for the decommissioning of the SAIL computer
system, preparatory to moving its hardware down to campus, would have very
unfortunate effects on the ongoing effort to bring up a LISP dialect (NIL,
or what remains of it) on the S-1 architecture. As you know, this effort
is a Summer 1979-only project, as at least 2 of its 3 major participants
(including its leader) are available for this effort only at the present:
Steele will be deep into thesis work next Summer, and Gabriel will be
degreed and gone. I have advertised LISP-on-the-S-1-at-end-FY79
sufficiently widely that, if it can't be delivered by the end of the
current fiscal year, minimization of political losses will require that I
bury it quickly and thoroughly, rather than try to keep it alive for
subsequent work in FY80.
Thus, if there is to be a language supportive of AI work when a S-1 Mark
IIA system comes to SAIL next Spring, it's seemingly going to be the
Steele-Gabriel-Brooks NIL, and NIL will come into existence only if the
team creating it has the SAIL computing facility with which to work
through the end of August.
Also, we (particularly the 2 local SAIL hardware experts, JBR and PMF) had
earlier promised Les to aid and assist any way we could in the SAIL move.
We will be much more available to do so in middle-to-late September (when
we will have completed our Navy review preparatory to releasing our FY80
funds) than we well be in early August (when we will be working
frantically to get into shape for our review).
Therefore, any consideration you may give to deferring the SAIL move by
4-8 weeks will certainly be appreciated. Such a deferral would seemingly
serve your interests as well as ours, and would (to within my limited
knowledge of these matters) entail no penalties to you or SAIL.
Thanks,
Lowell
∂19-Jun-79 2332 FB 10's and 20's versus VAXes
To: feigenbaum at SUMEX-AIM, JMC at SU-AI, REG at SU-AI,
rindfleisch at SUMEX-AIM
It is true that Gordon Bell and many other people at Digital see the VAX
as the only major family of machines in Digital's future. The main problem
right now is software and Gordon is now resorting to more traditional
means of encouraging software: identify some people that desperately need
the VAX address space and make them offers they can't refuse. That is why
you shouldn't be surprised to see VAXes at CMU, Caltech, and Stanford very
soon. Then some of us may have better ideas about what to do about VMS.
∂20-Jun-79 0503 Mark Crispin <Admin.MRC at SU-SCORE> Dialnet work
Date: 20 Jun 1979 0502-PDT
From: Mark Crispin <Admin.MRC at SU-SCORE>
Subject: Dialnet work
To: JMC at SU-AI, Les at SU-AI
I talked with IAZ and he's going to try to do some debugging at the
Business School this weekend. The problem is that he is doing work
for Knuth on TEX which is taking up much of his time.
I would really like some help on developing the user code for Dialnet,
preferably somebody who would take over the active hacking. There is
just too much that has to be worked on simultaneously for one person.
It isn't that DOING any one thing is hard - it's that doing it right
is. If you would like to see any one specific thing - TELNET for
example - I can have something that works in a day or so, but if you
want to have the "right" implementation, ie, something integrated with
the Arpanet software, that is a lot more work.
I also really do want to work for Ralph on SCORE full-time, and I get
the strong impression from Ralph that his reluctance is mostly because
of your unhappiness with Dialnet progress (ie, I am to make you happy
before I'm let off the hook). So, I'd like to discuss what exactly
you from me in more specific terms than "finish Dialnet" - because I
really do want out from an active role. Given the choice, I'd rather
do full-time system programming at SCORE simply because it is more
diversified - if I get stuck on one problem there are hundreds of
other ones which I can divert my efforts too. The problem with
Dialnet is that it is too one-dimensional and after almost two years
I'm pretty burnt-out with it.
Here's what I feel is reasonable to expect out of me:
. File transfer: My current program does work, although the user
interface is abysmal. I should assist IAZ in debugging the Tops-20
version of his program until it works as well.
. MAIL: I should explain to BH whatever he needs for him to put the
Dialnet support into MAIL at SAIL (he really wants to do it instead
of having me do it). I should write the mail receiving end for
Tops-20 and the modifications to MAILER to support Dialnet. I should
have some user interface for MAIL as well, preferably within MM.
. TELNET: I should write some kind of TELNET.
. Monitors: Continue to maintain and debug the Dialnet device service
routines for WAITS and Tops-20 regardless of working full time for
SCORE.
I feel that TELNET has the best chance of being done quickly. MAIL is
a thorny problem - GSB, LOTS, and SCORE are currently skew in the
MAILER programs which are running and it will take some work to put it
back together again. I believe too that I'd be in a better position
to work on Dialnet MAIL working full-time on Tops-20, since it would
fit in ideally with a redesign of the internals of Tops-20 MAIL I've
been contemplating for quite a while, and which would be a major project
worked on in cooperation with MMcM and other people (MMcM and I have
been discussing this idea for a while now).
So, what I propose doing is to write some kind of TELNET
implementation (I already DO have "some kind of FTP implementation")
and be around to answer questions, then I would like out of the
project with the understanding that I'll maintain the code that there
is now (the monitor stuff still does need some tuning) and that I am
committed to seeing MAIL work - in other words my work on the new MAIL
system will include Dialnet from ground 0. And of course that I will
continue to be available to answer questions about how the code works.
I believe it will be best for both me (letting me work on other stuff
that interests me) and the project to get somebody new in it.
How does that sound?
-------
∂20-Jun-79 0554 PJH via LONDON
thanks for yr message. You were right about the control characters, they
were being eaten by the operating system here. This is now fixed.
I wish I could find it in me to be sorry about Minskys defection, but I
cant. I met Dennet and he was rather worried also about spending a whole
year in the same building as Marvin. Replacing him with Ray Turner is
probably not possible, but i will investigate. Other interesting
people might be Brian Smith (but i guess he is too young?) or Yorick
Wilks, but again I bet this cannot now be fixed up becos of other
concerning concepts and belief, dont think too hard about it, i will try to
write something more coherent and give it to you at Michies do in edinburg.
an idea about next year. I was impressed at the idea which produced the
"beyond leibnitz" memo( if not at the content), viz. a file of discussion
and argument, to be edited later. Could we do something similar? No doubt
it would get read over arpanet, and maybe we could even let outsiders a
put in their 2 cents worth, why not? There seem to be 2 main
problems (1) getting the technical facilities, especially terminals at
the Center (2) getting the philosophers to actually do it. Can you
do anything about the former?
the job i now (as of this october) have is a senior lectureship ,
as opposed to a mere lectureship. Its roughly like the difference
between assistant and associate prof., i gather. It makes very little
difference to daytoday activities(ie i still teach, etc.) but i get
more clout in decision-making, get in on more longrange planning, and
get a bit more money. Chiefly, if I DONT get promoted by my age then Im
marked as a loser. For example, this is the route (via a readership)
to a chair, which is where im aiming in (say) 8-10 years. However,
it more and more looks like i will want to leave Essex before then.
All the bright people are leaving this place, and whats left is grey.
The funding prospects are not sufficiently good to enable one to build
a lab. on research money alone, except just possibly in very applied
robotics. In any case im not really suited by temperament to be the
thrusting, dynamic head of a lab.
I also hope we/you can get Sloman to come over for at least part of the
time. i think he would be stimulating.
