/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/FONT#10=ZERO30
␈↓ α∧␈↓␈↓ u1


␈↓ α∧␈↓α␈↓ β←Old and New Models of the Computerized World of the Future

␈↓ α∧␈↓␈↓ αTWhoever␈αasks␈α
someone␈αto␈αspeak␈α
about␈α"Where␈αare␈α
we␈αgoing?"␈αshould␈α
take␈αinto␈αaccount␈α
that
␈↓ α∧␈↓the␈α
prophet␈α
may␈α∞have␈α
prophesied␈α
before,␈α∞and␈α
may␈α
have␈α
some␈α∞responsibility␈α
to␈α
evaluate␈α∞his␈α
old
␈↓ α∧␈↓prophecies.␈α∪ This␈α∀is␈α∪the␈α∀present␈α∪situation,␈α∪because␈α∀today's␈α∪moderator,␈α∀Gene␈α∪Amdahl,␈α∀was␈α∪a
␈↓ α∧␈↓discussant␈αwhen␈αI␈αspoke␈αin␈αan␈αM.I.T.␈αseries␈α
on␈α"Management␈αand␈αthe␈αComputer␈αof␈αthe␈αFuture"␈α
in
␈↓ α∧␈↓1961.

␈↓ α∧␈↓␈↓ αTLike␈αcars,␈α
worlds␈αof␈α
the␈αfuture␈α
come␈αin␈αannual␈α
models,␈αand␈α
I␈αshall␈α
begin␈αwith␈αthe␈α
durability
␈↓ α∧␈↓of my 1961 model.

␈↓ α∧␈↓A 1961 model world of the future.

␈↓ α∧␈↓␈↓ αTThe␈α
1961␈α
future␈αwas␈α
time-sharing.␈α
 The␈α
concept␈αwas␈α
of␈α
a␈αcomputer␈α
system␈α
in␈α
which␈αeach
␈↓ α∧␈↓user␈α⊃at␈α⊃his␈α∩terminal␈α⊃could␈α⊃behave␈α⊃as␈α∩though␈α⊃he␈α⊃had␈α⊃a␈α∩whole␈α⊃computer␈α⊃at␈α⊃his␈α∩disposal,␈α⊃not
␈↓ α∧␈↓concerning␈α
himself␈α
with␈α
other␈α∞users.␈α
 I␈α
proclaimed␈α
this␈α∞superior␈α
to␈α
batch␈α
processing,␈α∞the␈α
current
␈↓ α∧␈↓mode␈αof␈αrunning␈αcomputers␈α
in␈αwhich␈αa␈αuser␈α
prepared␈αa␈αdeck␈αof␈α
cards␈αusing␈αa␈αkeypunch␈α
and␈αan
␈↓ α∧␈↓operator␈α⊂put␈α⊂it␈α⊂on␈α⊂magnetic␈α⊂tape␈α⊂in␈α⊂sequence␈α⊂with␈α⊂the␈α⊂jobs␈α⊂of␈α⊂other␈α⊂users.␈α⊂ Several␈α⊂hours␈α⊂to
␈↓ α∧␈↓several␈α⊂days␈α⊂later,␈α⊂the␈α⊂user␈α⊂would␈α⊂get␈α⊂back␈α⊂output.␈α⊂ My␈α⊂discussion␈α⊂concerned␈α⊂the␈α⊂scientific␈α⊂or
␈↓ α∧␈↓engineering␈α∂programmer,␈α∂who␈α∞needed␈α∂to␈α∂write␈α∞and␈α∂debug␈α∂programs.␈α∞ It␈α∂did␈α∂not␈α∂emphasize␈α∞the
␈↓ α∧␈↓user of already prepared programs.

␈↓ α∧␈↓␈↓ αTIn␈α∞1961␈α∞there␈α
were␈α∞no␈α∞general␈α
purpose␈α∞time-sharing␈α∞systems,␈α
although␈α∞there␈α∞were␈α
systems
␈↓ α∧␈↓that␈α∞time-shared␈α∞between␈α
various␈α∞jobs␈α∞that␈α
had␈α∞been␈α∞debugged␈α
to␈α∞run␈α∞together.␈α∞ An␈α
important
␈↓ α∧␈↓example␈α∂was␈α∂the␈α∂SAGE␈α⊂air␈α∂defense␈α∂system␈α∂which␈α⊂time-shared␈α∂among␈α∂radars␈α∂and␈α⊂people␈α∂with
␈↓ α∧␈↓titles␈αlike␈α"Senior␈αWeapons␈αController"␈αwho␈αsat␈αat␈αspecial␈αpurpose␈αterminals␈αpushing␈αbuttons␈αthat
␈↓ α∧␈↓told fighter pilots where they would find the bombers.

