/LMAR=0/XLINE=3/FONT#0=BAXL30/FONT#1=BAXM30/FONT#2=BASB30/FONT#3=SUB/FONT#4=SUP/FONT#5=BASL35/FONT#6=NGR25/FONT#7=MATH30/FONT#8=FIX25/FONT#9=GRKB30
␈↓ α∧␈↓␈↓ u1


␈↓ α∧␈↓␈↓ ∧\THE REAL COMPUTER REVOLUTION

␈↓ α∧␈↓␈↓ ¬xProf. John McCarthy
␈↓ α∧␈↓␈↓ ¬.Artificial Intelligence Laboratory
␈↓ α∧␈↓␈↓ ¬?Computer Science Department
␈↓ α∧␈↓␈↓ ¬8Stanford University, California

␈↓ α∧␈↓␈↓ αTI'll␈αbegin␈αmy␈αtalk␈α
about␈αthe␈αcomputer␈αrevolution␈α
by␈αsaying␈αthat␈αthere␈α
has␈αnot␈αbeen␈αany;␈αit␈α
is
␈↓ α∧␈↓still␈αin␈αthe␈αfuture.␈αIn␈α
fact,␈αwhile␈αthere␈αis␈αa␈α
literary␈αconvention␈αthat␈αtechnology␈αis␈αaccelerating␈α
faster
␈↓ α∧␈↓and␈αfaster␈αand␈αpeople␈αare␈αhaving␈αdifficulties␈αadapting␈αto␈αit,␈αin␈αterms␈αof␈αtechnology␈αaffecting␈αdaily
␈↓ α∧␈↓life,␈α⊃there␈α⊂has␈α⊃been␈α⊂a␈α⊃slow␈α⊂down␈α⊃and␈α⊂not␈α⊃a␈α⊂very␈α⊃recent␈α⊂slow␈α⊃down.␈α⊂ In␈α⊃terms␈α⊂of␈α⊃benefits␈α⊂of
␈↓ α∧␈↓technology␈α
and␈α
also␈α
in␈α
terms␈α
of␈α
stressing␈αpeople's␈α
capacity␈α
to␈α
adapt,␈α
less␈α
has␈α
happened␈αsince␈α
World
␈↓ α∧␈↓War␈α
II␈αthan␈α
happened␈αsay␈α
between␈α1890␈α
and␈α
1920.␈α We␈α
have␈αhad␈α
nothing␈αthat␈α
demands␈αas␈α
much
␈↓ α∧␈↓adaptation␈α⊃as␈α∩the␈α⊃automobile␈α∩did,␈α⊃as␈α∩the␈α⊃telephone␈α⊃did,␈α∩as␈α⊃having␈α∩electricity␈α⊃did,␈α∩as␈α⊃having
␈↓ α∧␈↓domestic␈α⊂refrigeration␈α⊂did.␈α⊂ It␈α∂seems␈α⊂to␈α⊂me␈α⊂this␈α⊂has␈α∂the␈α⊂paradoxical␈α⊂effect␈α⊂of␈α⊂promoting␈α∂anti-
␈↓ α∧␈↓technological␈αattitudes␈αbecause␈αthe␈αpublic␈αcan␈αquite␈αreasonably␈αask,␈α"What␈αhave␈αthey␈αdone␈αfor␈αus
␈↓ α∧␈↓lately"?␈α∞ Technology␈α∞has␈α∞had␈α∞large␈α∞recent␈α∞effects␈α∞in␈α∞increasing␈α∞prosperity␈α∞and␈α∞increasing␈α
health,
␈↓ α∧␈↓but both prosperity and health are things that people quite reasonably take for granted.

␈↓ α∧␈↓Perhaps␈α∂I␈α⊂was␈α∂expected␈α∂to␈α⊂speak␈α∂about␈α∂artificial␈α⊂intelligence␈α∂and␈α∂the␈α⊂social␈α∂problems␈α⊂it␈α∂might
␈↓ α∧␈↓produce,␈αbut␈αI␈αwould␈αlike␈αto␈αexplain␈αwhy␈αI␈αwon't.␈α I␈αthink␈αthat␈αartificial␈αintelligence,␈αand␈αby␈αthis␈αI
␈↓ α∧␈↓mean␈α∃intelligence␈α∃on␈α∃the␈α∃human␈α∃level,␈α∃is␈α∃achievable␈α∃and␈α∃when␈α∃it␈α∃comes␈α∃it␈α∃will␈α∃be␈α∀more
␈↓ α∧␈↓revolutionary␈α⊃than␈α⊃any␈α⊂previous␈α⊃scientific␈α⊃development,␈α⊃but␈α⊂I␈α⊃also␈α⊃think␈α⊂that␈α⊃I␈α⊃do␈α⊃not␈α⊂know
␈↓ α∧␈↓enough␈αabout␈αwhat␈αit␈αwill␈αreally␈αbe␈αlike␈αto␈αsay␈αanything␈αsensible␈αabout␈αwhat␈αits␈αsocial␈αeffects␈αwill
␈↓ α∧␈↓be.␈α I␈αthought␈αabout␈αit␈αquite␈αa␈αbit,␈αand␈αI␈αthink␈αalmost␈αeveryone␈αelse␈αknows␈αeven␈αless␈αthan␈αI␈αabout
␈↓ α∧␈↓it.

␈↓ α∧␈↓␈↓ αTSome␈α∃people,␈α∃for␈α⊗example␈α∃Ed␈α∃Fredkin␈α∃of␈α⊗M.I.T␈α∃and␈α∃Donald␈α∃Michie␈α⊗of␈α∃Edinburgh
␈↓ α∧␈↓University,␈α
beat␈α
the␈α
drum␈α∞about␈α
the␈α
social␈α
importance␈α
of␈α∞AI␈α
and␈α
hold␈α
conferences␈α
on␈α∞"what␈α
if"
␈↓ α∧␈↓and␈α⊃so␈α⊃forth.␈α⊂I␈α⊃deplore␈α⊃this␈α⊂and␈α⊃ask␈α⊃you␈α⊃to␈α⊂imagine␈α⊃that␈α⊃someone␈α⊂had␈α⊃managed␈α⊃in␈α⊃the␈α⊂last
␈↓ α∧␈↓American␈α⊃presidential␈α⊃election␈α∩or␈α⊃the␈α⊃forthcoming␈α∩one␈α⊃or␈α⊃in␈α⊃some␈α∩Israeli␈α⊃election␈α⊃to␈α∩ask␈α⊃the
␈↓ α∧␈↓leading␈α∞candidates␈α
their␈α∞positions␈α
on␈α∞artificial␈α∞intelligence.␈α
 After␈α∞they␈α
finish␈α∞saying␈α∞"huh?"␈α
and
␈↓ α∧␈↓decided␈α∞that␈α∞there␈α∞was␈α∞some␈α∞possible␈α∞advantage␈α
in␈α∞having␈α∞a␈α∞position,␈α∞they␈α∞would␈α∞turn␈α∞to␈α
some
␈↓ α∧␈↓underling␈α⊃who␈α∩would␈α⊃write␈α∩whatever␈α⊃came␈α∩to␈α⊃his␈α⊃mind␈α∩and␈α⊃whatever␈α∩would␈α⊃suit␈α∩better␈α⊃the
␈↓ α∧␈↓doctrine␈αof␈αthat␈α
party␈αand␈αwe␈αwould␈α
be␈αno␈αbetter␈αoff;␈α
it␈αprobably␈αwould␈αnot␈α
do␈αany␈αharm␈α
but␈αit
␈↓ α∧␈↓might,␈α⊂because␈α∂who␈α⊂knows,␈α⊂there␈α∂might␈α⊂be␈α∂some␈α⊂law␈α⊂that␈α∂would␈α⊂be␈α∂stupefying␈α⊂in␈α⊂its␈α∂silliness.
␈↓ α∧␈↓Computer␈α⊃scientists␈α⊃would␈α⊃at␈α⊃best␈α⊃find␈α⊂themselves␈α⊃in␈α⊃the␈α⊃position␈α⊃of␈α⊃the␈α⊃biologists␈α⊂regarding
␈↓ α∧␈↓recombinant DNA - spending full time explaining why various proposals would be harmful.

␈↓ α∧␈↓What␈αI␈αdo␈αwant␈αto␈αtalk␈αabout␈αis␈αhome␈αcomputer␈αterminals␈αand␈αunreconstructed␈αtime-sharing␈αand
␈↓ α∧␈↓to␈αcompare␈αthem␈αwith␈αhome␈αcomputers.␈α I␈αdon't␈αobject␈αto␈αhome␈αcomputers,␈αthat␈αis␈αto␈αsay␈αhaving␈αa
␈↓ α∧␈↓terminal␈αwith␈α
a␈αcomputer␈α
in␈αit,␈α
but␈αI␈α
think␈αwhat␈α
is␈αhappening␈α
with␈αhome␈α
computers␈αis␈αwhat␈α
could
␈↓ α∧␈↓have␈α⊃been␈α⊃expected;␈α⊃most␈α⊃people,␈α⊃even␈α⊃most␈α⊃scientifically␈α⊃educated␈α⊃people␈α⊃have␈α⊃little␈α∩use␈α⊃for
␈↓ α∧␈↓computering␈αper␈α
se␈αexcept␈αperhaps␈α
as␈αa␈α
hobby.␈α I␈αhave␈α
had␈αa␈α
terminal␈αin␈αmy␈α
home␈αfor␈αmore␈α
than
␈↓ α∧␈↓ten␈α⊂years,␈α⊂and␈α⊂I␈α⊂use␈α⊂it␈α⊂a␈α⊂lot␈α⊂for␈α⊂writing␈α⊂books,␈α⊂papers␈α⊂and␈α⊂letters␈α⊂and␈α⊂sending␈α⊂and␈α∂receiving
␈↓ α∧␈↓messages␈αand␈αkeeping␈α
up␈αwith␈αthe␈α
news␈αand␈αoccasionally␈αfor␈α
writing␈αprograms.␈αHowever,␈α
I␈αhave
␈↓ α∧␈↓never␈α
written␈α
a␈αprogram␈α
that␈α
has␈αanything␈α
to␈α
do␈αwith␈α
my␈α
domestic␈αlife␈α
though␈α
I␈αdo␈α
keep␈α
a␈αfew
␈↓ α∧␈↓files␈αon␈αhand␈αthat␈α
are␈αrelevant␈αto␈αthat,␈α
and␈αsometimes␈αthey␈αare␈α
helpful␈αif␈αI␈αactually␈α
remember␈αto
␈↓ α∧␈↓look␈α∂at␈α∂them.␈α∂ These␈α∂applications␈α∂don't␈α∞require␈α∂that␈α∂the␈α∂terminal␈α∂have␈α∂independent␈α∞computing
␈↓ α∧␈↓␈↓ u2


␈↓ α∧␈↓capability;␈α
mine␈α∞has,␈α
and␈α
I␈α∞haven't␈α
used␈α
it.␈α∞ They␈α
do␈α
require␈α∞communication␈α
capability␈α∞and␈α
the
␈↓ α∧␈↓availability of substantial private and public files on disk.

