/LMAR=0/XLINE=3/FONT#0=BASL30/FONT#1=BASI30/FONT#2=BASB30/FONT#10=BAXM30/FONT#11=ZERO30/FONT#3=STA200/FONT#4=NGB25
␈↓ ↓H␈↓␈↓βS␈↓∧ Artificial Intelligence Laboratory, STANFORD UNIVERSITY, Stanford, California 94305

␈↓ ↓H␈↓∧Telephone 415 497-4430␈↓ 	dDecember 7, 1978 




␈↓ ↓H␈↓Dr. William Carlson
␈↓ ↓H␈↓O≠ce of Information Processing Techniques Defense Advanced Research Agency
␈↓ ↓H␈↓1400 Wilson Blvd.
␈↓ ↓H␈↓Arlington, VA 20009

␈↓ ↓H␈↓Dear Bill:

␈↓ ↓H␈↓        As␈αI␈αsaid,␈αI␈αthought␈αyour␈αsummary␈αof␈αthe␈αsense␈αof␈αthe␈αmeeting␈αwas␈αexcellent.␈α The␈αmeeting
␈↓ ↓H␈↓itself␈α⊃was␈α⊂useful␈α⊃in␈α⊂making␈α⊃known␈α⊂what␈α⊃everyone␈α⊃was␈α⊂up␈α⊃to.␈α⊂ I␈α⊃think␈α⊂people's␈α⊃plans␈α⊃will␈α⊂be
␈↓ ↓H␈↓a≥ected␈αin␈αthe␈αlong␈αrun,␈αand␈αa␈αcertain␈αamount␈αof␈αduplicated␈αe≥ort␈αwill␈αbe␈αavoided.␈α Of␈αcourse,␈αit
␈↓ ↓H␈↓was a meeting of the tops of a lot of icebergs, and the bottom parts will be a≥ected more slowly.

␈↓ ↓H␈↓        As␈α
I␈α
mentioned,␈α
I␈α
would␈α
much␈α
like␈α
the␈α
opportunity␈α
to␈α
try␈α
out␈α
one␈α
or␈α
more␈α
LISP␈α
machines␈α
at
␈↓ ↓H␈↓Stanford.␈α It␈α
seems␈αto␈α
me␈αthat␈α
it␈αis␈αnow␈α
demonstrated␈αthat␈α
the␈αmachine␈α
can␈αbe␈α
an␈αe≥ective␈αtool␈α
for
␈↓ ↓H␈↓developing␈α∪and␈α∪running␈α∪large␈α∪symbolic␈α∪computation␈α∪programs␈α∪and␈α∪that␈α∪it␈α∪willl␈α∀provide␈α∪an
␈↓ ↓H␈↓e≥ective␈α
environment␈αfor␈α
a␈αuser␈α
who␈α
has␈αsu≠cient␈α
access␈αto␈α
one.␈α
 It␈αneeds␈α
to␈αbe␈α
seen␈α
how␈αmany
␈↓ ↓H␈↓people can e≥ectively share one, since they are still quite expensive.

␈↓ ↓H␈↓        Here are some comments on the contents of the meeting:

␈↓ ↓H␈↓        1.␈α∞I␈α∞do␈α∞not␈α∞share␈α∞the␈α∂view␈α∞that␈α∞environment␈α∞is␈α∞everything,␈α∞and␈α∞language␈α∂is␈α∞unimportant.
␈↓ ↓H␈↓Xerox␈αPARC␈αhas␈αcreated␈αan␈αextremely␈αpleasant␈αprogramming␈αenvironment␈αbut␈αat␈α
great␈αexpense
␈↓ ↓H␈↓per␈αuser.␈α This␈αenvironment␈αhas␈αin∨uenced␈αmany␈αattempts␈αat␈αemulation,␈αprominently␈αat␈αM.I.T.␈α It
␈↓ ↓H␈↓remains␈α
to␈α
be␈α
seen␈α∞how␈α
important␈α
quality␈α
of␈α
facilities␈α∞is␈α
compared␈α
to␈α
the␈α
number␈α∞of␈α
reasonable
␈↓ ↓H␈↓terminals␈αand␈αamount␈αof␈αcomputer␈αtime.␈α An␈αexperiment␈αto␈αdetermine␈αthe␈αe≥ect␈αof␈αsuper␈αfacilities
␈↓ ↓H␈↓on rate of program development would be worthwhile.