␈↓ α∧␈↓␈↓ αTThe␈αnovelty␈αwas␈αto␈αprovide␈αfor␈αrunning␈αundebugged␈αmachine␈αlanguage␈αprograms␈αtogether.
␈↓ α∧␈↓This␈α∞required␈α∞memory␈α∞protection␈α∞and␈α∞relocation␈α∞and␈α∞a␈α∞user␈α∞mode␈α∞that␈α∞couldn't␈α∂directly␈α∞initiate
␈↓ α∧␈↓input-output.␈α It␈αalso␈α
required␈αa␈αtime-sharing␈α
operator␈αsystem.␈α However,␈α
both␈αGene␈αAmdahl␈α
and
␈↓ α∧␈↓the␈α∂other␈α⊂discussant,␈α∂the␈α∂late␈α⊂John␈α∂Mauchly␈α∂who␈α⊂was␈α∂one␈α∂of␈α⊂the␈α∂inventors␈α∂of␈α⊂stored␈α∂program
␈↓ α∧␈↓computers, agreed on the desirability and feasibility of time-sharing.

␈↓ α∧␈↓␈↓ αTAt␈α⊃the␈α⊃time,␈α⊃I␈α⊃believed␈α⊃that␈α⊃time-sharing␈α⊃would␈α⊃become␈α⊃the␈α⊃dominant␈α⊃means␈α⊃of␈α⊃using
␈↓ α∧␈↓computers␈αby␈α1965,␈αbut␈αit␈αhappened␈αmuch␈αmore␈αslowly.␈α Our␈αArtificial␈αIntelligence␈αLaboratory␈αat
␈↓ α∧␈↓Stanford␈α∪developed␈α∀a␈α∪time-sharing␈α∀system␈α∪in␈α∪1963␈α∀and␈α∪1964,␈α∀and␈α∪we␈α∪were␈α∀able␈α∪to␈α∀buy␈α∪a
␈↓ α∧␈↓commercial␈α⊃time-sharing␈α⊃system␈α∩from␈α⊃Digital␈α⊃Equipment␈α∩Corporation␈α⊃in␈α⊃1965.␈α∩ However,␈α⊃the
␈↓ α∧␈↓undergraduate␈αcomputing␈αat␈αStanford␈αdid␈αnot␈αswitch␈αfrom␈αbatch␈αprocessing␈αto␈αtime-sharing␈αuntil
␈↓ α∧␈↓1976, and most other schools switched somewhat later.

␈↓ α∧␈↓␈↓ αTThere␈α∂are␈α∂still␈α∞important␈α∂bastions␈α∂of␈α∞batch␈α∂processing␈α∂in␈α∞the␈α∂commercial␈α∂world,␈α∂and␈α∞this
␈↓ α∧␈↓technological␈αfact␈αhas␈αa␈αsocial␈αconsequence.␈α Consider,␈αfor␈αexample,␈αa␈αbilling␈αsystem␈αin␈αwhich␈αthe
␈↓ α∧␈↓status␈αof␈αthe␈αvarious␈αaccounts␈αis␈αkept␈αon␈αmagnetic␈αtape,␈αeach␈αtape␈αstoring␈αthe␈αdata␈α
for␈αthousands
␈↓ α∧␈↓of␈α
accounts.␈α
 If␈α
a␈α
customer␈α
complains␈α
of␈α
an␈α
error,␈α
the␈α
clerks␈α
have␈α
no␈α
direct␈α
access␈α
to␈α
the␈αrecord.
␈↓ α∧␈↓They␈αmust␈αrefer␈αto␈αa␈αprinted␈αreport,␈αand␈αone␈αpossible␈αerror␈αis␈αa␈αdiscrepancy␈αbetween␈α
the␈αprinted
␈↓ α∧␈↓report␈αand␈αthe␈αtape␈αfile.␈α For␈αexample,␈αthe␈αwrong␈αtape␈αmay␈αhave␈αbeen␈αused.␈α Correcting␈αthe␈αerror
␈↓ α∧␈↓requires␈αmaking␈αa␈αcard␈αto␈α
modify␈αthe␈αrecord␈αand␈αwaiting␈α
for␈αthe␈αnext␈αupdate␈αrun.␈α Checking␈α
that
␈↓ α∧␈↓the␈α
correction␈α
has␈α
been␈α
made␈α
correctly␈α
must␈αawait␈α
the␈α
next␈α
printout␈α
of␈α
the␈α
account␈α
records.␈α No
␈↓ α∧␈↓␈↓ u2


␈↓ α∧␈↓wonder␈α∞persistent␈α∞errors␈α∞aren't␈α∞rare,␈α∞and␈α∞no␈α∞wonder␈α∞the␈α∞clerks␈α∞and␈α∞complaint␈α∞people␈α∞blame␈α∞the
␈↓ α∧␈↓computer.  In a sense they are correct in doing so.

␈↓ α∧␈↓␈↓ αTContrast␈α∂this␈α∂with␈α∂an␈α∂on-line␈α∞system␈α∂in␈α∂which␈α∂the␈α∂account␈α∞is␈α∂on␈α∂disk␈α∂file␈α∂that␈α∂is␈α∞always
␈↓ α∧␈↓directly␈α∞accessible␈α∞to␈α∞the␈α∂clerks.␈α∞ A␈α∞recognized␈α∞error␈α∞can␈α∂be␈α∞corrected␈α∞immediately,␈α∞and␈α∂the␈α∞new
␈↓ α∧␈↓record␈α∞can␈α∞immediately␈α
be␈α∞inspected.␈α∞ Today␈α∞there␈α
are␈α∞many␈α∞on-line␈α
systems,␈α∞but␈α∞there␈α∞are␈α
still
␈↓ α∧␈↓many too batch systems giving computers a bad name.