␈↓ α∧␈↓Let␈α∞me␈α∞begin␈α∞by␈α∞describing␈α∞some␈α∞of␈α∞the␈α∂things␈α∞we␈α∞have␈α∞done␈α∞at␈α∞Stanford␈α∞and␈α∞that␈α∂have␈α∞been
␈↓ α∧␈↓done␈α⊂at␈α⊂some␈α⊂other␈α⊂places␈α⊂that␈α⊂might␈α⊂be␈α⊂relevant␈α⊂to␈α⊂the␈α⊂question␈α⊂of␈α⊂home␈α⊂use␈α⊃of␈α⊂computers.
␈↓ α∧␈↓When␈α
we␈αstarted␈α
our␈α
laboratory␈αin␈α
1966,␈α
we␈αhad␈α
already␈α
decided␈αthat␈α
documentation␈α
was␈αgoing␈α
to
␈↓ α∧␈↓be␈α
an␈αimportant␈α
use␈αof␈α
our␈αcomputer.␈α
 We␈αarranged␈α
to␈αhave␈α
a␈αline␈α
printer␈αthat␈α
would␈αprint␈α
upper
␈↓ α∧␈↓and␈αlower␈α
case␈αand␈α
a␈αlarge␈αcharacter␈α
set,␈αwhich␈α
was␈αmuch␈α
more␈αof␈αan␈α
expense␈αat␈α
that␈αtime␈αthan␈α
it
␈↓ α∧␈↓is␈α∪today.␈α∪ When␈α∪the␈α∪opportunity␈α∪came␈α∪around␈α∪1970␈α∪we␈α∪planned␈α∪and␈α∪built␈α∪a␈α∪system␈α∩which
␈↓ α∧␈↓provided␈αfor␈αa␈αterminal␈αin␈αeach␈αoffice␈αrather␈αthan␈αterminals␈αin␈αterminal␈αrooms,␈αand␈αwe␈αarranged
␈↓ α∧␈↓that␈α
the␈α
secretaries␈αshould␈α
also␈α
have␈αterminals.␈α
 I␈α
became␈α
interested␈αin␈α
home␈α
computing␈αabout␈α
that
␈↓ α∧␈↓time␈α
and␈α
foresaw␈α
a␈α
very␈α
substantial␈α
chicken␈αand␈α
egg␈α
problem,␈α
because␈α
the␈α
applications␈α
of␈αhome
␈↓ α∧␈↓computing␈αreally␈αare␈αcommunication␈αand␈αdata␈αbase␈αapplications,␈αand␈αif␈αyou␈αdo␈αnot␈αhave␈αthe␈αdata
␈↓ α∧␈↓bases then how are you going to demonstrate the applications.

␈↓ α∧␈↓␈↓ αTWe␈αfinally␈αfound␈α
one␈αwhere␈αthe␈αnecessary␈α
data␈αbases␈αare␈α
electronically␈αavailable,␈αand␈αthis␈α
is
␈↓ α∧␈↓a␈α⊂news␈α∂service.␈α⊂ the␈α∂news␈α⊂service.␈α∂ We␈α⊂have␈α⊂coming␈α∂into␈α⊂our␈α∂computer␈α⊂the␈α∂news␈α⊂wires␈α⊂of␈α∂the
␈↓ α∧␈↓Associated␈αPress␈αand␈αthe␈α␈↓↓New␈αYork␈αTimes␈↓.␈α Our␈α
computer␈αis␈αconnected␈αto␈αthe␈αwire␈αjust␈αas␈α
though
␈↓ α∧␈↓it␈αwere␈αa␈αteletype␈αin␈αa␈αnewspaper␈αoffice.␈α As␈αeach␈αstory␈αcomes␈αinto␈αthe␈αcomputer,␈αit␈αis␈αindexed␈αby
␈↓ α∧␈↓all␈αnon-trivial␈αwords,␈αi.e.␈αall␈αbut␈αwords␈αlike␈α"the"␈αand␈α"he".␈α The␈αresult␈αis␈αa␈αcontinuously␈αupdated
␈↓ α∧␈↓concordance␈α
of␈αthe␈α
day's␈αnews␈α
stories.␈α Each␈α
midnight␈αthe␈α
file␈αof␈α
the␈αday's␈α
stories␈αis␈α
closed,␈αand␈α
we
␈↓ α∧␈↓keep␈αtwo␈αweeks;␈αif␈αwe␈αhad␈αmore␈αdisk␈αfile␈αwe␈αwould␈αkeep␈αmore.␈α Aftter␈αthe␈αprogram␈αuser␈αtypes␈α
NS
␈↓ α∧␈↓for␈α"news␈αservice",␈αhe␈αcan␈αgive␈αthe␈αprogram␈αa␈α"key␈αword␈αexpression".␈α Thus␈αif␈αhe␈αwere␈α
intereseted
␈↓ α∧␈↓in␈α∂Israel␈α∞but␈α∂felt␈α∂he␈α∞had␈α∂heard␈α∂enough␈α∞about␈α∂Dayan's␈α∞resignation␈α∂then␈α∂he␈α∞might␈α∂type␈α∂the␈α∞key
␈↓ α∧␈↓word␈α∞expression␈α
"israel-dayan"␈α∞which␈α
would␈α∞give␈α
all␈α∞stories␈α
that␈α∞had␈α
the␈α∞word␈α
"Israel"␈α∞in␈α
them
␈↓ α∧␈↓but␈α∪did␈α∪not␈α∪have␈α∪the␈α∪word␈α∪"Dayan"␈α∩in␈α∪them.␈α∪ You␈α∪can␈α∪make␈α∪arbitrarily␈α∪complex␈α∩Boolean
␈↓ α∧␈↓expressions.  Actually the data base is small enough so that false hits do not bother you.

␈↓ α∧␈↓␈↓ αTThis␈αsystem␈αis␈αquite␈αcrude,␈αbut␈αin␈αsome␈αrespects␈αit␈αcompares␈αfavorably␈αwith␈αthe␈αmuch␈αmore
␈↓ α∧␈↓sophisticated␈α⊃␈↓↓New␈α⊃York␈α⊂Times␈↓␈α⊃Data␈α⊃Bank.␈α⊂ The␈α⊃latter␈α⊃employs␈α⊂human␈α⊃abstractors␈α⊃who␈α⊂make
␈↓ α∧␈↓abstracts␈α∂of␈α∂everything␈α∞that␈α∂appears␈α∂in␈α∂the␈α∞␈↓↓Times␈↓␈α∂and␈α∂many␈α∞other␈α∂publications␈α∂and␈α∂files␈α∞them
␈↓ α∧␈↓under␈α∩key␈α∩words␈α∩that␈α∩seem␈α∩important␈α∩to␈α∩the␈α∩abstractors␈α∩at␈α∩the␈α∩time.␈α∩ If␈α∩something␈α⊃becomes
␈↓ α∧␈↓important␈αin␈α
retrospect,␈αfor␈α
example␈αthe␈α
building␈αWatergate,␈α
then␈αyou␈α
cannot␈αgo␈α
back␈αin␈αtime␈α
and
␈↓ α∧␈↓say␈α∞what␈α∞stories␈α∞mention␈α∞Watergate␈α∞before␈α∞the␈α∞time␈α∞it␈α∞became␈α∞important␈α∞because␈α∞the␈α
abstractor
␈↓ α∧␈↓didn't then consider it important.

␈↓ α∧␈↓␈↓ αTOne␈αcan␈αask␈αNS␈αto␈αbe␈αnotified␈αwhen␈αstories␈αthat␈αfit␈αspecified␈αkey␈αword␈αexpression␈αcome␈αin.
␈↓ α∧␈↓You␈αget␈αan␈αannouncement␈αif␈αyour␈αare␈αlogged␈αin␈αand␈αelectronic␈αmail␈αif␈αyou␈αare␈αnot.␈α You␈αcan␈αalso
␈↓ α∧␈↓move␈α
stories␈α
to␈α
files␈αor␈α
print␈α
them.␈α
 This␈αwas␈α
one␈α
of␈α
the␈αfew␈α
experiments␈α
aimed␈α
at␈αhome␈α
computer
␈↓ α∧␈↓use that could done in 1972.

␈↓ α∧␈↓␈↓ αTOur␈α
Dialnet␈αproject␈α
is␈αalso␈α
relevant.␈α It␈α
is␈αa␈α
project␈αfor␈α
communication␈α
between␈αcomputers
␈↓ α∧␈↓giving␈α∩services␈α∩like␈α∩those␈α∪of␈α∩the␈α∩ARPANET␈α∩to␈α∩any␈α∪computer␈α∩that␈α∩will␈α∩equip␈α∩itself␈α∪with␈α∩a
␈↓ α∧␈↓computer␈α∀operated␈α∀telephone␈α∃dialer␈α∀and␈α∀a␈α∀1200␈α∃baud␈α∀Bell␈α∀System␈α∀compatible␈α∃modem␈α∀and
␈↓ α∧␈↓implement␈αthe␈αDialnet␈αprotocols.␈α Given␈αthis,␈αa␈αuser␈αcan␈αtransfer␈αa␈αfile␈αbetween␈αhis␈αcomputer␈αand
␈↓ α∧␈↓any␈α
other␈α
computer.␈α
 He␈αneed␈α
only␈α
give␈α
the␈αlocal␈α
file␈α
name␈α
and␈α
the␈αforeign␈α
file␈α
name␈α
and␈αthen␈α
get
␈↓ α∧␈↓through the other computer's password fence.

␈↓ α∧␈↓␈↓ αTIt␈α
does␈α
require␈α
something␈α
that␈α
one␈α
might␈α
take␈α
as␈α
obvious␈α
-␈α
that␈α
every␈α
computer␈αsystem␈α
have
␈↓ α∧␈↓␈↓ u3


␈↓ α∧␈↓a␈α
systemwide␈α
naming␈α
of␈α
files.␈α
 However␈α
curiously␈α∞enough,␈α
and␈α
this␈α
will␈α
be␈α
my␈α
first␈α
nibble␈α∞on␈α
a
␈↓ α∧␈↓hand␈α∂that␈α∂have␈α∂been␈α∂very␈α∂generous.␈α∂ Few␈α∂if␈α∂any␈α∂IBM␈α∂operating␈α∂systems␈α∂have␈α⊂this␈α∂elementary
␈↓ α∧␈↓characteristic␈α∞of␈α
having␈α∞systemwide␈α∞file␈α
names.␈α∞ What␈α∞we␈α
actually␈α∞need␈α∞and␈α
can␈α∞get␈α∞by␈α
suitable
␈↓ α∧␈↓prefexing␈αis␈α
world␈αwide␈α
file␈αnames␈α
so␈αthat␈α
you␈αcan␈αrefer␈α
by␈αname␈α
to␈αany␈α
file␈αin␈α
the␈αworld␈α
-␈αnot
␈↓ α∧␈↓necessarily that everybody would let you have any file or even admit that it exists.