␈↓ ↓H␈↓        Moreover,␈α∞I␈α
think␈α∞that␈α
LISP␈α∞as␈α
a␈α∞language␈α
for␈α∞computing␈α
with␈α∞symbolic␈α∞information␈α
has
␈↓ ↓H␈↓many␈α∂advantages␈α∂that␈α∂have␈α⊂not␈α∂been␈α∂realizedd␈α∂in␈α∂any␈α⊂other␈α∂languages␈α∂so␈α∂far.␈α∂ An␈α⊂attempt␈α∂at
␈↓ ↓H␈↓formulating these was included in my ␈↓↓History of LISP␈↓ paper which I enclose.

␈↓ ↓H␈↓        2.␈α
I␈α
think␈α
it␈α
would␈α
be␈α
worthwhile␈α
to␈αtry␈α
to␈α
bring␈α
up␈α
ADA␈α
in␈α
a␈α
time-sharing␈αenvironment
␈↓ ↓H␈↓and provide it with analogous facilities to LISP.  These facilities should include

␈↓ ↓H␈↓                a.␈α∃A␈α∃standard␈α⊗representation␈α∃for␈α∃the␈α∃ADA␈α⊗language␈α∃in␈α∃ADA␈α⊗data␈α∃structures.
␈↓ ↓H␈↓Moreover,␈αlists␈αof␈αADA␈α
expressions␈αmust␈αalso␈αbe␈αrepresentable␈α
and␈αcomputable␈αwith,␈αso␈α
that␈αthe
␈↓ ↓H␈↓language␈α∩essentially␈α∩has␈α∩to␈α∩contain␈α∪LISP.␈α∩ Primitives␈α∩for␈α∩composing␈α∩and␈α∪decomposing␈α∩ADA
␈↓ ↓H␈↓expressions must also be available and standard.

␈↓ ↓H␈↓                b.␈α∞It␈α∞must␈α∞be␈α∂possible␈α∞to␈α∞run␈α∞single␈α∞ADA␈α∂statements␈α∞in␈α∞analogy␈α∞with␈α∞the␈α∂ability␈α∞to
␈↓ ↓H␈↓αDr. William Carlson␈↓ ¬VDecember 7, 1978␈↓ 
nPage 2␈↓ 

␈↓ ↓H␈↓evaluate␈α
single␈α
just␈α∞de≡ned␈α
LISP␈α
functions.␈α∞ Since␈α
ADA␈α
has␈α∞declarations␈α
and␈α
block␈α∞structure␈α
(I
␈↓ ↓H␈↓presume)␈α∂while␈α∞LISP␈α∂is␈α∞single␈α∂level,␈α∂there␈α∞will␈α∂be␈α∞additional␈α∂problems␈α∞in␈α∂making␈α∂a␈α∞convenient
␈↓ ↓H␈↓system.␈α∃ This␈α∃problem␈α∃is␈α∃one␈α∃reason␈α∀why␈α∃the␈α∃outcome␈α∃of␈α∃a␈α∃"friendly␈α∃ADA␈α∀environment"
␈↓ ↓H␈↓experiment cannot be predicted a priori.

␈↓ ↓H␈↓                c.␈α∪The␈α∩full␈α∪ability␈α∩to␈α∪transfer␈α∩AI␈α∪applications␈α∩will␈α∪require␈α∩the␈α∪ability␈α∪for␈α∩ADA
␈↓ ↓H␈↓programs␈α
to␈α
create␈αnew␈α
ADA␈α
code,␈αattach␈α
it␈α
to␈αwhatever␈α
is␈α
used␈αfor␈α
property␈α
lists␈αand␈α
to␈α
run␈αit
␈↓ ↓H␈↓from inside the program.  However, this facility of LISP is not used in all AI programs.

␈↓ ↓H␈↓                d.␈α
These␈α
facilities␈αneed␈α
to␈α
be␈αa␈α
separable␈α
ADA␈αpackage,␈α
because␈α
their␈αassociated␈α
large
␈↓ ↓H␈↓run-time␈α⊃routines␈α⊃will␈α⊃not␈α⊃be␈α⊃wanted␈α⊃in␈α⊃ordinary␈α⊃uses␈α⊃of␈α⊃ADA␈α⊃that␈α⊃often␈α⊃have␈α⊃to␈α⊃be␈α⊂space
␈↓ ↓H␈↓e≠cient.

␈↓ ↓H␈↓        Thanks for inviting me to the meeting.


␈↓ ↓H␈↓Best regards,



␈↓ ↓H␈↓John McCarthy
␈↓ ↓H␈↓Director
␈↓ ↓H␈↓Professor of Computer Science