/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␈↓ 	cNovember 4, 1977 




␈↓ ↓H␈↓Prof. Andrei P. Ershov
␈↓ ↓H␈↓Computation Center
␈↓ ↓H␈↓Novosibirsk 630090
␈↓ ↓H␈↓USSR

␈↓ ↓H␈↓Dear Andrei:

␈↓ ↓H␈↓        I'm␈αsorry␈αI␈αcouldn't␈αmake␈α
it␈αto␈αthe␈αIFIP␈αconference␈αthis␈α
year,␈αand␈αyou␈αcouldn't␈αcome␈α
to␈αthe
␈↓ ↓H␈↓AI␈α⊂conference.␈α⊂ In␈α⊃my␈α⊂opinion,␈α⊂this␈α⊂AI␈α⊃conference␈α⊂was␈α⊂the␈α⊂best␈α⊃yet,␈α⊂especially␈α⊂in␈α⊂the␈α⊃area␈α⊂of
␈↓ ↓H␈↓problem solving systems.

␈↓ ↓H␈↓        I␈αhave␈αbeen␈αfurther␈αpursuing␈α
the␈αideas␈αin␈αthe␈αpapers␈αI␈α
gave␈αyou␈αin␈αRepino,␈αwith␈αonly␈α
small
␈↓ ↓H␈↓advances beyond the results reported in the papers.

␈↓ ↓H␈↓        I␈α
enclose␈α
a␈α
prospectus␈α
for␈α
a␈α
one␈α
year␈α
study␈α
on␈α
arti≡cial␈α
intelligence␈α
and␈α∞philosophy␈α
which
␈↓ ↓H␈↓the␈αCenter␈α
for␈αAdvanced␈αStudy␈α
in␈αthe␈αBehavioral␈α
Sciences␈αhas␈αasked␈α
me␈αto␈αorganize.␈α
 Candidates
␈↓ ↓H␈↓for␈α
participation␈α
include,␈α
besides␈α
myself,␈α
Daniel␈α
Dennet␈α
of␈α
Tufts␈α
University,␈α
Margaret␈α
Boden␈αof
␈↓ ↓H␈↓Sussex␈α
University␈αwho␈α
has␈αjust␈α
written␈αan␈α
excellent␈α
popular␈αbut␈α
constructively␈αcritical␈α
book␈αon␈α
AI
␈↓ ↓H␈↓but␈α⊃probably␈α⊃can't␈α⊃come,␈α⊃and␈α⊃Zenon␈α⊃Pylyshyn␈α⊃of␈α⊃the␈α⊃University␈α⊃of␈α⊃Western␈α⊃Ontario,␈α∩an␈α⊃AI
␈↓ ↓H␈↓oriented theoretical psychologist.  A really outstanding Soviet candidate could be considered.

␈↓ ↓H␈↓        Here is some additional news:

␈↓ ↓H␈↓        1.␈α
The␈α∞Fletcher␈α
Jones␈α
Foundation␈α∞has␈α
endowed␈α
a␈α∞chair␈α
at␈α
Stanford,␈α∞and␈α
Don␈α∞Knuth␈α
has
␈↓ ↓H␈↓been␈αappointed␈αto␈αit.␈α Jones␈αwas␈αone␈αof␈αthe␈αfounders␈αof␈αan␈αearly␈αsuccessful␈αsoftware␈αcompany␈αand
␈↓ ↓H␈↓was killed in an airplane crash.

␈↓ ↓H␈↓2.␈α
Knuth␈α
is␈α
spending␈α
a␈α
good␈α
part␈αof␈α
his␈α
time␈α
this␈α
year␈α
developing␈α
an␈α
automatic␈αtype-setting␈α
system
␈↓ ↓H␈↓good␈αenough␈αfor␈αhis␈αbooks.␈α He␈αwants␈αto␈αget␈αa␈αVideocomp␈αtypesetter␈αfrom␈αIII␈αthat␈αwill␈αpermit␈αus
␈↓ ↓H␈↓to␈α∂produce␈α∞good␈α∂mathematical␈α∂quality␈α∞books.␈α∂ Winston's␈α∂recent␈α∞book␈α∂on␈α∂AI␈α∞was␈α∂typeset␈α∂by␈α∞the
␈↓ ↓H␈↓Videocomp.

␈↓ ↓H␈↓3.␈α
Dialnet␈α
is␈αproceeding␈α
well.␈α
 Here␈α
is␈αa␈α
draft␈α
of␈αthe␈α
low␈α
level␈α
protocols.␈α They␈α
are␈α
subject␈αto␈α
much
␈↓ ↓H␈↓further discussion before even an experimental implementation.

␈↓ ↓H␈↓3. Dave Poole and friends have ≡nished building Foonly at III and are now debugging it.