␈↓ α∧␈↓␈↓ αTDialnet␈αis␈αalso␈αused␈αto␈αsend␈αcomputer␈αmail.␈α Thus␈αa␈αcomputer␈αmay␈αreceive␈αa␈αtelephone␈αcall
␈↓ α∧␈↓from␈α∞another␈α∂computer␈α∞in␈α∞which␈α∂it␈α∞says␈α∂"I␈α∞have␈α∞mail␈α∂for␈α∞your␈α∞user␈α∂named␈α∞so-and-so",␈α∂and␈α∞of
␈↓ α∧␈↓course␈α
the␈α
receiving␈αcomputer␈α
can␈α
reply␈αor␈α
just␈α
hang␈αup␈α
or␈α
it␈α
can␈αreply␈α
that␈α
it␈αhas␈α
no␈α
such␈αuser␈α
or
␈↓ α∧␈↓it␈αcan␈αaccept␈αthe␈αmail.␈α Perhaps␈αthis␈αis␈α
the␈αway␈αcomputer␈αmail␈αshould␈αhave␈αbeen␈αdeveloped␈αin␈α
the
␈↓ α∧␈↓first␈α∞place␈α∞instead␈α∞of␈α∞basing␈α∞an␈α∂ARPAnet␈α∞on␈α∞leased␈α∞dedicated␈α∞facilities.␈α∞ The␈α∂standard␈α∞dial-up
␈↓ α∧␈↓telephone␈α⊂network␈α⊂much␈α⊂more␈α⊂capability␈α⊂than␈α⊂has␈α⊂been␈α⊂exercised.␈α⊂ For␈α⊂example,␈α⊂it␈α⊂is␈α⊂already
␈↓ α∧␈↓international.

␈↓ α∧␈↓␈↓ αTAnother␈α∞sort␈α∞of␈α
protoproduct␈α∞project␈α∞is␈α
the␈α∞Common␈α∞Business␈α∞Communication␈α
Language.
␈↓ α∧␈↓When␈α∩I␈α⊃first␈α∩thought␈α⊃of␈α∩it,␈α⊃it␈α∩didn't␈α⊃seem␈α∩to␈α⊃have␈α∩much␈α⊃scientific␈α∩interest␈α⊃and␈α∩certainly␈α⊃no
␈↓ α∧␈↓connection␈α∩with␈α∪artificial␈α∩intelligence␈α∪research.␈α∩ It␈α∪seemed␈α∩to␈α∪be␈α∩just␈α∪another␈α∩idea␈α∪for␈α∩using
␈↓ α∧␈↓computers␈α∂to␈α∂increase␈α∂human␈α⊂productivity.␈α∂ It␈α∂was␈α∂set␈α∂off␈α⊂by␈α∂reading␈α∂a␈α∂1965␈α∂article␈α⊂about␈α∂the
␈↓ α∧␈↓world␈α∞of␈α∞the␈α∞future.␈α∂ It␈α∞was␈α∞a␈α∞1965␈α∞model␈α∂world␈α∞of␈α∞the␈α∞future,␈α∞and␈α∂it␈α∞referred␈α∞to␈α∞a␈α∂time␈α∞when
␈↓ α∧␈↓businesses would be quite computerised.

␈↓ α∧␈↓␈↓ αTThere␈αwas␈αa␈αscenario␈αin␈αwhich␈αa␈αclerk␈αin␈αthe␈αpurchasing␈αdepartment␈αof␈αa␈αcompany␈αhears␈αa
␈↓ α∧␈↓beep␈αand␈αturns␈αto␈αa␈αdisplay␈αscreen␈αthat␈αsays,␈α"We␈αneed␈α5000␈αpencils.␈α Order␈αthem␈αfrom␈αcompany
␈↓ α∧␈↓B".␈α∞ She␈α∞(Sorry␈α∂Pat,␈α∞this␈α∞is␈α∞a␈α∂1965␈α∞model␈α∞world␈α∂of␈α∞the␈α∞future.)␈α∞turns␈α∂from␈α∞her␈α∞terminal␈α∂to␈α∞her
␈↓ α∧␈↓typewriter␈αand␈αtypes␈α
out␈αa␈αpurchase␈α
order␈αwhich␈αis␈α
duly␈αsent␈αto␈α
company␈αB␈αwhere␈α
another␈αclerk
␈↓ α∧␈↓turns from her incoming mail basket to a terminals and types "Send IBM 5000 pencils".

␈↓ α∧␈↓␈↓ αTWe␈α
would␈α
like␈α
to␈α
let␈α
both␈α
clerks␈α∞do␈α
something␈α
more␈α
worthy␈α
of␈α
their␈α
talents␈α
and␈α∞have␈α
the
␈↓ α∧␈↓computers␈αcommunicate␈αwith␈αeach␈αother␈αdirectly.␈α In␈α
order␈αto␈αdo␈αthis␈αwe␈αneed␈αa␈αstandard␈α
business
␈↓ α∧␈↓communication␈α
language␈α∞wherein␈α
one␈α∞computer␈α
can␈α∞call␈α
up␈α
another␈α∞and␈α
can␈α∞say␈α
"What␈α∞is␈α
your
␈↓ α∧␈↓price␈αand␈αdelivery␈α
for␈α5000␈αpencils?"␈α
and␈α"I␈αhereby␈α
order␈α5000␈αpencils"␈α
and␈αthings␈αlike␈αthat.␈α
 This
␈↓ α∧␈↓may␈α
seem␈α
like␈αa␈α
grubby␈α
bit␈α
of␈αcommercial␈α
standardization␈α
of␈α
no␈αinterest␈α
to␈α
computer␈αscientists,␈α
but
␈↓ α∧␈↓in fact it raises all sorts of interesting questions.

␈↓ α∧␈↓␈↓ αTOne␈α∂question␈α∞is␈α∂how␈α∞such␈α∂a␈α∞system␈α∂can␈α∞ever␈α∂change.␈α∞ I␈α∂began␈α∞to␈α∂think␈α∞about␈α∂this␈α∞while
␈↓ α∧␈↓visiting␈α⊃a␈α⊂San␈α⊃Diego␈α⊂naval␈α⊃training␈α⊃facility.␈α⊂ There␈α⊃is␈α⊂something␈α⊃called␈α⊂the␈α⊃fleet␈α⊃data␈α⊂system
␈↓ α∧␈↓whereby␈α
every␈α∞ship␈α
in␈α
a␈α∞fleet␈α
would␈α∞know␈α
about␈α
all␈α∞of␈α
the␈α∞airplanes␈α
and␈α
ships␈α∞that␈α
any␈α∞of␈α
the
␈↓ α∧␈↓ships␈α
in␈αthe␈α
fleet␈αhad␈α
detected␈α
by␈αexchanging␈α
messages.␈αAt␈α
the␈α
time␈αNATO␈α
was␈αgoing␈α
to␈αjoin␈α
this
␈↓ α∧␈↓system␈αand␈α
suitable␈αcommittees␈α
had␈αdetermined␈αthat␈α
the␈αsystem␈α
should␈αbe␈α
updated␈αor␈αimproved␈α
at
␈↓ α∧␈↓that␈αtime␈αand␈α
there␈αwas␈αto␈α
be␈αa␈αcutover␈α
day.␈α There␈αwas␈α
the␈αhope␈αthat␈α
the␈αRussians␈αwould␈αnot␈α
do
␈↓ α∧␈↓any␈αthing␈αon␈α
this␈αcutover␈αday␈α
because␈αthings␈αwere␈α
surely␈αgoing␈αto␈α
be␈αin␈αa␈α
great␈αstate␈αof␈α
confusion.
␈↓ α∧␈↓Presumably␈αthis␈αhas␈αhappened␈αby␈αnow,␈αsince␈αthe␈αvisit␈αwas␈α7␈αor␈α8␈αyears␈αago.␈α Massive␈αchanges␈αon
␈↓ α∧␈↓cutover␈α∞dates␈α∂are␈α∞quite␈α∞unsatisfactory␈α∂for␈α∞large␈α∂communication␈α∞and␈α∞data␈α∂systems␈α∞and␈α∂would␈α∞be
␈↓ α∧␈↓almost␈α∀impossible␈α∀for␈α∀loosely␈α∀coupled␈α∀systems␈α∪involving␈α∀tens␈α∀of␈α∀thousands␈α∀of␈α∀computers␈α∪of
␈↓ α∧␈↓differing ages, sizes and manufacturers under no central management.

␈↓ α∧␈↓␈↓ αTGrowth␈α⊂requires␈α∂that␈α⊂when␈α∂a␈α⊂new␈α∂kind␈α⊂of␈α∂message␈α⊂is␈α∂introduced␈α⊂message␈α⊂sources␈α∂begin
␈↓ α∧␈↓transmitting␈α⊂both␈α⊃kinds␈α⊂of␈α⊂message.␈α⊃ When␈α⊂enough␈α⊂new␈α⊃style␈α⊂messages␈α⊂are␈α⊃being␈α⊂transmitted,
␈↓ α∧␈↓receivers␈α⊂begin␈α∂using␈α⊂them.␈α∂ When␈α⊂no-one␈α∂requires␈α⊂old␈α∂style␈α⊂messages,␈α∂their␈α⊂transmission␈α∂may
␈↓ α∧␈↓stop.
␈↓ α∧␈↓␈↓ u4


␈↓ α∧␈↓␈↓ αTAnother␈α
design␈α
idea␈α
is␈α
that␈αeach␈α
message␈α
should␈α
be␈α
a␈α
list,␈αand␈α
every␈α
item␈α
should␈α
be␈αa␈α
either
␈↓ α∧␈↓an␈α
atom␈α
or␈α
a␈α
list␈α
whose␈α
first␈α
item␈α
designates␈αwhat␈α
kind␈α
of␈α
a␈α
thing␈α
the␈α
list␈α
represents.␈α
 You␈αmay
␈↓ α∧␈↓have␈α
heard␈α
of␈α
such␈α
lists;␈α
they␈α
begin␈α
and␈α
end␈α
with␈α
parentheses.␈α
 Remember␈α
that␈α
only␈αcomputers␈α
get
␈↓ α∧␈↓to see them.