Other names just occured to me as minsky replacements: Herb Simon,
maggie Boden, randy davis, drew mcdemott.
∂20-Jun-79 1206 MFB kl → m.j.hall
my own personal interest is that it move as late as possible.
im in the middle of my thesis implementation, and will soon
be in the middle of thesis writing.
any sustained down time is going to screw me up.
how long willit be down for (from a user's point of view) ?
martin
∂20-Jun-79 1219 PAT
John, have you ever received this?
∂13-Apr-79 1554 JMC
Please get me "Integer Programming" by T.C. Hu from the library.
---
Is checked out to someone, Richard has put a recall on it and will
call me when it is returned.
∂20-Jun-79 1250 DEK KL move date
To: JMC at SU-AI, EAF at SU-AI
From my own standpoint, everything would be much better if the KL were to move
at a date ≥ September 1. The compromise that John spoke of last week (moving
the tip on Sep 1 and the Lab on Oct 1) sounds most reasonable to me based on
all the inputs I've heard; but the decision isn't mine, of course, so I'm
merely sending this note as part of the data for John's decision.
Every day I get to use the KL during the period now to Sep 1 is "critical path
time" for me, and I am expecting to be a heavy user during this time since
METAFONT is coming on the air.
∂20-Jun-79 1327 BPM SAIL move
To: JMC
CC: LES
I realize that this would negate some of the advantages of moving either
as early or as late as possible, but moving between August 15 and October 1
makes sense as far as impacting the work of as few people as possible.
The autumn quarter starts around Oct. 1, so moving after that will be
suboptimal. The summer quarter ends around Aug. 15, which includes
deadlines for things like thesis submission and oral exams. In my case,
my orals are scheduled for Aug. 8, so I will be working hard right up
until that date on code, thesis draft, slides for the oral presentation,
etc. Through approximately Aug. 15, others will be preparing talks for
IJCAI too.
∂20-Jun-79 1336 RPG Move
To optimize the chances of getting the S1 Lisp running this summer,
Between August 23 and September 10 would be best for the S1 Lisp group.
-rpg-
∂20-Jun-79 1353 RAK Moving date
August 16 is the deadline for summer submission of theses which I strongly
hope to meet (as undoubtdly do sveral others). Therefore, I propose that
the moving date be at least August 17.
Dick
∂20-Jun-79 1619 LGC ARPA Access to SAIL
∂15-Jun-79 1022 JMC
Whoever would be seriously inconvenienced by a month without ARPAnet if
occasioned by moving the TIP before moving the Lab should reply to JMC.
---------------
Since Stanford lies outside my local telephone calling area, use of the
Ames TIP is currently the only practical means I have of working on the
SAIL computer from home, something which I do a lot. For this reason, it
would be a great help to me if the period of SAIL inaccessibility via
Ames TIP were minimized, and occurred during the month of August.
∂20-Jun-79 1812 Mark Crispin <Admin.MRC at SU-SCORE> SAIL Arpanet downtime
Date: 20 Jun 1979 1807-PDT
From: Mark Crispin <Admin.MRC at SU-SCORE>
Subject: SAIL Arpanet downtime
To: JMC at SU-AI
This is a message sent to me irt SAIL Arpanet downtime -
---------------
Mail from SU-AI rcvd at 20-Jun-79 1709-PDT
Date: 20 Jun 1979 1706-PDT
From: Dick Sites <RLS at SU-AI>
Subject: arpanet
To: MRC at SU-AI
I'm working on optimizer for S-1 project remote from San Diego. If the
Arpanet is down for a month, that would cut into my work, depending on the
actual dates involved. We will be finished by Oct 1, but will be working
heavily over the summer. A two-week gap is probably no sweat, so long
as we know the dates early enough to work around them.
Dick Sites / U.C.S.D.
---------------
-------
∂20-Jun-79 1827 Mark Crispin <Admin.MRC at SU-SCORE>
Date: 20 Jun 1979 1828-PDT
From: Mark Crispin <Admin.MRC at SU-SCORE>
To: JMC at SU-AI, MRC at SU-AI
Cc: LES at SU-AI
In-reply-to: Your message of 20-Jun-79 1247-PDT
Well, as I have said many times I already have a file transfer program
working, and I believe that my responsibility there now is mostly to
help Ignacio bring his program up on Tops-20. I explained about the
problems with MAIL - BH is possessive about the SAIL MAIL program and
it'll take a while for me to learn enough about its internals to do it
myself anyway. You DO want to say .MAIL G.Gorin@LOTS, don't you? On
the Tops-20 end, the problem is not so much incoming Dialnet MAIL as
it is outgoing Dialnet MAIL - several programs are involved in this.
LOTS, GSB, SCORE, and the outside world are all skew with respect to
mail systems for Tops-20. I am working on making SCORE's way of doing
things the Stanford standard, but even SCORE's way needs to be cleaned
up a bit.
I agree that telnet is pretty useless for Dialnet, but there aren't
any implemention problems there - doing it "right" on Tops-20 involves
such horrible monitor changes that it's acceptable to do it wrong.
I believe Dialnet is in good shape for somebody else to take over the
main development effort from me. The difficult and hairy monitor code
involving getting the data over reliably, creating and deleting processes,
etc. all works. There is also a good set of debugging tools for Dialnet
that I have written, and there is also some documentation (although I
admit it does need to be gone over a bit). I really don't think you can
get that much more out of me and pressing me is just going to make things
worse in the long run.
-------
∂20-Jun-79 1918 HPM M.J.H. move
If we move in the next few months, my thesis will either be
delayed by about two months, or else be finished off without actual
obstacle avoiding runs. If the move is put off until October or November,
I can almost certainly finish without problems.
I want to complete my degree this December. I had intended that my
final demonstrations, which would be filmed and edited into an expository
movie, would involve obstacle avoiding runs of the cart outdoors.
The work is currently held up by a hardware problem with the cart
steering motor (when our last low speed motor burned out, and was replaced
by a faster one, i hastily and poorly designed a speed control servo to
regulate its speed. This circuit has ruined itself several times. I have a
new and much better design, but it awaits a fancy tachometer, which has
been ordered).
In a month or two the new hardware should be working, and I can
begin debugging the final path planning parts of the cart program, then
spend another 2 or 3 months doing and filming long runs. After that,
barring the need for re-takes, I should no longer require the physical
cart, and the lab can move without seriously inconveniencing me.
If we move before the final runs, it would require somehow finding
and setting up an environment for the cart at least 100 ft square, that's
free of moving objects like people. It would be necessary to run at least
two coax cables to the vicinity of this area from the SAIL computer, for
the CB and TV antennas, plus many other details. I can't imagine all that
getting done in less than a few months, in the crowded environment of the
campus. Possibly it can't be done at all. If it looks too difficult, I
might opt to finish of my work in pretty much it currnt state, world
mapping, but not obstacle avoiding. This would make me very unhappy, since
I've promised myself, all these many years, not to quit until the cart
successfully and reliably ran obstacle courses.
∂20-Jun-79 2037 BIS Best Regards
To: JMC, LES
Just got a note from Philip R Bagley, now of Philadelphia PA,
who, upon learning of my coming to Stanford, asked me to give you two
his best regards -- if you're "still here".