␈↓ ↓H␈↓5.␈α⊃Another␈α⊃grooup␈α⊃of␈α⊃Stanford␈α⊃students␈α⊃have␈α∩designed␈α⊃a␈α⊃big␈α⊃computer␈α⊃called␈α⊃the␈α⊃S-1.␈α∩ It␈α⊃is
␈↓ ↓H␈↓intended␈α
to␈αbe␈α
a␈α
multi-processor␈αmachine␈α
with␈α
each␈αprocessor␈α
about␈α
the␈αspeed␈α
of␈α
a␈α370/168.␈α
 They
␈↓ ↓H␈↓have␈α
or␈α∞are␈α
about␈α
to␈α∞get␈α
enough␈α
money␈α∞for␈α
a␈α
16␈α∞processor␈α
system.␈α
 The␈α∞≡rst␈α
processor␈α∞is␈α
about
␈↓ ↓H␈↓half␈αbuilt.␈α We␈αare␈αmaking␈αa␈αproposal␈αto␈αdesign␈αand␈αimplement␈αa␈αtime-sharing␈αsystem␈αfor␈αit,␈αand
␈↓ ↓H␈↓αProf. Andrei P. Ershov␈↓ ¬TNovember 4, 1977␈↓ 
nPage 2␈↓ 

␈↓ ↓H␈↓would␈α
get␈α
a␈α
system␈α
in␈α
addition␈α
to␈α
money␈α
for␈α
people.␈α
 The␈α
machine␈α
has␈α
a␈α
30␈α
bit␈α
address␈α
to␈αthe␈α
byte
␈↓ ↓H␈↓level,␈α∞but␈α∞it␈α∞is␈α∞more␈α∞like␈α∞a␈α∞PDP-10␈α∞in␈α∂structure␈α∞than␈α∞like␈α∞a␈α∞370.␈α∞ I␈α∞think␈α∞it␈α∞embodies␈α∂the␈α∞right
␈↓ ↓H␈↓approach␈α
to␈αmulti-processor␈α
architecture␈α-␈α
each␈α
processor␈αhas␈α
independent␈αcontrol␈α
and␈αcan␈α
address
␈↓ ↓H␈↓the whole of memory.

␈↓ ↓H␈↓6.␈α
The␈α∞LOTS␈α
time-sharing␈α∞system␈α
is␈α
now␈α∞working␈α
well,␈α∞and␈α
Ralph␈α
Gorin␈α∞is␈α
doing␈α∞an␈α
excellent
␈↓ ↓H␈↓job␈α
as␈α
manager.␈α
 We␈α
now␈α∞have␈α
512K␈α
words␈α
of␈α
core,␈α
2␈α∞disk␈α
channels,␈α
and␈α
3␈α
disks␈α∞each␈α
holding
␈↓ ↓H␈↓200␈α
megabytes.␈α It␈α
runs␈αvery␈α
well␈αwith␈α
about␈α45␈α
logged␈αin␈α
users,␈αand␈α
we␈αwill␈α
shortly␈αtry␈α
it␈αout␈α
with
␈↓ ↓H␈↓about 20-25 more terminals.  We expect its performance will still be adequate.

␈↓ ↓H␈↓7.␈α⊃Rumor␈α⊃has␈α⊃it␈α⊃that␈α⊃the␈α⊃next␈α⊃generation␈α⊃of␈α⊃Intel␈α⊃and␈α⊃Zilog␈α⊃microprocessors␈α⊃will␈α⊃have␈α⊂large
␈↓ ↓H␈↓address␈α∞spaces␈α∂-␈α∞some␈α∞say␈α∂up␈α∞to␈α∞32␈α∂bits␈α∞which␈α∞will␈α∂provide␈α∞room␈α∞for␈α∂growth␈α∞and␈α∂for␈α∞upwards
␈↓ ↓H␈↓compatible machines.  Other rumors contradict these.  We'll know early next year.

␈↓ ↓H␈↓        How␈α⊃are␈α⊂things␈α⊃in␈α⊂Novosibirsk?␈α⊃ Is␈α⊂there␈α⊃any␈α⊂scienti≡c␈α⊃computer␈α⊂much␈α⊃better␈α⊃than␈α⊂the
␈↓ ↓H␈↓BESM-6 in the o≠ng?  How is it with big disks and with time-sharing?

␈↓ ↓H␈↓        How␈αare␈αyou␈αand␈αhow␈α
are␈αNina␈αand␈αVasya␈αand␈αAnyuta?␈α
 I␈αthink␈αyou␈αare␈αvery␈αfortunate␈α
in
␈↓ ↓H␈↓your children.


␈↓ ↓H␈↓Best regards,



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