␈↓ α∧␈↓␈↓ αTAnother␈α∞characteristic␈α∞was␈α∂suggested␈α∞in␈α∞N.␈α∞ Chomsky's␈α∂book␈α∞␈↓↓Reflections␈α∞on␈α∂Language␈↓;␈α∞he
␈↓ α∧␈↓claims␈α∞that␈α∞human␈α∞languages␈α∞have␈α
it,␈α∞other␈α∞linguists␈α∞dispute␈α∞it,␈α
but␈α∞anyway␈α∞we␈α∞want␈α∞it␈α∞for␈α
our
␈↓ α∧␈↓Common␈αBusiness␈α
Communication␈αLanguage.␈α It␈α
is␈αthat␈αthe␈α
syntax␈αnever␈αrequires␈α
an␈αatom␈αat␈α
any
␈↓ α∧␈↓place.␈α∞ For␈α
example,␈α∞if␈α
a␈α∞price␈α
is␈α∞to␈α
be␈α∞designated␈α
then␈α∞of␈α
course␈α∞␈↓∧1␈↓3.12␈α
is␈α∞a␈α
reasonable␈α∞thing␈α
to
␈↓ α∧␈↓appear␈α⊂as␈α⊂a␈α∂list␈α⊂item,␈α⊂but␈α∂any␈α⊂expression␈α⊂designating␈α∂a␈α⊂price␈α⊂should␈α∂also␈α⊂be␈α⊂allowed,␈α⊂such␈α∂as
␈↓ α∧␈↓translations␈α⊂of␈α⊂␈↓↓"the␈α∂same␈α⊂price␈α⊂as␈α∂last␈α⊂week"␈↓␈α⊂or␈α∂␈↓↓"in␈α⊂accordance␈α⊂with␈α∂our␈α⊂contract"␈↓␈α⊂or␈α⊂␈↓↓"the␈α∂price
␈↓ α∧␈↓↓granted to the most favored customer"␈↓ or whatever.

␈↓ α∧␈↓␈↓ αTWhen␈α∞I␈α∞tried␈α
to␈α∞specify␈α∞a␈α
simple␈α∞version␈α∞of␈α
a␈α∞CBCL␈α∞that␈α
would␈α∞be␈α∞expandable,␈α∞I␈α
found
␈↓ α∧␈↓substantial␈αdifficulties␈αwhich␈αled␈αto␈αsome␈αinteresting␈αscientific␈αproblems.␈α Consider␈αdesigning␈αhow
␈↓ α∧␈↓delivery␈α∪is␈α∀specified.␈α∪ First,␈α∪the␈α∀conditions␈α∪might␈α∪be␈α∀unspecified,␈α∪meaning␈α∪that␈α∀the␈α∪delivery
␈↓ α∧␈↓conditions␈αare␈αstandard,␈αor␈αit␈αmight␈αsay␈α␈↓↓"the␈αbest␈αway"␈↓.␈α However,␈αif␈αyou␈αwant␈αto␈αelaborate␈αwhat␈α
is
␈↓ α∧␈↓really␈αto␈αbe␈αallowed␈α
in␈αdelivery␈αspecifications,␈αthere␈αare␈α
all␈αkinds␈αof␈αconditional␈αspecifications␈α
such
␈↓ α∧␈↓as␈α␈↓↓"to␈αbe␈αsent␈αby␈αFederal␈αExpress␈αunless␈αthey␈αare␈αon␈αstrike,␈αin␈αwhich␈αcase␈αit␈αgoes␈αby␈αpost␈αoffice"␈↓␈αor
␈↓ α∧␈↓vice␈α∩versa.␈α∩ It␈α∩turns␈α∪out␈α∩that␈α∩a␈α∩substantial␈α∩and␈α∪interesting␈α∩part␈α∩of␈α∩the␈α∩␈↓↓semantics␈↓␈α∪of␈α∩natural
␈↓ α∧␈↓language is involved in formalizing what one business computer might say to another.

␈↓ α∧␈↓␈↓ αTThis␈α∀is␈α∀interesting␈α∀from␈α∀a␈α∀scientific␈α∪point␈α∀of␈α∀view,␈α∀because␈α∀interest␈α∀in␈α∪computational
␈↓ α∧␈↓linguistics␈α
has␈α
mainly␈α
focused␈α
on␈α
the␈α
syntax␈α
of␈αlanguage.␈α
 Here␈α
is␈α
a␈α
case␈α
where␈α
we␈α
are␈α
going␈αto
␈↓ α∧␈↓use␈α
a␈α
perfectly␈α
rigid␈α
syntax␈α
decided␈α
in␈α
advance,␈α
but␈α
we␈α
have␈α
all␈α
the␈α
semantic␈α
problems␈αof␈α
deciding
␈↓ α∧␈↓exactly␈α
what␈α
so␈αyou␈α
want␈α
one␈αcomputer␈α
to␈α
be␈αable␈α
to␈α
say␈α
to␈αanother.␈α
 This␈α
has␈αled␈α
me␈α
to␈αthe␈α
belief
␈↓ α∧␈↓that␈αputting␈α
English␈αlanguage␈α
"front␈αends"␈α
on␈αprograms␈α
that␈αbasically␈α
speak␈αin␈α"computerese"␈α
may
␈↓ α∧␈↓miss␈α
the␈α
point␈α
scientifically,␈α
because␈α
more␈α
interesting␈α
than␈α
how␈α
to␈α
say␈α
in␈α
natural␈α
language␈αwhat␈α
we
␈↓ α∧␈↓already␈α∞say␈α∞in␈α
various␈α∞computer␈α∞input-output␈α∞jargons,␈α
is␈α∞how␈α∞to␈α
make␈α∞the␈α∞computer␈α∞say␈α
things
␈↓ α∧␈↓said in natural language that we do not presently know how to make computers say at all.

␈↓ α∧␈↓This␈α∩was␈α∩by␈α∩way␈α∩of␈α∩an␈α∩advertisement␈α∩for␈α∩the␈α∩scientific␈α∩interest␈α∩of␈α∩some␈α∩of␈α∩these␈α⊃problems.
␈↓ α∧␈↓Another␈αimportant␈αfeature␈αof␈α
our␈αenvironment␈αis␈αthe␈α
mail␈αsystem␈αthat␈αpermits␈α
sending␈αelectronic
␈↓ α∧␈↓mail␈αall␈α
over␈αthe␈α
ARPANET␈αand␈αto␈α
other␈αplaces.␈α
 I␈αcertainly␈αhave␈α
gotten␈αinto␈α
the␈αstate␈α
where␈αI
␈↓ α∧␈↓divide␈α∞the␈α∞world␈α∂into␈α∞two␈α∞classes␈α∂of␈α∞people␈α∞-␈α∞those␈α∂to␈α∞whom␈α∞I␈α∂can␈α∞send␈α∞ARPANET␈α∂mail␈α∞and
␈↓ α∧␈↓those␈αwith␈αwhom␈αI␈α
have␈αto␈αrely␈αupon␈α
more␈αordinary␈αkinds␈αof␈α
communication.␈α I␈αthink␈αI␈α
will␈αnot
␈↓ α∧␈↓sing␈αmy␈αsong␈αabout␈αwhy␈αit␈αis␈αoften␈αbetter␈α
than␈αtelephoning␈αand␈αthings␈αlike␈αthat.␈αI␈αwill␈αsay␈α
that␈αI
␈↓ α∧␈↓suspect␈α
that␈α∞IBM␈α
may␈α
have␈α∞gone␈α
into␈α
some␈α∞kind␈α
of␈α
blind␈α∞alley␈α
if␈α
they␈α∞suppose␈α
that␈α∞the␈α
speech
␈↓ α∧␈↓filing␈α⊃system␈α⊃will␈α⊂obviate␈α⊃the␈α⊃need␈α⊃for␈α⊂written␈α⊃electronic␈α⊃mail␈α⊂facilities␈α⊃for␈α⊃everybody␈α⊃-␈α⊂even
␈↓ α∧␈↓executives␈α
-␈α
who␈α
can␈α
type␈α
as␈α
well␈α
as␈α
engineers␈α
when␈α
they␈α
have␈α
the␈α
motivation␈α
-␈α
and␈αunder␈α
certain
␈↓ α∧␈↓circumstances, I believe they will.

␈↓ α∧␈↓I␈α∞would␈α
like␈α∞to␈α
do␈α∞as␈α
the␈α∞other␈α
speakers␈α∞have␈α
done␈α∞and␈α
talk␈α∞about␈α
social␈α∞effects,␈α
but␈α∞I␈α∞want␈α
to
␈↓ α∧␈↓express␈αthe␈α
opinion␈αthat␈αmuch␈α
of␈αthe␈α
talk␈αabout␈αthe␈α
effects␈αof␈αtechnology␈α
has␈αbeen␈α
harmful␈αand
␈↓ α∧␈↓even␈α∪immoral.␈α∪ I␈α∪believe␈α∀that␈α∪people␈α∪should␈α∪be␈α∀cautious␈α∪about␈α∪talking␈α∪about␈α∀social␈α∪effects,
␈↓ α∧␈↓especially␈α⊂about␈α⊃talking␈α⊂about␈α⊃planning␈α⊂desirable␈α⊃social␈α⊂effects␈α⊃and␈α⊂planning␈α⊃that␈α⊂undesirable
␈↓ α∧␈↓social␈α⊃effects␈α⊂should␈α⊃be␈α⊂avoided.␈α⊃ The␈α⊂danger␈α⊃is␈α⊂that␈α⊃this␈α⊂will␈α⊃come␈α⊂down␈α⊃to␈α⊃planning␈α⊂other
␈↓ α∧␈↓people's␈αlives␈αfor␈αthem,␈αand␈αit␈αseems␈αto␈αme␈αthat␈αthis␈αdisease␈αafflicts␈αmany␈αof␈αthose␈αAmericans␈αwho
␈↓ α∧␈↓refer to themselves as the "␈↓↓public policy community"␈↓.
␈↓ α∧␈↓␈↓ u5


␈↓ α∧␈↓␈↓ αTIn␈α
particular,␈α∞I␈α
doubt␈α∞the␈α
virtue␈α∞of␈α
the␈α∞Office␈α
of␈α∞Technology␈α
Assessment␈α∞which␈α
Congress
␈↓ α∧␈↓created␈α∞to␈α
advise␈α∞about␈α∞the␈α
consequences␈α∞of␈α
new␈α∞technology.␈α∞ Its␈α
attitude␈α∞seems␈α
to␈α∞be␈α∞␈↓↓"Don't␈α
do
␈↓ α∧␈↓↓anything␈α
new␈αwithout␈α
consulting␈αus,␈α
and␈α
our␈αnew␈α
director␈αwill␈α
take␈α
office␈αin␈α
a␈αfew␈α
months,␈αand␈α
we
␈↓ α∧␈↓↓will be able to consider these things when we have decided in what order to start to consider them"␈↓.

␈↓ α∧␈↓␈↓ αTLet␈αme␈αput␈αit␈αthis␈αway:␈αhardly␈αany␈αtechnology␈αhas␈αspecific␈αsocial␈αeffects,␈αgood␈αor␈αbad;␈αwhat
␈↓ α∧␈↓technology␈α∞does␈α
is␈α∞provide␈α
opportunities.␈α∞ Some␈α
people␈α∞can␈α
use␈α∞these␈α
opportunities␈α∞to␈α∞do␈α
things
␈↓ α∧␈↓that␈α∞will␈α∂turn␈α∞out␈α∞to␈α∂be␈α∞good␈α∞and␈α∂other␈α∞people␈α∞can␈α∂use␈α∞them␈α∞to␈α∂do␈α∞things␈α∞that␈α∂turn␈α∞out␈α∂to␈α∞be
␈↓ α∧␈↓harmful.␈α∞ Which␈α∞you␈α∞emphasize␈α∞depends␈α∞less␈α∞on␈α
specific␈α∞technologies␈α∞than␈α∞on␈α∞how␈α∞you␈α∞look␈α
at
␈↓ α∧␈↓people and institutions and the ways they can be expected to use any opportunity.