Phil gave me my first programming job.
∂20-Jun-79 2144 RPG Solution
I suggest that you make your comments as below. The reason
is that "|" is a string delimiter, and the COMMENT mechanism cannot
affect the reader whereas ";" can.
-rpg-
;;; (<Move> ::= (S|F <number>)|(<number> <number>)
;;; <number> ::= 1|2|3|4. The first designates a move from
;;; the hand to a storage or final pile, and the second designates a
;;; move from a storage pile to a final pile.
∂21-Jun-79 0824 Ijcai79 at SUMEX-AIM IJCAI Travel:Visas
Date: 21 Jun 1979 0824-PDT
From: Ijcai79 at SUMEX-AIM
Subject: IJCAI Travel:Visas
To: to:
cc: Buchanan
A reminder that you will need a visa for travel to Japan.
A TOURIST visa to Japan is much easier to obtain than a
visa for business reasons.
The IJCAI-79 Program Committee
-------
∂21-Jun-79 0950 Bmoore at SRI-KL (Bob Moore)
Date: 21 Jun 1979 0949-PDT
From: Bmoore at SRI-KL (Bob Moore)
To: jmc at SAIL
Thursday (today) at 4:00 will be fine.
Bob
-------
∂21-Jun-79 0952 Dbrown at SUMEX-AIM (Response to message)
Date: 21 Jun 1979 0943-PDT
From: Dbrown at SUMEX-AIM
Subject: (Response to message)
To: JMC at SU-AI
In response to your message sent 20 Jun 1979 1504-PDT
I'm not sure. But possibly. I have done next to nothing re visitors.
I have been unable to determine what can be done, given our space
shortage. I won't be back in California til about 5 July.
I can answer about hung then. -Denny
-------
∂21-Jun-79 1000 CLT*
meeting at noon in room 383p
I will be going down around 11:40 if anyone wants a ride --- Carolyn
∂21-Jun-79 1207 ARK S-1 Meeting Notice
To: S1 at SU-AI
CC: pickering at SUMEX-AIM
There will be an S-1 meeting on Wednesday, 27 June at 1:30 pm in 301 Margaret
Jacks Hall. HAD and KO have been removed from the S-1 mailing list.
Arthur
∂21-Jun-79 1247 PAT todorovich
is at LLL, he will be at Hoover Institute about 5pm and I gave him
directions to your house.:
∂21-Jun-79 1249 RPG
∂21-Jun-79 1107 JMC
Is there a good way to edit the value of a variable using (edit)?
You can say (EDITV <variable>).
-rpg-
∂21-Jun-79 1253 PN Pail-5,Ptrans
To: S1 at SU-AI
1. PAIL5, which describes the U-code that SOPA will accept, is now up.
XSPOOL PAIL5.DOC[DOC,S1]/nohead. XSPOOL PAIL5.XGP to get a cover for it.
2. A P-code to U-code translator has been written, called PTRANS[SOP,S1].
It will be used primarily to produce Ucode to help in testing the
interpreter and the optimizer until the new compiler is stable.
Relevent new DO files in [DO,S1]
PUSE: Pascal to P-code to U-code to Interpreter.
FUSE: Fortran to U-code to Interpreter.
∂22-Jun-79 1212 ARK S-1 meeting: location corrected
To: S1 at SU-AI
CC: pickering at SUMEX-AIM
The S-1 meeting will be on Wednesday, 27 June at 1:30 pm at the AI Lab
Small conference room, not at Margaret Jacks as originally announced.
Arthur
∂22-Jun-79 1327 LES Dastek visit
To: FB, JBR
CC: JMC
Some people from a newly-formed company called Dastek plan to visit
SAIL on Tuesday, June 26, at 3pm to see what we have and to discuss
their planned entry into the high-density disk business.
They are apparently an IBM spin-off -- their President (who will be
here) supposedly played a central role in the development of IBM's
latest thin-film recording scheme.
The person who contacted me was Jim Koerber, who used to be the IBM
rep at Stanford. They are apparently open to the idea of planting
some of their stuff here. John McCarthy would like to look into the
possibility of a big CSD disk system, probably hung on the Ethernet.
It appears to me that their offerings might also be of interest to
the S-1 crew.
So, show up if you are interested. It might be best for me to give
them a 20 minute tour first, so we will probably get underway around
3:20.
∂22-Jun-79 1330 HVA (1) Sunday, July 8th; (2) Carolyn Talcott-Summenr Quarter
(1) Bechtel has just informed me that facilities will be available on
Sunday, July 8th, from 2-4 p.m.(it had to be moved up because of I.Center
dinner), and I have confirmed this to Martha Coulton--we will probably meet
at Bechtel late Mon.afternoon (6/25),to discuss details.
(2) Carolyn Talcott - Summer Quarter. I have to take list of Su. Qtr.SRA's
to Graduate Study office before 3 p.m. TODAY. My notes (of many weeks ago),
tell me that you wanted to put Carolyn on your NSF Basic Research Grant for
Summer (I gather the IBM fellowship is used up); wld. you pls. confirm aye or
nay. Thank you.
Carolyn should be on my NSF or ARPA according to whichever is convenient
but for the summer less the 5 or 6 weeks that she will be working for
IBM. (It won't necessarily exactly correspond to the amount of time I
will be working for IBM). Ask her for an estimate on Monday.
∂22-Jun-79 1458 CSD.JJF at SU-SCORE 1980 AI Qual
Date: 22 Jun 1979 1449-PDT
From: CSD.JJF at SU-SCORE
Subject: 1980 AI Qual
To: Buchanan at SUMEX-AIM, Feigenbaum at SUMEX-AIM, Lenat at SUMEX-AIM,
To: JMC at SU-AI, TW at SU-AI
Cc: Kunz at SUMEX-AIM, csd.jjf at SU-SCORE
In beginning our preparation for next year's AI Qual, we
would like to streamline the reading list so that it is both
(1) representative of the field and
(2) can be covered adequately in a
reasonable length of time.
This fall we will start a formal study program, and
we would like to work on the reading list this summer.
Could you let us know who the next chairman will be so that we
might work with him on this?
John Kunz
Jeff Finger
-------
∂22-Jun-79 1628 MDD SAIL MEMO
To: PAT
CC: JMC
John has suggested that two short memos I've written here be combined to form
(a) a SAIL MEMO and (b) a paper to be submitted to a special forthcoming issue
of Artificial Intelligence. He said you could take care of merging them.
They are stored in my file [1/MDD] as REP and CRIT. REP "requires"
PUBMAC.PUB from my file which in turn calls on some fonts still on tape.
I'll use "PUMPKIN" (see LES if you don't know about PUMPKIN) to try to get
them restored (probably will happen next week). Thanks and have a good summer.