␈↓ α∧␈↓You␈αsay␈αthat␈α
if␈αwe␈αgive␈αthem␈α
the␈αchance␈αthey␈α
are␈αsure␈αto␈αabuse␈α
it,␈αand␈αyou␈α
can␈αhave␈αtwo␈αkinds␈α
of
␈↓ α∧␈↓worry: one is what they will do to each other and the other is what they will do to themselves.

␈↓ α∧␈↓␈↓ αTSome␈α∞people␈α∞worry␈α∞that␈α∞if␈α∞people␈α∞have␈α∞too␈α∞much␈α∞access␈α∞to␈α∞computer␈α∞files␈α∞they␈α∞will␈α
waste
␈↓ α∧␈↓their␈α
time␈α
or␈α
if␈α∞they␈α
have␈α
time-sharing␈α
they␈α
will␈α∞try␈α
out␈α
programs␈α
without␈α∞adequately␈α
thinking.
␈↓ α∧␈↓They␈α∪are␈α∪really␈α∪all␈α∪out␈α∪to␈α∪help␈α∪each␈α∪other␈α∪by␈α∪disciplining␈α∪each␈α∪other.␈α∪ Dijkstra's␈α∀book␈α∪␈↓↓"A
␈↓ α∧␈↓↓Discipline␈α
of␈α
Programming"␈↓␈α
suggests␈α
to␈αme␈α
a␈α
book␈α
jacket␈α
picture␈αof␈α
someone␈α
standing␈α
on␈α
a␈αdais
␈↓ α∧␈↓with␈α
a␈α
whip␈α
menacing␈α
a␈α
crowd␈α
of␈αhalf-naked␈α
programmers␈α
at␈α
computer␈α
terminals.␈α
 So␈α
if␈αyou␈α
want
␈↓ α∧␈↓discpline, put an ad in the ␈↓↓Berkeley Barb␈↓.

␈↓ α∧␈↓␈↓ αTMy␈αidea␈αof␈αsocial␈αconsciousness␈αis␈αdifferent,␈αand␈αinstead␈αof␈αinventing␈αways␈αin␈αwhich␈αthings
␈↓ α∧␈↓might␈αgo␈αwrong,␈αI␈αwould␈αlike␈αto␈αdiscuss␈αsome␈αuseful␈αopportunities.␈α After␈αall,␈αthe␈αprobability␈αthat
␈↓ α∧␈↓an␈αopportunity␈αwill␈αbe␈αmissed␈αis␈αfar␈αgreater␈αthan␈αthe␈αprobability␈αthat␈αan␈αunfortunate␈αaspect␈αof␈αan
␈↓ α∧␈↓application will go unnoticed.

␈↓ α∧␈↓␈↓ αTFor␈αsome␈αyears␈αnow,␈αdisk␈αstorage␈αhas␈αbeen␈αcheap␈αenough␈αso␈αthat␈αit␈αhas␈αbeen␈αeconomical␈αto
␈↓ α∧␈↓have␈αa␈αnational␈αlibrary␈αin␈αwhich␈αeverything␈αthat␈αhas␈αbeen␈αwritten␈αis␈αstored␈αon␈αdisk␈αand␈αavailable
␈↓ α∧␈↓from␈αa␈αterminal␈αanywhere␈αin␈αa␈αcountry␈αor␈α
even␈αin␈αthe␈αworld.␈α Such␈αa␈αfile␈αwould␈αbe␈α
the␈αeffective
␈↓ α∧␈↓equivalent␈αof␈αa␈αvery␈αlarge␈αnumber␈αof␈αcopies␈αof␈αthe␈αbook.␈α Recently␈αit␈αhas␈αbecome␈αcheaper␈αto␈αstore
␈↓ α∧␈↓even␈α
a␈α
single␈αcopy␈α
on␈α
disk␈α
than␈αon␈α
a␈α
library␈α
shelf␈α-␈α
counting␈α
the␈α
cost␈αof␈α
the␈α
library,␈α
staffing␈αit,
␈↓ α∧␈↓cataloging␈α
the␈αbook␈α
and␈α
the␈αcost␈α
of␈α
the␈αbook␈α
itself.␈α Rumor␈α
has␈α
it␈αthat␈α
IBM␈α
can␈αmake␈α
disk␈αfiles␈α
of
␈↓ α∧␈↓more␈α
than␈α
six␈α
times␈α
the␈α
density␈α
of␈α
the␈α
current␈α
best␈α
that␈α
would␈α
cost␈α
little␈α
more␈α
to␈α∞produce.␈α
 Our
␈↓ α∧␈↓society␈αhas␈αsuffered␈α
from␈αmissing␈αthe␈α
1970␈αopportunity␈αto␈α
create␈αsuch␈αa␈α
library,␈αand␈αthe␈αblame␈α
for
␈↓ α∧␈↓this␈α
belongs␈α
mainly␈α
to␈α
the␈α
technology␈α
assessors,␈α
whose␈α
imaginations␈α
extend␈α
only␈α
to␈α
what␈α
may␈αgo
␈↓ α∧␈↓wrong with new technology.

␈↓ α∧␈↓␈↓ αTThe␈α∩simple␈α⊃convenience␈α∩of␈α⊃having␈α∩everything␈α⊃in␈α∩the␈α⊃Library␈α∩of␈α∩Congress␈α⊃immediately
␈↓ α∧␈↓available␈α⊃is␈α⊃important␈α∩in␈α⊃itself;␈α⊃much␈α∩misinformation␈α⊃would␈α⊃be␈α⊃avoided␈α∩if␈α⊃such␈α⊃facts␈α∩as␈α⊃are
␈↓ α∧␈↓available␈α∃could␈α∃be␈α⊗easily␈α∃looked␈α∃up.␈α⊗ However,␈α∃its␈α∃effect␈α∃on␈α⊗the␈α∃future␈α∃of␈α⊗publication␈α∃is
␈↓ α∧␈↓sociologically␈α∞even␈α∞more␈α∞interesting.␈α∞ If␈α∞all␈α∂publication␈α∞is␈α∞available␈α∞via␈α∞terminals␈α∞from␈α∂disk␈α∞file,
␈↓ α∧␈↓then␈α
publication␈αconsists␈α
merely␈αof␈α
declaring␈αpublic␈α
a␈αfile␈α
that␈αyou␈α
have␈αput␈α
into␈αthe␈α
system.␈α In
␈↓ α∧␈↓order␈αto␈αget␈αanybody␈αto␈αread␈α
it␈αyou␈αhave␈αto␈αget␈αit␈αto␈α
his␈αattention,␈αand␈αsince␈αwe␈αcan't␈αsuppose␈α
that
␈↓ α∧␈↓people␈α
will␈α
read␈α
more␈α
than␈α
now,␈α
this␈αwill␈α
continue␈α
to␈α
be␈α
complicated␈α
and␈α
uncertain.␈α
 As␈αnow,␈α
some
␈↓ α∧␈↓people␈α∞will␈α∞read␈α∞files␈α∞constituting␈α∞book␈α∞reviews,␈α∞and␈α∞the␈α∞reviewers␈α∞will␈α∞review␈α∞what␈α∞books␈α∞they
␈↓ α∧␈↓would␈α
consider␈α
worthwhile,␈α
and␈α
so␈α
forth.␈α
 People␈α
who␈α
write␈α
something␈α
eccentric␈α
or␈α
not␈α
of␈αmuch
␈↓ α∧␈↓interest␈αto␈αother␈αpeople␈αwill␈αhave␈αas␈αmuch␈αtrouble␈αgetting␈αtheir␈αstuff␈αread␈αunder␈αa␈αnew␈αsystem␈αas
␈↓ α∧␈↓they do now.
␈↓ α∧␈↓␈↓ u6


␈↓ α∧␈↓␈↓ αTThe␈α∞important␈α
fact␈α∞is␈α
that␈α∞such␈α
a␈α∞system␈α
will␈α∞make␈α
the␈α∞writers,␈α
especially␈α∞the␈α∞well␈α
known
␈↓ α∧␈↓ones,␈αsubstantially␈αindependent␈αof␈αpublishers.␈αThe␈αcost␈αof␈αgetting␈αa␈αbook␈αout␈αfor␈αa␈αreader␈α
is␈αnow
␈↓ α∧␈↓something␈α∞like␈α∞between␈α
10%␈α∞and␈α∞20%␈α∞to␈α
the␈α∞writer␈α∞and␈α∞80%␈α
printing␈α∞and␈α∞distribution␈α∞costs,␈α
and
␈↓ α∧␈↓these␈αnumbers␈αcan␈αeasily␈α
be␈αreversed.␈α I␈αthink␈α
this␈αwould␈αmake␈αfor␈α
fewer␈αfads␈αand␈αit␈α
would␈αalso
␈↓ α∧␈↓have␈α⊃the␈α⊃important␈α∩social␈α⊃effect␈α⊃that␈α⊃a␈α∩smaller␈α⊃public␈α⊃would␈α⊃be␈α∩needed␈α⊃in␈α⊃order␈α∩to␈α⊃support
␈↓ α∧␈↓someone␈αwho␈αis␈αwriting␈αbecause␈αhe␈αcould␈αget␈αhis␈αincome␈αwith␈αroughly␈αone-fifth␈αof␈αthe␈αnumber␈αof
␈↓ α∧␈↓readers␈αthat␈αhe␈αpresently␈α
needs.␈α Another␈αaspect␈αis␈αthat␈α
much␈αmore␈αdetailed␈αinformation␈αwould␈α
be
␈↓ α∧␈↓available for those who want it.

␈↓ α∧␈↓When␈αwe␈αfirst␈αput␈αthe␈αAssociated␈αPress␈αinto␈αour␈αcomputer␈αI␈αhad␈αthis␈αhope␈αthat␈αthere␈αmust␈αbe␈αall
␈↓ α∧␈↓these␈αgreat␈αstories␈αthat␈αthe␈α
AP␈αput␈αout␈αthat␈αthe␈αnewspapers␈α
don't␈αprint,␈αbut␈αof␈αcourse␈αthat␈α
was␈αa
␈↓ α∧␈↓dumb␈α⊂idea.␈α∂AP␈α⊂is␈α∂a␈α⊂co-op␈α⊂of␈α∂newspapers␈α⊂and␈α∂it␈α⊂is␈α∂not␈α⊂going␈α⊂to␈α∂put␈α⊂out␈α∂any␈α⊂stories␈α⊂that␈α∂the
␈↓ α∧␈↓newspapers␈α⊃aren't␈α⊂likely␈α⊃to␈α⊂print.␈α⊃So␈α⊂it␈α⊃turns␈α⊃out␈α⊂really␈α⊃shallow,␈α⊂and␈α⊃if␈α⊂you␈α⊃consider␈α⊃the␈α⊂AP
␈↓ α∧␈↓correspondent␈αin␈α
a␈αplace␈α
like␈αMoscow␈α
or␈αPeking␈αor␈α
Jerusalem,␈αyou␈α
must␈αsay␈α
that␈αhe␈α
must␈αget␈αa␈α
bit
␈↓ α∧␈↓bored␈αwith␈αhis␈αjob.␈αbecause␈αhe␈αwrites␈αthe␈αmuch␈αsame␈αthing␈αevery␈αday,␈αand␈αhe␈αdoes␈αnot␈αwrite␈αvery
␈↓ α∧␈↓much. It is just packaged in what a newspaper is likely to print.