∂22-Jun-79 1808 MDD future addresses
To: PAT
CC: HVA, JMC
Until July 8:(mail drop only) Computer Sccience Division; Evans Hall,
University of California,Berkeley,CA 94720
July 9-13:Courant Institute,251 Mercer Street,New York,N.Y.10012,(212)-460-7100
July 16-27:495 Littleton Street,West Lafayette,IN 47906,(317)-743-2206
August 1-31: Berkeley address same as above
After September 15:New York address same as above
∂22-Jun-79 1904 REM Shannon's original paper
To: JMC at SU-AI, RIVEST at MIT-ML
The only paper by Shannon in Bell Sys Tech J for 1948 was his original
paper on communication theory. The Markovian model was introduced,
although no optimized variable-length-context model was proposed, no
Huffman or better (IRSM) code was suggested, and no mention of
cryptography (except that cr. books were cited because they contain
tables of bigram etc. frequencies). One invention of his is fractured
english obtained by picking letter at random from book, then repeatedly
opening book at random and using left-context (fixed depth) to select
one more random letter from book. Other than the general idea of
using a Markovian model and computing entropy for it (and rather bad
ideas for obtaining an entropy match for said Markovian model), and
the analogy between transformer (impedance match) and cruncher (entropy
match), I didn't see anything relevant to my inventions.
∂22-Jun-79 2232 Feigenbaum at SUMEX-AIM AI Qual for next year
Date: 22 Jun 1979 2231-PDT
From: Feigenbaum at SUMEX-AIM
Subject: AI Qual for next year
To: buchanan, lenat, jmc at SAIL, tw at SAIL
The troops want a leader.
The following excuses apply:
McCarthy is on sabbatical the whole year
Buchanan is on sbbatical in Winter and Spring
Feigenbaum is on sabbatical in the Fall and Chairmaning in Winter and Spring
Winograd just did it
Do we have a solution by the rule-out heuristic?
Comments???
Ed
-------
∂23-Jun-79 1034 DEW lab moving date
I have two projects in the works: my thesis and an AIJ article. I hope to
have them both done by September 1. Therefore it would be very helpful
if the lab did not move til after that date. Despite the optimistic
predictions on the ease of bringing the system back up at Jacks, I think
my work would be disrupted for some weeks by the move. Dave
∂23-Jun-79 1656 Rindfleisch at SUMEX-AIM DEC GIFT VAX - LICENSE ARRANGEMENTS
Date: 23 Jun 1979 1650-PDT
From: Rindfleisch at SUMEX-AIM
Subject: DEC GIFT VAX - LICENSE ARRANGEMENTS
To: FEIGENBAUM, FB at SU-AI, JMC at SU-AI,
To: ADMIN.GORIN at SU-SCORE
DEC is trying hard to ship a preliminary VAX by the end of June (11/780,
250K bytes of memory, 2 RK07 disks, 8 TTY lines). They need a no cost
purchase order that I'm trying to arrange with Stanford purchasing. The
question of patent or copyright ownership for systems developed on that
machine has come up with Marvin Geistlinger (new head of SU purchasing).
a) Does anyone have a prior agreement with DEC about this?
b) What are your opinions about what should be written into the agreement?
Tom R.
-------
Has D.E.C. has made some proposals that have to be responded to?
Unless D.E.C. raises an issue, nothing need be done,
since we incur no obligation from accepting a gift. When D.E.C. gave us
the KL-10, nothing was said and nothing happened. More to the point is
what software they will let us use and whether we can get sources, but
this can be negotiated later.
∂23-Jun-79 2325 Feigenbaum at SUMEX-AIM (Response to message)
Date: 23 Jun 1979 2324-PDT
From: Feigenbaum at SUMEX-AIM
Subject: (Response to message)
To: JMC at SU-AI, rindfleisch, admin.gorin at SU-SCORE,
To: EAF at SU-AI
In response to the message sent 23 Jun 1979 1801-PDT from JMC at SU-AI
My attitude is exactly the same as John's. Don't raise any issue re a gift
unless DEC does. A gift is a gift. The fewer strings the better (preferably
zero).
Ed
-------
∂23-Jun-79 2335 Feigenbaum at SUMEX-AIM a bit of jmc history
Date: 23 Jun 1979 2328-PDT
From: Feigenbaum at SUMEX-AIM
Subject: a bit of jmc history
To: jmc at SAIL
John, when did you leave Dartmouth f
for MIT? I'm writing a brief history of early AI for IJCAI
proceedings and want to get that straight. Do you know
when Marvin left the Society of Fellows at Harvard for MIT?
Ed
p.s. I'll have to get some special tourist advice from you re
Kathmandu,i.e. how to stay healthy and have a good time and
how to eat without getting sick.
-------
I went to M.I.T. as a Sloan Fellow in September 1957 while still on
Dartmouth faculty. I didn't return to Dartmouth but stayed as
Assistant Professor of Communication Sciences in the EE Dept. I
don't know when Marvin left Society of Fellows, but I assume he
was a fellow for the standard period, which I believe is three
years.
∂24-Jun-79 0905 FB Digital gift
To: JMC at SU-AI, EAF at SU-AI, admin.gorin at SU-SCORE,
rindfleisch at SUMEX-AIM
The gift of a small configuration VAX 11/780 is part of the package that I
have been negotiating with Digital for a large VAX system for computer aided
design for LSI. Digital wants to get some of that discount into this fiscal
year's tax writeoff so they have to ship something by the end of this month
(the end of their fiscal year). Since we don't have any ARPA charge numbers
for this project yet, they are giving us the first part of the system with
discounts to be adjusted on later augmentations. This is an understanding
with Digital saying that if something goes wrong we get the gift anyway.
The configuration is 256KB of main store, 2 RK07's (20MB each), eight terminal
lines, and (I hope) some kind of tape drive. We will put it in ERL 318 and
move it later the MJ computer room. My arrangements with Digital include all
of their software and sources that they give away or sell. They are making
similar shipments to CMU (2 small systems) and Caltech (1 small system).
We are planning a meeting for late July to discuss system software issues.
The major issues are Digital's stand on UNIX and whether VMS can be made into
a reasonable system. Who wants to go? (Probably in Pittsburgh or Washington)
Forest
∂24-Jun-79 1355 Feigenbaum at SUMEX-AIM VAX VLSI/CAD facility proposal
Date: 24 Jun 1979 1354-PDT
From: Feigenbaum at SUMEX-AIM
Subject: VAX VLSI/CAD facility proposal
To: kahn at ISI, fb at SAIL, jmc at SAIL, rindfleisch,
To: admin.gorin at SCORE
cc: engelmore at ISI
Friends,
Herein is my first draft of the proposal for the VAX facility that has been
much discussed among us. Please send me comments so that I can quickly
revise it, get it through the Sponsored Projects Office, and off to Bob
Kahn as soon as possible. Note that it is short and to the point.
Forest and Tom should feed me the exact configuration (specify the gift
portion, the first augmentation, and the second augmentation).
Tom and I will iterate on the budget and get it to you pronto.
Right now I just need comments on the text.
Thanks,
Ed
-------------------------------------------------------------------------
INTRODUCTION
The Stanford University Computer Science Department herein proposes
the acquisition, maintenance, and operation of a computer facility to
support the development of a network-based resource for VLSI
computer-aided-design (CAD) programs. Equipment purchased under the
proposed contract will augment a substantial gift of equipment by
Digital Equipment Corporation to Stanford for VLSI CAD research,
thereby leveraging the proposed ARPA funding by a substantial amount.
The equipment proposed is a DEC VAX11/780 configuration.