␈↓ α∧␈↓Another␈α⊃desirable␈α∩social␈α⊃effect␈α⊃of␈α∩on-line␈α⊃publication␈α⊃is␈α∩to␈α⊃make␈α⊃an␈α∩effective␈α⊃right␈α∩of␈α⊃reply.
␈↓ α∧␈↓Politicians␈α∂and␈α∂journalists␈α∂often␈α⊂engage␈α∂in␈α∂hit-and-run␈α∂controversy,␈α⊂in␈α∂what␈α∂he␈α∂says␈α⊂or␈α∂writes
␈↓ α∧␈↓depends␈αto␈αa␈αlarge␈αextent␈αfor␈αits␈αeffect␈αon␈αthe␈αfact␈αthat␈αpeople␈αwho␈αdisagree␈αwith␈αit␈αor␈αthe␈αpeople
␈↓ α∧␈↓who␈αare␈αspecifically␈αattacked␈αwill␈αnot␈αbe␈αable␈αto␈αmount␈αan␈αinstant␈αreply.␈α With␈αthe␈αon-line␈αsystem,
␈↓ α∧␈↓if␈α
you␈α
read␈α
(say)␈α
an␈α
attack␈α
on␈α
the␈α
"Eastern␈α
Establishment"␈α
by␈α
a␈α
Republican␈α
President,␈α∞then␈α
you
␈↓ α∧␈↓could␈α
ask␈αyour␈α
terminal␈α
␈↓↓"What␈αdo␈α
the␈α
Democrats␈αreply␈α
to␈αthat?"␈↓.␈α
 I␈α
think␈αthis␈α
would␈α
enhance␈αthe
␈↓ α∧␈↓quality␈αof␈αdebate␈α
which␈αhas␈αgreatly␈α
declined␈αin␈αthe␈α
United␈αStates␈αat␈α
least␈αin␈αthe␈α
past␈α100␈αyears.␈α
 If
␈↓ α∧␈↓you␈α
read␈α
even␈α
part␈αof␈α
the␈α
Lincoln-Douglas␈α
debates␈α
you␈αdiscover␈α
that␈α
one␈α
of␈α
them␈αcould␈α
propound
␈↓ α∧␈↓to␈αthe␈αother␈αa␈αvery␈αcomplex␈αhypothetical␈αquestion␈αabout␈αwhat␈αthe␈αcountry␈αshould␈αdo␈α"if"␈αand␈αthe
␈↓ α∧␈↓other would actually feel obliged to answer.  That certainly has not happened at all recently.

␈↓ α∧␈↓The␈α∞third␈α
aspect␈α∞of␈α
such␈α∞a␈α
new␈α∞system␈α
which␈α∞I␈α
consider␈α∞desirable␈α
is␈α∞that␈α
it␈α∞would␈α∞enhance␈α
the
␈↓ α∧␈↓ability␈αof␈αa␈αgroup␈αto␈αmaintain␈αits␈αseparateness␈αif␈αit␈αso␈αdesires.␈α They␈αcould␈αbe␈αphysically␈αscattered
␈↓ α∧␈↓but␈α
intellectually␈αmuch␈α
closer␈α
together;␈αthe␈α
cost␈α
of␈αhaving␈α
a␈α
newspaper␈αor␈α
magazine␈α
or␈αwhatever
␈↓ α∧␈↓kind of communication seemed best to them would be greatly reduced from what it is now.

␈↓ α∧␈↓␈↓ αTNow␈α∞I␈α∞want␈α∞to␈α∞finish␈α∞off␈α∞by␈α∂giving␈α∞some␈α∞advice,␈α∞unsolicited␈α∞of␈α∞course.␈α∞We␈α∞will␈α∂give␈α∞this
␈↓ α∧␈↓advice␈α⊃to␈α∩three␈α⊃entities,␈α⊃the␈α∩first␈α⊃of␈α∩which␈α⊃is␈α⊃IBM␈α∩which␈α⊃has␈α⊃generously␈α∩invited␈α⊃me␈α∩to␈α⊃this
␈↓ α∧␈↓symposium.

␈↓ α∧␈↓␈↓ αTThe␈αfirst␈αquestion␈αis␈αwhen␈αare␈αthey␈αgoing␈αto␈αgive␈αup␈αthe␈αpunch␈αcard␈αcheck?␈α The␈αpunched
␈↓ α∧␈↓card␈α
check␈αis␈α
not␈αimportant␈α
in␈αitself␈α
but␈αas␈α
a␈α
symbol␈αof␈α
attitude.␈α Its␈α
use␈αrequires␈α
an␈αadaptation␈α
by
␈↓ α∧␈↓humans␈α⊂to␈α⊂efficient␈α∂data␈α⊂processing␈α⊂by␈α⊂machines.␈α∂ It␈α⊂was␈α⊂acceptable␈α∂in␈α⊂the␈α⊂age␈α⊂of␈α∂mechanical
␈↓ α∧␈↓accounting␈α↔machines,␈α_when␈α↔machines␈α_were␈α↔clumsy␈α↔and␈α_difficult␈α↔to␈α_adapt␈α↔to␈α_people.␈α↔ Its
␈↓ α∧␈↓maintenance␈α∃today␈α⊗serves␈α∃no␈α⊗efficiency␈α∃purpose␈α∃but␈α⊗is␈α∃merely␈α⊗a␈α∃bureaucratic␈α⊗tradition␈α∃in
␈↓ α∧␈↓organizations␈α∩that␈α∩see␈α∩computers␈α∪only␈α∩as␈α∩a␈α∩means␈α∩of␈α∪doing␈α∩old␈α∩things␈α∩cheaper␈α∩and␈α∪put␈α∩no
␈↓ α∧␈↓emphasis␈α
on␈α
using␈α
the␈α∞flexibility␈α
of␈α
computers␈α
too␈α
allow␈α∞he␈α
people␈α
with␈α
whom␈α∞the␈α
organization
␈↓ α∧␈↓deals more convenience and flexibility.

␈↓ α∧␈↓␈↓ αTA␈α∩key␈α∩technological␈α∩requirement␈α∪for␈α∩making␈α∩computers␈α∩and␈α∩bureaucracies␈α∪treat␈α∩people
␈↓ α∧␈↓politely␈α
is␈α
good␈α
time␈αsharing,␈α
and␈α
my␈α
recent␈α
experience␈αwith␈α
IBM␈α
computer␈α
systems␈αhas␈α
confirmed
␈↓ α∧␈↓my␈αold␈α
belief␈αthat␈α
IBM␈αis␈αnot␈α
up␈αto␈α
the␈αstate␈α
of␈αthe␈αart.␈α
 Its␈αterminals␈α
are␈αtoo␈α
complex,␈αit␈αuses␈α
half
␈↓ α∧␈↓␈↓ u7


␈↓ α∧␈↓duplex␈αinteraction,␈αits␈αeditors␈αare␈αinefficient␈αand␈αhard␈αto␈αuse,␈αetc.,etc.␈α I␈αdon't␈αwant␈αto␈αgo␈αinto␈αthat
␈↓ α∧␈↓right now; if they ask me I will tell them.  (Note: they aren't interested).

␈↓ α∧␈↓␈↓ αTIn␈α
this␈α∞connection,␈α
Lewis␈α
Branscomb␈α∞gave␈α
an␈α
interesting␈α∞example␈α
this␈α
morning␈α∞having␈α
to
␈↓ α∧␈↓do␈α∞with␈α∞an␈α∞on-line␈α∞application␈α
that␈α∞was␈α∞interrupted␈α∞for␈α∞some␈α
batch␈α∞processing.␈α∞ If␈α∞there␈α∞is␈α
any
␈↓ α∧␈↓technology␈α∂trade-off␈α∂which␈α∞has␈α∂a␈α∂social␈α∂impact␈α∞it␈α∂is␈α∂the␈α∞distinction␈α∂between␈α∂batch␈α∂and␈α∞on-line
␈↓ α∧␈↓processing.␈α∞ Whatever␈α∞they␈α∞did␈α∂do,␈α∞updating␈α∞this␈α∞data␈α∂base␈α∞with␈α∞batch␈α∞processing␈α∂should␈α∞have
␈↓ α∧␈↓been␈αdone␈αon-line.␈α Indeed␈αI␈αwould␈αclaim␈αthat␈αbatch␈αprocessing␈αhas␈αbeen␈αresponsible␈αfor␈αmany␈αof
␈↓ α∧␈↓the␈αevils␈αthat␈αhave␈αbeen␈αascribed␈αto␈αcomputers.␈α If␈αthe␈αinformation␈αrequired␈αto␈αanswer␈αa␈αquestion
␈↓ α∧␈↓or␈α
correct␈α
an␈αerror␈α
is␈α
on␈αa␈α
tape␈α
sitting␈αon␈α
a␈α
shelf,␈αyou␈α
can't␈α
solve␈αthe␈α
person's␈α
problem␈α
until␈αthe
␈↓ α∧␈↓next␈α∞time␈α∞you␈α∞run␈α∞the␈α∞data␈α∞through.␈α∞ Whereas␈α∂if␈α∞the␈α∞data␈α∞are␈α∞in␈α∞on-line␈α∞files␈α∞available␈α∂to␈α∞the
␈↓ α∧␈↓people␈αwho␈αreceive␈α
the␈αrequests,␈αthey␈α
can␈αanswer␈αquestions␈α
and␈αmake␈αchanges␈α
in␈αthe␈αdata␈αbase␈α
on
␈↓ α∧␈↓the␈αspot.␈α
 Before␈αdisk␈α
files␈αbecame␈αlarge␈α
and␈αcheap,␈α
batch␈αprocessing␈αwas␈α
necessary.␈α If␈α
I␈αweren't
␈↓ α∧␈↓an␈α∞opponent␈α∞of␈α∞regulation,␈α∞I␈α∞would␈α∞favor␈α∞making␈α∞it␈α∞illegal␈α∞in␈α∞applications␈α∞that␈α∞concern␈α∞current
␈↓ α∧␈↓records about people.

␈↓ α∧␈↓␈↓ αTHowever,␈α∞one's␈α∞social␈α∞coonscience␈α∞should␈α∂not␈α∞merely␈α∞be␈α∞concerned␈α∞with␈α∞fighting␈α∂off␈α∞evils.
␈↓ α∧␈↓This␈αis␈αgood␈α
enough␈αfor␈αunimaginative␈αpeople,␈α
but␈αscientists␈αand␈α
technologists␈αshould␈αput␈αmost␈α
of
␈↓ α∧␈↓their␈αefforts␈αinto␈αcreatively␈α
taking␈αadvantage␈αof␈αtechnological␈α
opportunities␈αto␈αimprove␈αthe␈α
quality
␈↓ α∧␈↓of life.