LONG TERM OBJECTIVES
The nation currently faces major opportunities, and serious
scientific problems, in the move to VLSI circuitry. The DOD
(particularly DARPA) and the NSF are both moving to seize these
opportunities and solve the problems. Major computer science groups,
such as ours, are proposing to contribute to this effort. The
national interest is best served if the fruits of their work,
represented as software packages, are made available to the national
community of scientists and engineers working on these problems. Only
in this way can the science and technology cumulate in the nation
rapidly enough to make the necessary impact.
Experience of almost a decade has shown that the best way to make
this software available to a national community of interest is via
the national computer networks (of which the ARPAnet is the major
example). The evidence from the various ARPA-supported research
laboratories, and from the NIH-supported SUMEX-AIM National Computer
Resource, makes this abundantly clear.
Thus, our long range objective is to establish and operate a
network-based resource for CAD software of the VLSI era. It will have
a network-graphics orientation. It will be operated as a national
resource, along the lines of the successful SUMEX-AIM model, and will
be available to all researchers, whether supported by DOD or by NSF.
It is planned that the national resource will utilize much higher
speed VAX equipment from Digital Equipment Corporation that will
become available sometime toward the end of the coming three year
period. The equipment proposed herein is relatively low-capacity and
therefore will support only the software development work and
experimental network use (to develop and test the network software
systems). A follow-on proposal for the national resource will be sent
in the third year.
For development purposes, we will use the current ARPAnet. What
network to use for the national facility will become clearer as
events unfold, particularly with respect to the new wide-band
network.
SHORT TERM OBJECTIVES
This particular proposal asks for funds to establish and operate an
experimental VAX11/780 facility to support research on CAD software
at Stanford University and experimentation with the intended network-
based graphics access.
Specifically our objectives over the next three years are:
1. to augment the forthcoming DEC gift of a VAX11/780 with sufficient
main memory, disc memory, ARPAnet access electronics, and peripherals
to make the CAD research viable.
2. to use the facility in support of the work of four Stanford groups
on CAD: the Heuristic Programming Project group working on layout;
the Artificial Intelligence Laboratory group working on an augmented
SUDS (Stanford University Drawing System) system; and groups of the
Computer Systems Laboratory and Information Systems Laboratory
working on CAD at the architectural and functional specification
levels. The software produced by these groups will constitute the
core of the software made availble in the ultimate national resource.
3. to develop the necessary network driver software packages that
will allow the graphics interaction with the CAD software on the
network chosen for the national resource.
4. to adapt existing VAX software, and/or develop new base-level
software, to support the programming necessary for (2) above. The
VAX-line software is not yet adequate in this regard.
MILESTONES
October,1979: complete the initial augmentation of the VAX
configuration. (Meeting this milestone is conditional on DEC's
ability to deliver the necessary equipment, but if there is any
slippage it will not be long).
December, 1979: complete a second-stage of augmentation of the
configuration. Make final decisions regarding the base-level software
(operating system and languages).
March, 1979: Complete integration of graphics display equipment with
VAX configuration.
September, 1980: First demonstration of VLSI/CAD software on VAX.
September, 1981: Second demonstration of VLSI/CAD software on VAX.
March, 1982: Complete the plan for the national resource, including
selection of the appropriate machine, the appropriate network, and a
management structure for the national access. This plan will be
formulated in consultation with ARPA-IPTO, and will lead to a
follow-on proposal for the national resource.
September, 1982: Third demonstration of VLSI/CAD software with
experimental network-graphics use.
MANAGEMENT PLAN
The Principal Investigator will be Professor Edward A. Feigenbaum. He
is Chairman of the Computer Science Department as well as one of the
research investigators using the facility. He has experience as
Principal Investigator of the SUMEX-AIM National Resource and as
former Director of the Stanford Computation Center.
The facility director will be Mr. Thomas Rinsfleisch. He has been
Director of the SUMEX-AIM facility since its inception, and will
continue in that role as he adds the proposed facility to his
responsibilities.
There will be a steering committee, chaired by the P.I., and
consisting of:
Mr. Ralph Gorin, the Computer Science Department's Director of
Computer Facilities.
Professor Forest Baskett
Professor John McCarthy
The steering committee will discuss and decide all major issues, and
advise the P.I. on non-major issues.
When the national resource comes into being, the management structure
will be changed somewhat, along the lines of the SUMEX-AIM structure
that has worked so well.
The facility will be housed in the computer room of the new Computer
Science Department building at Stanford, and will serve graphics
terminals located elsewhere on campus by local digital network.
EQUIPMENT CONFIGURATION: INITIAL (GIFT) PLUS PROPOSED (*)
(here follows the configuration)
BUDGET
(Here follows the budget)
-------
∂24-Jun-79 2104 CLT predpar substitution seems to have the catching bv
disease
To: RWW, CLT, JMC
DECLARE INDVAR r s t x;
DECLARE PREDCONST R 2;
DECLARE PREDPAR A 2;
AXIOM RE: ∃s.A(s,r);;
∧i RE[A←λx t.R(x,t)];
∧i RE[A←λx s.R(x,s)];
SHOW PRF→PROOF.FOL;
%1 ∃s.R(s,r) ∧I RE
%2 ∃s.R(r,r) ∧I RE
∂25-Jun-79 0421 MRC via Dialnet Dialnet TELNET
To: JMC, LES
This message is being sent via Dialnet TELNET from SAIL over the kludge
port. If this isn't proof that Dialnet works, I don't know what is.
Obviously, there is both a server and a user program up. To run it, you
say R DTLNET. It will ask you for a host name. Type S for SAIL, L for
LOTS and G for GSB. Currently there is no hardware at LOTS and GSB
doesn't have a server yet. I'll try to get one at GSB sometime soon.
I don't have much control about hardware at LOTS.
Are you convinced now that Dialnet works and is in good shape?
∂25-Jun-79 0742 PJH via LONDON dates and turner
The later date seems more likely to be acceptable as is v. close
to the end of term, a nd that endofterm is a slack time. But not
crucial.
About theposibility of Ray Turner visiting. There is a possibility
of his getting away from here for the second half of the year, ie
starting in april-may until september or october. I know this is
not optimal from point of view of getting intensive group discussions,
but in the event of not finding somebody reasonable for the whole year,
which seems quite likely this late, consider the scenario where two
people arrive halfway through the year to bring fresh ideas to the
group. Ie use one years money for two halfyears. Turner and
Sloman? I think this might be quite a god scenario for several reasons
(1) insurance against staleness in the discussions (2) new arrivals
will have some prepared positions to react against (3) preparation
for their arrival might be good discipline for the group at a crucial
point in the year.
correction to last msg: ray turner could maybe come late March 80.
Bob:
Here is a message from Pat Hayes who was advocating Ray
Turner. My own inclination, perhaps from laziness, is to postpone
action on this until the group meets - or until and if Lindzey gets
the money he has asked from the Sloan Foundation. What do you
think?