␈↓ α∧␈↓␈↓ αTOne␈α
idea␈α
is␈α
to␈α
urge␈α
businesses␈α
and␈α∞government␈α
agencies␈α
to␈α
make␈α
more␈α
and␈α
more␈α∞of␈α
their
␈↓ α∧␈↓data␈α
bases␈α
directly␈αaccessible␈α
to␈α
any␈α
person␈αwith␈α
a␈α
terminal␈αat␈α
any␈α
hour␈α
of␈αthe␈α
day␈α
or␈α
night.␈α A
␈↓ α∧␈↓customer␈α
should␈α
be␈αable␈α
to␈α
find␈α
the␈αstate␈α
of␈α
his␈α
account␈αeven␈α
at␈α
2:00␈α
in␈αthe␈α
morning,␈α
if␈αthat's␈α
what
␈↓ α∧␈↓he␈α
feels␈α
like␈α
doing,␈α
of␈α
course␈α
with␈α
suitable␈α
safeguards␈α
against␈α
wiping␈α
out␈α
one's␈α
debts␈α
or␈α
reading
␈↓ α∧␈↓other␈α∞people's␈α∞files.␈α∞ A␈α∞simple␈α∞password␈α∂form␈α∞of␈α∞security␈α∞against␈α∞snooping␈α∞will␈α∞be␈α∂adequate␈α∞for
␈↓ α∧␈↓almost␈α
all␈α
people,␈αbut␈α
those␈α
requiring␈α
more␈αshould␈α
be␈α
able␈α
to␈αget␈α
it␈α
by␈α
paying␈αthe␈α
cost␈α
in␈αmoney␈α
or
␈↓ α∧␈↓in lack of access to their own data.  *

␈↓ α∧␈↓␈↓ αTNow␈α⊂I␈α⊂want␈α⊂to␈α⊃give␈α⊂some␈α⊂advice␈α⊂to␈α⊃governments.␈α⊂ First␈α⊂a␈α⊂piece␈α⊃of␈α⊂advice␈α⊂to␈α⊂the␈α⊃U.␈α⊂S.
␈↓ α∧␈↓Government:␈α
there␈α
is␈α
all␈α
this␈α
information␈α
that␈α
is␈α
supposed␈α
to␈α
be␈α
available␈α
according␈α
to␈αFreedom␈α
of
␈↓ α∧␈↓Information,␈αbut␈αof␈αcourse␈αit␈αis␈αsitting␈αin␈αlibraries␈αor␈αreading␈αrooms␈αin␈αthe␈αagencies.␈α While␈αone␈αis
␈↓ α∧␈↓prodding␈αthe␈αgovernment␈αto␈αdo␈αsomething␈αmaybe␈αone␈αshould␈αrequire␈αthat␈αall␈αof␈αthis␈αinformation
␈↓ α∧␈↓should be available on-line on a terminal anywhere in the country.

␈↓ α∧␈↓I have talked enough, so I will skip the rest of my advice.

␈↓ α∧␈↓QUESTIONS AND ANSWERS FOLLOWING PRESENTATION:

␈↓ α∧␈↓J. Raviv:

␈↓ α∧␈↓John␈αMcCarthy␈αsaid␈αthat␈αhe␈αdidn't␈αwant␈αto␈αtalk␈αabout␈αartificial␈αintelligence,␈αbut␈αI␈αdon't␈αthink␈αwe
␈↓ α∧␈↓should␈α
let␈α
him␈α
off␈α
so␈α
easily.␈αI␈α
would␈α
like␈α
to␈α
ask␈α
the␈αfollowing␈α
question:␈α
There␈α
has␈α
been␈α
a␈α
lot␈αof
␈↓ α∧␈↓research␈α
going␈α
on␈αin␈α
artificial␈α
intelligence␈α
and␈αon␈α
robotics␈α
at␈α
Stanford␈αunder␈α
your␈α
direction␈αat␈α
SRI
␈↓ α∧␈↓and␈α
at␈α
MIT.␈α
 I␈α
was␈α
wondering␈α
if␈α
you␈α
could␈α
summarize␈α
what␈α
some␈α
of␈α
the␈α
benefits␈α
are␈α
that␈αwe␈α
have
␈↓ α∧␈↓obtained␈αfrom␈αthis␈αresearch␈αbut␈αnot␈αto␈αdirect␈αthe␈αanswer␈αtowards␈αindustrial␈αrobots␈αbecause␈αthis␈αis
␈↓ α∧␈↓not␈α∩the␈α∪same␈α∩kind␈α∪of␈α∩sophisticated␈α∪work␈α∩that␈α∩you␈α∪are␈α∩leading.␈α∪What␈α∩is␈α∪the␈α∩benefit␈α∪to␈α∩our
␈↓ α∧␈↓profession and to man and society, either right now or in the future?

␈↓ α∧␈↓J. McCarthy:
␈↓ α∧␈↓␈↓ u8


␈↓ α∧␈↓␈↓ αTI␈α∞think␈α∞the␈α∞practical␈α∂benefits␈α∞today␈α∞have␈α∞been␈α∂minimal.␈α∞There␈α∞have␈α∞certainly␈α∂been␈α∞many
␈↓ α∧␈↓beneficial␈α
by-products␈αof␈α
artificial␈αintelligence,␈α
for␈αexample,␈α
time␈αsharing,␈α
LISP,␈αMACSYMA,␈α
etc.,
␈↓ α∧␈↓but␈α
I␈α
believe␈α
that␈α
artifical␈α
intelligence␈α
is␈α
in␈α∞a␈α
basic␈α
research␈α
stage␈α
and␈α
will␈α
remain␈α
in␈α∞that␈α
stage
␈↓ α∧␈↓quite␈α⊂a␈α⊂long␈α∂time␈α⊂to␈α⊂come.␈α∂ Some␈α⊂people␈α⊂are␈α⊂saying␈α∂that␈α⊂since␈α⊂it␈α∂has␈α⊂not␈α⊂produced␈α⊂any␈α∂great
␈↓ α∧␈↓practical␈α∞importance␈α∂in␈α∞twenty␈α∂years␈α∞we␈α∂should␈α∞stop.␈α∞ Should␈α∂we␈α∞have␈α∂said␈α∞that␈α∂about␈α∞genetics
␈↓ α∧␈↓twenty years after Gregor Mendel, which would have been actually before the subject got started.

␈↓ α∧␈↓␈↓ αTOne␈α
slogan␈α
which␈α
I␈α
put␈α
forth␈α
on␈α
that␈αis␈α
that␈α
for␈α
artifical␈α
intelligence␈α
to␈α
reach␈α
a␈αhuman␈α
level
␈↓ α∧␈↓will␈α⊃require␈α⊃1.7␈α⊂Einsteins␈α⊃and␈α⊃.3␈α⊂of␈α⊃a␈α⊃Manhattan␈α⊃Project,␈α⊂and␈α⊃it␈α⊃would␈α⊂be␈α⊃best␈α⊃to␈α⊃have␈α⊂the
␈↓ α∧␈↓Einstein␈α
before␈α
the␈α
Manhattan␈α
Project.␈α
 I␈α
think␈α
that␈α
artificial␈α
intelligence␈α
has␈α
already␈α
had␈αsome
␈↓ α∧␈↓scientific␈αand␈αsome␈αphilosophical␈αbenefits.␈α The␈αscientific␈αbenefits␈αare␈αthat␈αthe␈αcognitive␈αapproach
␈↓ α∧␈↓to␈α∀psychology␈α∀has␈α∪been␈α∀an␈α∀outgrowth␈α∪of␈α∀work␈α∀in␈α∪artificial␈α∀intelligence,␈α∀although␈α∀the␈α∪main
␈↓ α∧␈↓contributors␈αto␈α
this␈αarea,␈αNewell␈α
and␈αSimon␈α
of␈αthe␈αCarnegie-Mellon␈α
group,␈αhave␈α
been␈αinterested
␈↓ α∧␈↓in␈α
both␈α
all␈α∞along.␈α
This␈α
has␈α∞really␈α
demolished␈α
behaviorism␈α
as␈α∞the␈α
paradigm␈α
for␈α∞psychology,␈α
and
␈↓ α∧␈↓this␈α∂demolition␈α∂was␈α∂something␈α∞that␈α∂was␈α∂clearly␈α∂required␈α∂for␈α∞progress.␈α∂ The␈α∂same␈α∂thing␈α∂is␈α∞true,
␈↓ α∧␈↓even␈αthough␈αthe␈αprogress␈αis␈αless␈αadvanced,␈αin␈αphilosophy.␈α There␈αwere␈αmany␈αforms␈αof␈αphilosophy
␈↓ α∧␈↓in␈αempiricism␈αand␈αpositivism,␈αetc.␈αthat␈αwere␈αakin␈αto␈αbehaviorism␈αin␈αmethodology␈αand␈αwhich␈αhave
␈↓ α∧␈↓now␈α∩given␈α∩way␈α∩to␈α∩the␈α∩willingness␈α∩of␈α∩philosophers␈α∩to␈α∩understand␈α∩that␈α∩if␈α∩you␈α∩really␈α∪want␈α∩to
␈↓ α∧␈↓understand␈α⊃thought␈α⊂or␈α⊃belief␈α⊂or␈α⊃something␈α⊃like␈α⊂that␈α⊃then␈α⊂you␈α⊃have␈α⊂to␈α⊃admit␈α⊃rather␈α⊂complex
␈↓ α∧␈↓mechanisms.

␈↓ α∧␈↓␈↓ αTOne␈α∂potential␈α∞benefit␈α∂I␈α∞see␈α∂in␈α∞the␈α∂distance␈α∞may␈α∂be␈α∞just␈α∂mathematical␈α∞imperialism␈α∂on␈α∞my
␈↓ α∧␈↓part.␈α
 Why␈α
can't␈α
society␈α
do␈α
the␈α
right␈α
thing␈αby␈α
establishing␈α
the␈α
basic␈α
facts␈α
as␈α
axioms␈α
of␈α
a␈αtheory
␈↓ α∧␈↓and␈α∂deducing␈α∂that␈α∂a␈α∂certain␈α∂course␈α∂of␈α⊂action␈α∂will␈α∂have␈α∂desirable␈α∂results.␈α∂ We␈α∂could␈α⊂end␈α∂many
␈↓ α∧␈↓controversies,␈α∞because␈α∂everyone␈α∞could␈α∂examine␈α∞the␈α∂basis␈α∞for␈α∂these␈α∞axioms␈α∂and␈α∞decide␈α∂that␈α∞they
␈↓ α∧␈↓agree␈αwith␈α
them␈αand␈α
examine␈αthe␈α
proof␈αthat␈α
the␈αdesired␈α
course␈αof␈α
action␈αwould␈α
have␈αgood␈α
results,
␈↓ α∧␈↓and then we would not be squabbling so much.