∂25-Jun-79 1103 Moore at SRI-KL (J Moore) TAK over the Integers
Date: 25 Jun 1979 1104-PDT
From: Moore at SRI-KL (J Moore)
Subject: TAK over the Integers
To: JMC at SU-AI
Cc: RWW at SU-AI
I am sorry that you haven't had time to entertain the
question of TAK's termination when viewed as a function over
the Naturals instead of over the Integers. I agree that
(SUB1 0) = 0 throws a wrench into the behavior of my version
of TAK near 0. However, I thought it interesting that a
lexicographic combination of linear measures explained a
close cousin of your version of TAK. (And even more
interesting that a proof had been done mechanically. After
all, as you pointed out at IJCAI-77, it matters less how
well women preach and dogs walk about on two legs than that
they do it at all!)
When I got your response I decided to have a go at proving
the termination of TAK over the Integers, formalizing the
intuitions that lead me to believe that the lexicographic
measure I'd found didn't take unwarranted advantage of
(SUB1 0) = 0.
A simple extension of the lexicographic measure for the
Naturals works for the Integers. Recall that for the
Naturals, my argument went like this (informally): In each
recursive call of TAK, either (a) the call computes out
without further recursion, or (b) the max of the args gets
closer to 0, or (c) the max stays fixed and the first
argument gets closer to 0.
The extension to the Integers is just: either (a) the call
computes out without further recursion, or (b) the max of the args
gets closer to the min, or (c) the max stays fixed with respect to the
min and the first argument gets closer to the min.
Our theorem prover has proved the theorems implied by (a),
(b), and (c). For this exercise we defined the Integers as
pairs of Naturals (standing for their difference), and
defined Integer less-than, difference, and plus in terms of
pairwise operations on Naturals.
J
P.S. I am writing up a brief note on the exercise and will
send it along when it's done. I'd also like to send it to
whomever else you might think interested. May I have the
address of the author of TAK and anybody else you think
might be interested?
-------
∂25-Jun-79 1102 RWW bugs
To: CLT, JMC
I am sorry about not responding sooner but I have been pretty much staying at
home with my kids. I will be in tues and deal with it all thanks
richard
∂25-Jun-79 1335 Bmoore at SRI-KL (Bob Moore) Hayes's idea about Ray Turner
Date: 25 Jun 1979 1336-PDT
From: Bmoore at SRI-KL (Bob Moore)
Subject: Hayes's idea about Ray Turner
To: jmc at SAIL
If the Center does have money that they were planning to spend on Minsky
the idea of having two new people come half way through the year sounds
reasonable, but:
Since Minsky has an endowed chair, my guess is that he was going to be
more or less free for the Center, in which case his not coming does not
create any new funds--we could check with Lindzey on this.
Half way through the year ought to be the 9 month academic year, i.e.,
January rather than March or April. I for one will be going back to SRI
in June, since I don't have any independent support to pay another three
months' salary. I would think that several of the others would be in
the same position (including Pat!)
In short, I think it would be a good idea to approach Lindzey about the
possibility of having some people join us for half the year right away,
since the Center seems to require long lead times. My guess is that he
will turn us down flat, but if he agrees there should be a general discussion
of who to invite. My own first choice would be Mitch Marcus, and I know
that Dennett is very high on him also.
Why don't you give Lindzey a call and see what he says.
--Bob
P.S. On the conference dates, we ought to check whether either of the
proposed dates is a football weekend, which might make it difficult to
obtain motel rooms for visitors. Could you get Patty to check on that?
-------
∂25-Jun-79 1416 CLT equiv
it is almost ready to tex. I'll finish it tonight
∂26-Jun-79 0109 DLW at MIT-AI (Daniel L. Weinreb)
Date: 26 JUN 1979 0408-EDT
From: DLW at MIT-AI (Daniel L. Weinreb)
To: jmc at SU-AI
CC: hwc at SU-AI
I was forwarded a request from you for comments from anyone who
would be inconvenienced seriously by the ArpaNet's being down for a month.
I still use MIT-AI for mail receiving and connections to people
at MIT with whom I (and other S-1 people) discuss issues of
the S-1 operating system; I generally log into MIT-AI about once
a day for a short session. So unless we get our own Arpanet connection
up quite soon, I would be seriously inconvenienced.
Thank you.
∂26-Jun-79 0137 LES
∂25-Jun-79 2257 ME LLL drive fixed
To: LES, JBR, LCW, HWC, TM, ROB
Al Barrett of Ampex came out and replaced head 17 of the LLL drive, and
now it works. Unfortunately, we need to replace or fix the even head
preamp of the same drive, since the orginal one in that drive was swapped
into another drive a little while ago to keep the other drive going. So
the LLL drive isn't up yet, but it will be as soon as it has a good even
head preamp (Rob is taking care of that -- he may try to fix it, having
gotten some good clues from Al Barrett).
∂26-Jun-79 1336 RWW
To: CLT, JMC
This is not a disease but (as usual!!!) a feature that you have used many
times. Let me illustrate with ordinary mathematical induction.
P(0) ∧ ∀n.(P(n)⊃P(n+1)) ⊃ ∀n.P(n)
The usual kind of predicate parameter you use is
P←λn.A(n)
This produces
A(0) ∧ ∀n.(A(n)⊃A(n+1)) ⊃ ∀n.A(n);;
Notice all the "caught" bound variables. You get the same result if
[P←λm.A(m)]. This is because the λ-variables are considered to be
the "formal" arguments of the predicate parameter - they specify
what variables in the body of the λ-expression are to be instantiated
with the arguments of the predicate parameter AS IT APPEARS in the axiom.
Thus the substitution allows bound variables to be caught. However,
if the formal parameters conflict in some other way the bound variables
are changed. Consider the folowing which I believe is exhaustive.
*****∧I INDUCT[P←λm.m+a=a+m];
3 ((0+a)=(a+0)∧∀n.((n+a)=(a+n)⊃((n+1)+a)=(a+(n+1))))⊃∀n.(n+a)=(a+n)
Here m is the parameter and a is free. Notice n remains the same.
*****∧I INDUCT[P←λm.m+n=n+m];
4 ((0+n)=(n+0)∧∀n1.((n1+n)=(n+n1)⊃((n1+1)+n)=(n+(n1+1))))⊃∀n1.(n1+n)=%
(n+n1)
Here m is the parameter and n is free. This produced a change in bound
variables to leave n free.
*****∧I INDUCT[P←λm.∀n.m+n=n+m];
5 (∀n.(0+n)=(n+0)∧∀n.(∀n1.(n+n1)=(n1+n)⊃∀n1.((n+1)+n1)=(n1+(n+1))))⊃∀%
n n1.(n+n1)=(n1+n)
Here m is the parameter and n is bound. Notice that it was changed so
as not to conflict.
*****∧I INDUCT[P←λm n.m+n=n+m];
6 ∀n.(((0+n)=(n+0)∧∀n1.((n1+n)=(n+n1)⊃((n1+1)+n)=(n+(n1+1))))⊃∀n1.(n1%
+n)=(n+n1))
Here both m and n are λ-variables. Since P only takes 1 argument the
remaining ones are treated as universal parameters. This answers John's
question about the parameters in the replacement axiom of ZF.
*****∧I INDUCT[P←λn m.m+n=n+m];
7 ∀m.(((m+0)=(0+m)∧∀n.((m+n)=(n+m)⊃(m+(n+1))=((n+1)+m)))⊃∀n.(m+n)=(n%
+m))
Notice the effect of changing the order of the parameters on the change
of bound variable and the instantiated axiom.
I believe that this explains the problem reported by Carolyn. This may
all seem complicated but I believe (says Richard defensively) it is
coherent, correct and comprehensive. The commands for creating these steps
are below.
MARK BEGIN;
DECLARE INDVAR n m a b c;
DECLARE PREDPAR P 1;
DECLARE OPCONST + 2[INF];
DECLARE PREDCONST A 1;
AXIOM INDUCT: P(0) ∧ ∀n.(P(n)⊃P(n+1)) ⊃ ∀n.P(n);;
∧I INDUCT[P←λn.A(n)];
∧I INDUCT[P←λm.A(m)];
∧I INDUCT[P←λm.m+a=a+m];
∧I INDUCT[P←λm.m+n=n+m];
∧I INDUCT[P←λm.∀n.m+n=n+m];
∧I INDUCT[P←λm n.m+n=n+m];
∧I INDUCT[P←λn m.m+n=n+m];
MARK END;
I understand the examples of your message, which are all correct
but irrelevant, because the predicate parameter of the mathematical
induction schema has only one argument. The replacement schema, which
gave rise to the problem, Carolyn's example and my elaboration of it
involve predicate parameters with two arguments. Had you but read her
example, . . .
∂26-Jun-79 1429 RWW bug
yes! I apologize I believe there is a bug but it is not
essentially a catch bound variables bug although it looks that way.
I am in the process of fixing it. Please read my message anyway
as it describes about axiom schemas with parameters.
richard
∂26-Jun-79 1608 RWW BUG
Did you get my message about the bug. It is not with change of
bound variables but instead it does repeated substitution rather
than simultaneous substitution. This is clearly incorrect.
I am determining the best way to fix it now.
richard
Yes, and I assume you got mine.
∂27-Jun-79 0927 DEW thesis
John,
I'm nearing the end of my rewriting of my thesis to include everyone's suggestions.
I agreed with your comments in WILKIN[S79,JMC] and incorporated them.
Two things: 1) if you have any other suggestions or requests for the final draft,
please let me know this week, and 2) I'm not sure of the procedure that should
be followed. Unless you want it some other way, I plan to bring you a velo-bound
copy of the final draft in about 2 or 3 weeks and then come around in about a week
to get your signature on the necessary forms. There will of course be time for
minor changes.
By the way, many sections have been rewritten since the draft you have. Chapter
1 and the first half of chapter 2 have been completely reorganized and should
be much clearer now.
Cheers, Dave
I leave for Edinburgh on July 8 and will be at IBM Research Laboratory
in Yorktown Heights, N.Y. from July 16 to about the beginning of
September, so it looks like part of this will have to be done by mail.
∂27-Jun-79 1022 DEW
I also have to mail the signature sheets to Hans at CMU so they will be
making the rounds. Perhaps I can get you a final (or nearly so) draft
by July 8, and you can sign the forms before you go with my word of honor
that I won't use the forms for 3 weeks and not even then if you have any
objections.
∂27-Jun-79 1335 PN Proposal for Version markers in Pascal programs
To: S1 at SU-AI
This is a proposal for a version comment convention for Pascal programs
that will be transported to the S-1. Besides adding clarity, it will
enable the versions to be changed automatically through a program I have
written. It will also be easy to convert to the version feature of the
Ucode compiler when that comes up.
All code that should be compiled by the DEC-10 compiler should be
bounded by the comments (*D10...*) (*...D10*). Similaraly, code
to be compiled by PCPASC would have (*S1...*) (*...S1*).
The conversion program will work for any other version markers of the
same format.
If I don't hear any comments to the contrary in a week, I will assume
that this new convertion will be adopted. I also plan to change the
program header page to wipe out any mention of SLAC.
∂27-Jun-79 1418 REM via SU-TIP
To: JMC, RWW
See 79627A.WRU[1,REM]
∂27-Jun-79 1445 DEK moving day
To: REG
CC: JMC
I ran into John this noon and asked if he had decided about moving day.
He said there were a convincing number of cases to make the conclusion
that the tip shouldn't move before the lab. So the day for Lab+Tip to move
is now October 1, as I understand it.
Les says there is somebody named Ron (?) who also is trying to make plans
for fitting things into the computer room in Jacks... I'm send this note
in an attempt to improve communications, in case they are as shaky as they
appear to be between so many busy people!
∂27-Jun-79 2238 RPG
∂27-Jun-79 2224 JMC broken differently
Now (refile count (solit lsp)) elicits the reply
;(OPEN (SOLIT LSP DSK (S79 JMC)) IN) Wrong number of arguments to form.
Right. OPEN is a built in LISP function that opens a file in NEWIO.
I suggest renaming yours or not expecting LISP to be able to open files.
-rpg-
∂27-Jun-79 2329 RPG
∂27-Jun-79 2325 JMC
REFILEV seems to delete the rest of the file into which refile takes place.
You're right. You might try doing a LOAD-FILE on that file [(load-file foo bar)]
which might fix that, or just use another file for the
variables. Otherwise it will take about an hour's worth of hacking,
which is not convenient at the moment since I'm working on the S1 LISP
fairly strongly tonight.
-rpg-
∂28-Jun-79 0435 PJH via LONDON martindavis
john, in haste, about martin davis papr for special issue. I like the
idea very much, but last word must be dannys. There is a time problem,
so let me suggest that you act as primary reviewer, I as secondary.
bring a spare copy to edinburgh, the mails are v. slow. OK? seeu , pat.
∂28-Jun-79 1558 Bmoore at SRI-KL (Bob Moore) Ascribing Mental Qualities to Machines
Date: 28 Jun 1979 1553-PDT
From: Bmoore at SRI-KL (Bob Moore)
Subject: Ascribing Mental Qualities to Machines
To: jmc at SAIL
cc: hendrix
John,
Thank you for reminding me about your "Ascribing Mental
Qualities..." paper. I did, in fact, read an early version
of it about three years ago, but this was quite a bit before
I began the serious thinking about philosophical and
semantical problems involving belief that led to our paper.
The most interesting point of comparison between the two
papers seems to be between your notion of a second order
structural definition and (1) our discussion of interpreting
a physical system as having an internal language, and our appeal
to functional explanations of (2) the notion of a belief set
and (3) the notion of a sentence of a natural language
expressing a formula of an internal language. In each of
these three cases, we raise the question of whether a
"higher" level of description can legitimately be applied to
a system that is specified at a "lower" level. We could
state our position in your terms by treating each of the
higher level descriptions as a second order structural
theory of the system as described at the lower level, and
claiming that for the second order theory to be true, there
must be one "best" first order theory that satisfies it.
Your citation of Putnam in regard to second order theories
is particularly interesting, because my ideas about
functional theories in psychology have been strongly
influenced by Putnam, although I don't think I have read the
particular paper you cite.
In addition to this interesting technical point, you and we
of course share more or less the same particular model of
belief, which we said (p. 7) "should be familiar to anyone
acquainted with developments in artificial intelligence or
cognitive simulation over the past few years." You may be
interested to know that during his recent visit to
Stanford, Putnam discussed essentially the same model,
calling it the "Carnap-Reichenbach model", attributing its
invention to the work of those two in the forties. (I
suppose your "Programs with Common Sense" was the first
place it was explicitly set forth in the AI literature.)