␈↓ α∧␈↓Now␈α∂this␈α⊂goal␈α∂of␈α⊂a␈α∂scientific␈α⊂approach␈α∂to␈α⊂social␈α∂problems␈α⊂is␈α∂one␈α⊂which␈α∂requires␈α⊂an␈α∂enormous
␈↓ α∧␈↓advance␈α
in␈αour␈α
ability␈αto␈α
formulate␈α
facts␈αabout␈α
the␈αworld␈α
before␈α
it␈αbecomes␈α
even␈αremotely␈α
feasible.
␈↓ α∧␈↓It␈α∞has␈α∞a␈α∂bad␈α∞name␈α∞because␈α∂it␈α∞is␈α∞very␈α∂easy␈α∞to␈α∞claim␈α∞that␈α∂you␈α∞already␈α∞have␈α∂it,␈α∞and␈α∞you␈α∂can␈α∞put
␈↓ α∧␈↓people in jail who claim that your theory is not scientific.

␈↓ α∧␈↓M. Rabin:

␈↓ α∧␈↓Would␈α⊃you␈α⊃care␈α⊃to␈α⊃comment␈α⊃on␈α⊃the␈α⊂various␈α⊃programming␈α⊃languages␈α⊃which␈α⊃arose␈α⊃out␈α⊃of␈α⊂the
␈↓ α∧␈↓artificial␈αintelligence␈αeffort,␈αthe␈αvarious␈αcapabilities␈αand␈αflexibility␈αthat␈αis␈αafforded␈αto␈αuse␈αthrough
␈↓ α∧␈↓these␈αlanguages␈αand␈αwould␈αyou␈αsay,␈αin␈αyour␈αopinion,␈αdoes␈αthis␈αget␈αus␈αcloser␈αto␈αthe␈αsolution␈αof␈αthe
␈↓ α∧␈↓problem of making the man machine communication, that the first speakers spoke about, easier?

␈↓ α∧␈↓J. McCarthy:

␈↓ α∧␈↓␈↓ αTLISP␈αwas␈αthe␈αfirst␈αlanguage␈αoriented␈αtoward␈αartificial␈αintelligence,␈αand␈αfor␈αalmost␈αten␈αyears,
␈↓ α∧␈↓it␈α
had␈αno␈α
competitor␈α
that␈αsurvived␈α
at␈α
all.␈α Then␈α
we␈α
had␈αMICROPLANNER,␈α
and␈α
then␈αthere␈α
were
␈↓ α∧␈↓a␈α≤quick␈α≥succession␈α≤of␈α≤such␈α≥languages␈α≤that␈α≤killed␈α≥each␈α≤other␈α≤off.␈α≥ I␈α≤kind␈α≥of␈α≤liked
␈↓ α∧␈↓MICROPLANNER;␈α
I␈α
thought␈α
that␈α
in␈α
many␈α
respects␈αit␈α
was␈α
an␈α
improvement␈α
on␈α
LISP,␈α
but␈αthen
␈↓ α∧␈↓things got very complicated and people went back to programming in raw LISP.

␈↓ α∧␈↓The␈αsystems␈αwhich␈αare␈αcombined␈αdata␈αbases␈αand␈αlanguages␈αlike␈αKRL,␈αin␈αmy␈αopinion,␈αsuffer␈αfrom
␈↓ α∧␈↓␈↓ u9


␈↓ α∧␈↓technical␈αdeficiencies;␈αthey␈αare␈αtoo␈αmuch␈αbased␈αon␈αMinsky's␈αframe␈αparadigm␈αand␈αSchank's␈αscripts.
␈↓ α∧␈↓I␈α
don't␈α
think␈α
that␈α
languages␈α
are␈α
artificial␈αintelligence␈α
per␈α
se,␈α
nor␈α
do␈α
I␈α
think␈α
that␈αcommunication
␈↓ α∧␈↓with␈α⊃computers␈α⊃should␈α⊂often␈α⊃be␈α⊃in␈α⊃natural␈α⊂language.␈α⊃ Here␈α⊃is␈α⊂an␈α⊃analogy␈α⊃we␈α⊃that␈α⊂sometimes
␈↓ α∧␈↓convinced␈α
ARPA.␈α
 It␈α
is␈α
indeed␈α
possible␈α
to␈α
program␈α
a␈α
computer␈α
in␈α
English;␈α
it␈α
is␈α
also␈α
possible␈α
to
␈↓ α∧␈↓make an airplane that is controlled by whip and spurs, but neither may be desirable.

␈↓ α∧␈↓J. Raviv:

␈↓ α∧␈↓There␈α∞is␈α
a␈α∞friend␈α
of␈α∞mine␈α
who␈α∞worked␈α∞in␈α
an␈α∞area␈α
of␈α∞artificial␈α
intelligence,␈α∞and␈α
I␈α∞asked␈α∞him␈α
in
␈↓ α∧␈↓1964␈αwhat␈αhe␈αthought␈αwould␈αhappen,␈αwhat␈αare␈αthe␈αhopes,␈αthe␈αpractical␈αapplications.␈αHe␈αsaid␈αthat
␈↓ α∧␈↓in␈αten␈α
years␈αthere␈α
will␈αbe␈α
a␈αrobot␈αdelivering␈α
mail.␈α He␈α
didn't␈αmean␈α
in␈αa␈α
hospital␈αor␈αsomething␈α
that
␈↓ α∧␈↓is␈αhappening␈αnow,␈α
but␈αlike␈αdelivering␈αmail␈α
in␈αthe␈αsuburbs.␈αWas␈α
a␈αstatement␈αlike␈αthat␈α
irresponsible
␈↓ α∧␈↓on his side or was there a hope of sorts or have things changed?

␈↓ α∧␈↓J. McCarthy:

␈↓ α∧␈↓␈↓ αTI␈α∪don't␈α∪think␈α∪the␈α∪statement␈α∪was␈α∪irresponsible␈α∪if␈α∪it␈α∪was␈α∪responsibly␈α∪made.␈α∪ Namely,␈α∪if
␈↓ α∧␈↓someone␈α
asks␈α
you␈α
how␈α
long␈α
something␈αwill␈α
take,␈α
one␈α
way␈α
of␈α
doing␈αit␈α
is␈α
adding␈α
up␈α
the␈α
time␈αyou
␈↓ α∧␈↓imagine␈α⊗it␈α⊗will␈α⊗take␈α⊗to␈α⊗overcome␈α⊗the␈α∃difficulties␈α⊗you␈α⊗recognize.␈α⊗ If␈α⊗it␈α⊗happens␈α⊗there␈α∃are
␈↓ α∧␈↓unrecognized␈αdifficulties␈αor␈αif␈αit␈αhappens␈αthat␈αyour␈αidea␈αof␈αwhat␈αis␈αrequired␈αis␈αmistaken,␈αthen␈αyou
␈↓ α∧␈↓will␈αcome␈αup␈αwith␈αincorrect␈α
estimates.␈α I␈αwill␈αadmit␈αthat␈α
I␈αexpected␈αgreater␈αprogress␈αin␈αthe␈α
robotics
␈↓ α∧␈↓area than has occurred.

␈↓ α∧␈↓␈↓ αTWhen␈αwe␈α
first␈αstarted␈α
our␈αproject␈α
in␈α1965,␈α
I␈αthought␈α
we␈αwould␈α
have␈αprograms␈α
that␈αwould
␈↓ α∧␈↓be␈α
able␈αto␈α
assemble␈αa␈α
Heathkit␈αout␈α
of␈αparts␈α
in␈α
a␈αcouple␈α
of␈αyears.␈α
 Part␈αof␈α
the␈αreason␈α
that␈αwe␈α
aren't
␈↓ α∧␈↓here␈α
yet␈αis␈α
that␈α
the␈αproblem␈α
was␈α
more␈αdifficult␈α
than␈α
I␈αexpected.␈α
 Another␈α
part␈αwas␈α
that␈α
I␈αdidn't
␈↓ α∧␈↓understand␈α⊃what␈α⊃I␈α⊃read␈α⊃somewhere␈α⊃that␈α⊃it␈α⊃takes␈α⊃ten␈α⊃years␈α⊃to␈α⊃develop␈α⊃a␈α⊃scientific␈α⊃laboratory.
␈↓ α∧␈↓There␈α
are␈α
only␈αnow␈α
beginning␈α
to␈αbe␈α
people␈α
who␈αare␈α
really␈α
professional␈αin␈α
robotics.␈α
 There␈αwere
␈↓ α∧␈↓dilettantes␈αlike␈αme␈αwho␈αhad␈αideas␈αin␈αthe␈αfield␈αbut␈αwere␈αnot␈αtotally␈αcommitted␈αto␈αit,␈αand␈αthere␈αwere
␈↓ α∧␈↓people␈α∂who␈α∂had␈α∂disliked␈α∂the␈α∂previous␈α∂branches␈α∂of␈α∂science␈α∂they␈α∂had␈α∂tried␈α∂and␈α∂very␈α⊂often␈α∂they
␈↓ α∧␈↓turned␈α∞out␈α∞to␈α∞dislike␈α
robotics␈α∞also.␈α∞ Now␈α∞we␈α
are␈α∞beginning␈α∞to␈α∞get␈α
a␈α∞core␈α∞of␈α∞professionals␈α∞in␈α
the
␈↓ α∧␈↓field,␈α
and␈α
maybe␈α
progress␈α
will␈α
be␈α
faster␈α
now.␈α
 As␈α
I␈α
said␈α
in␈α
my␈α
lecture␈α
in␈α
Tel-Aviv,␈α
there␈α∞is␈α
too
␈↓ α∧␈↓much␈α∩concentration␈α∩now␈α∩on␈α∩short␈α∩range␈α∪demonstration␈α∩projects.␈α∩In␈α∩some␈α∩sense,␈α∩there␈α∪is␈α∩less
␈↓ α∧␈↓thinking␈αabout␈αbasic␈αscientific␈αproblems␈αof␈αheuristics␈αand␈αepistomology␈αthan␈αthere␈αwas␈αin␈αthe␈αlate
␈↓ α∧␈↓50's when artificial intelligence first got started. I think this has to be reversed.

␈↓ α∧␈↓P. Goldberg:

␈↓ α∧␈↓To␈αwhat␈αextent␈αdo␈αyou␈αthink␈αthat␈αthis␈αproblem␈αwhich␈αyou␈αjust␈αdescribed␈αstems␈αfrom␈αthe␈αfact␈αthat
␈↓ α∧␈↓there are too many people in the field which is not ready for that assault?

␈↓ α∧␈↓J. McCarthy:

␈↓ α∧␈↓I␈αdon't␈αknow.␈α It␈αis␈αhard␈αfor␈αme␈αto␈αsee␈αthat␈αthere␈αare␈αtoo␈αmany␈αrelative␈αto␈αother␈αfields.␈α I␈αthink␈αall
␈↓ α∧␈↓these smart people would do better if they would only work on the real major problems of AI.