Beyond these points of commonality, though, I think that the
basic thrusts of your paper and ours are really quite
different. As you point out, ascribing mental qualities to
machines is the converse of reduction, and it is reduction
that specifically concerns us. Moreover, we discuss
computational models of belief not so much for their own
sake, as to be able to make some more general points about
the relationship between psychology and semantics. If you
have any comments to make about these aspects of our paper,
I certainly would enjoy discussing them with you, and I am
sure Gary would too.
Bob
-------
∂28-Jun-79 1721 PAT Edinburgh flights
To: JMC
CC: PAT
Sunday July 8
British Airways 286 leave SFO 7:30pm arrive Heathrow 1:30pm
British Airways 4812 leave Heathrow 5:40pm arrive Edinburgh 6:50pm
Sunday July 15
British Airways 4733 leave Edinburgh 9:40am arrive Heathrow 10:50am
Pan Am 101 leave Heathrow 1:30pm arrive Kennedy 4:05pm
The reservations were made through Franklin Hirsch, Dina Bolla Travel Agency,
134 Town and Country Village, 329-0950.
Please check into the following: It occurs to me that it might be
advantageous for me to take a flight that stops in NY and leave some
baggage in NY, since I am not coming back to California till the end
of summer. Is there a reaonably good connection that will do it and
still get me to Edinburgh Monday evening? Does Hirsch know whether
I can check a suitcase or two at Kennedy for a week? Anyway, please
try to find out about it.
∂28-Jun-79 1900 JMC*
Church reference
∂29-Jun-79 0157 Feigenbaum at SUMEX-AIM AI Lab's design project budget
Date: 29 Jun 1979 0157-PDT
From: Feigenbaum at SUMEX-AIM
Subject: AI Lab's design project budget
To: jmc at SAIL, les at SAIL
cc: kahn at ISI, fb at SAIL, rindfleisch
Let me try to summarize in a few words (for Les' proposal writing effort) the
fruits of numerous bilateral and multilateral conversations I have had recently.
Try to keep the SAIL design project budget in the range of $100-$150 Thousand
per year for three years. Try to keep the budget to about the $100K level
in Year 1, so as to give us all a little more flexibility on the VAX machine
acquisition in year 1 ( time in which the total monies available
to us are more limited than they will be later).
Ed
-------
∂29-Jun-79 1023 Jrobinson at SRI-KL (Jane Robinson) Dinner tomorrow nnight
Date: 29 Jun 1979 1023-PDT
From: Jrobinson at SRI-KL (Jane Robinson)
Subject: Dinner tomorrow nnight
To: jmc at SU-AI
Just got back from vacation and want to finnd out results of your checking
with Caroline about proposal we get together Saturday evening, my house,
for supper. If we can, what time would suit you?
Jane
-------
∂29-Jun-79 1420 ARK S-1 Meeting Notice
To: S1 at SU-AI
CC: pickering at SUMEX-AIM
The next S-1 Meeting will be on Thursday, 12 July at 1:30 pm in the SAIL
small conference room.
Arthur
∂29-Jun-79 1645 Buchanan at SUMEX-AIM Request for help at IJCAI
Date: 29 Jun 1979 1615-PDT
From: Buchanan at SUMEX-AIM
Subject: Request for help at IJCAI
To: Session heads:
cc: ijcai79
I would very much appreciate your chairing one of the technical sessions
at IJCAI in Tokyo. Having you participate will enhance the
quality of the meetings, and distribute the overall effort
of making the conference run smoothly.
More papers were submitted for this conference than for IJCAI-77 at
MIT and fewer were accepted, so we expect you would find many
interesting talks.
I'd be glad to try to accommodate your preference for times if I can.
Would you please let me know as soon as possible whether or not you
can help us out. I hope you will.
Sincerely,
Bruce Buchanan,
Program Chairman
-------
∂29-Jun-79 1903 LCW SAIL Move
What is the latest decision regarding the date that SAIL will move?
Curt
A final decision will be announced when reached, but the move won't be
before Sept. 1 and may be substantially later.
∂30-Jun-79 0906 FB proposals
To: LES, JMC
We need the ARPA design system proposal. Is the AI Lab going to be a cost
center? If so, what is the cost? LLL will need to know and approve. I
am told that the administrative requirements will make a cost center a
slow thing to realize.
∂01-Jul-79 0048 Feigenbaum at SUMEX-AIM various
Date: 1 Jul 1979 0043-PDT
From: Feigenbaum at SUMEX-AIM
Subject: various
To: jmc at SAIL
cc: les at SAIL
1. Can you live with a 100K budget for the design project for the first year?
Budget can rise by 35-40K in years 2 and 3. Are these amounts enough?
2. Forest agreed to be Acting Chairman for the Fall.
3. I still don't like (very much don't like) the Oct. 15 move date for SAIL.
It's really the celay of the TIP move that distresses me so much.
Ed
-------
∂01-Jul-79 0207 LLW Fourth of July Party
To: "@S1P[1,LLW]" at SU-AI
This is (relatively) Early Notice of a Fourth-of-July Party next Sunday, 8
July, starting at 1300 at Del Valle Regional Park, continuing there until
approx. 1800, and proceeding thereafter at Firmwedge Manor. *YOU* (and a
spouse or vaguely similar-type companion) are invited! Picnic-type
goodies will be provided at both locations, so come hungry; burgers, dogs,
pop, beer, basic salad and cake/ice cream will be provided--you should
feel free to embellish these staples with anything that will appeal to
your colleagues. RSVP to CDO@SU-AI, or risk running out of food. Basic
game materials will be provided, but you should bring any exotic (legal)
ones which you believe will appeal more. Cindy will provide detailed
directions; general ones are to take Interstate 580 east from the Bay Area
toward Stockton, get off on Vasco Rd and take it to its end on Tesla Rd.
Go right on Tesla until Mines Rd appears on the left; follow the signs to
Del Valle Regional Park, and head toward the principal source of
commotion. Firmwedge Manor is located at 639 Hazel St. in Livermore.
Lowell
∂01-Jul-79 1632 REM via AMES-TIP Quote of the day
To: RPG
CC: JMC
I showed DOCTOR to a 9-yr old black kid who lives in EPA and let him
play with it a little on my terminal. After a while he said
"This guy's really dumb".
(DOCTOR flunked the 9-yr old Turing test sort-of)
∂01-Jul-79 1722 Wiederhold at SUMEX-AIM Parties ...
To: S1 at SU-AI
Date: 1 Jul 1979 1722-PDT
From: Wiederhold at SUMEX-AIM
Subject: Parties ...
To: s1 at SAIL
cc: pickering
In addition to celebrating the fourth of July we propose an expansion of
our horizons:
On Bastille day , July 14th , you are all invited to the annual ( we did miss
last year, but ... ) S1 picnic on Kings Mountain - better to see Skylab -
at Gio and Voy Wiederhold.
The formal time is 12 - 4 pm, individual flexibility exists of course,
but no guarantees of goodies.
The address is 155 Marine Road, Woodside; the phone number is 851-8363.
We'd appreciate an RSVP, to get an approximate headcount, and will mail
a map to those interested.
Come all and enjoy ...
Gio
-------