␈↓ α∧␈↓␈↓ αTWhy␈α⊂has␈α⊂it␈α⊂taken␈α⊂so␈α⊂long␈α⊃for␈α⊂the␈α⊂time-sharing␈α⊂mode␈α⊂of␈α⊂computer␈α⊂operation␈α⊃to␈α⊂become
␈↓ α∧␈↓dominant, and why is the process still incomplete?  Here are some speculations.

␈↓ α∧␈↓␈↓ αT1.␈α
There␈α∞were␈α
no␈α
operating␈α∞time-sharing␈α
systems␈α
in␈α∞1961.␈α
 The␈α
first␈α∞experimental␈α
systems
␈↓ α∧␈↓were␈α
a␈α
year␈α∞later,␈α
and␈α
the␈α∞Digital␈α
Equipment␈α
Corporation␈α∞PDP-6␈α
didn't␈α
become␈α∞available␈α
until
␈↓ α∧␈↓1964, and then it wasn't entirely satisfactory.

␈↓ α∧␈↓␈↓ αT2.␈αTime-sharing␈αturned␈αout␈αto␈αrequire␈αgreater␈αcomputer␈αresources␈αthan␈αanticipated.␈α This␈αis
␈↓ α∧␈↓partly␈αa␈α
fact␈αof␈α
technology,␈αbut␈αit␈α
is␈αpartly␈α
a␈αpsychological␈αfact.␈α
 Computer␈αscientists␈α
and␈αhackers
␈↓ α∧␈↓have␈α⊃shown␈α⊃great␈α⊃interest␈α⊃in␈α⊃adding␈α⊃features␈α⊃to␈α⊃time-sharing␈α⊃and␈α⊃other␈α⊃on-line␈α⊃systems␈α⊂and
␈↓ α∧␈↓almost␈αno␈αinterest␈αin␈αfinding␈αthe␈αsources␈αof␈αinefficiency␈αand␈αcorrecting␈αthem.␈α To␈αgive␈αa␈αtechnical
␈↓ α∧␈↓example,␈α∂in␈α∞one␈α∂of␈α∂our␈α∞systems␈α∂at␈α∞Stanford,␈α∂the␈α∂time␈α∞required␈α∂for␈α∞the␈α∂system␈α∂to␈α∞open␈α∂a␈α∂file␈α∞is
␈↓ α∧␈↓proportional,␈α
for␈α
no␈α
good␈αreason,␈α
to␈α
the␈α
number␈αof␈α
files␈α
already␈α
open.␈α With␈α
a␈α
small␈α
number␈αof
␈↓ α∧␈↓users␈α
on␈α
the␈α
machine␈αthe␈α
effect␈α
is␈α
not␈αnoticeable,␈α
but␈α
when␈α
one␈αbuys␈α
a␈α
compatible␈α
machine␈αfive
␈↓ α∧␈↓times␈α
as␈α
fast,␈α
one␈α
is␈α
unpleasantly␈α
surprised␈α
that,␈α
for␈α
unknown␈α
reasons,␈α
it␈α
won't␈α
handle␈α
five␈α
times␈α
as
␈↓ α∧␈↓many users with the same service.

␈↓ α∧␈↓␈↓ αT3.␈αIBM's␈αnew␈αcomputer␈αof␈αthe␈α1960s,␈αthe␈α360,␈αdid␈αnot␈αhave␈αthe␈αrelocation␈αfeatures␈αrequired
␈↓ α∧␈↓for␈α∂time-sharing␈α∂except␈α∂in␈α∂the␈α∂special␈α∂Model␈α∂67.␈α∂ These␈α∂features␈α∂became␈α∂standard␈α∂only␈α∂in␈α∂the
␈↓ α∧␈↓middle␈α⊃1970s.␈α∩ Gene␈α⊃Amdahl␈α⊃can␈α∩say␈α⊃better␈α∩than␈α⊃I␈α⊃can␈α∩how␈α⊃this␈α⊃came␈α∩about.␈α⊃ We␈α∩are␈α⊃still
␈↓ α∧␈↓suffering␈αfrom␈αthis␈αmistake␈αtoday,␈αbecause,␈α
according␈αto␈αmy␈αinformants,␈αmany␈αcompanies␈α
are␈αstill
␈↓ α∧␈↓wedded to older IBM software in which time-sharing is grafted onto a basically batch system.

␈↓ α∧␈↓␈↓ αT4.␈α
IBM␈αfollowed␈α
M.I.T.␈αinto␈α
the␈αswamp␈α
of␈αover-elaboration␈α
in␈αtheir␈α
time-sharing␈αsystems␈α
of
␈↓ α∧␈↓the middle 1960s.  The reader is warned that this opinion may not receive universal assent.

␈↓ α∧␈↓␈↓ αT5.␈α
I␈α
must␈α
confess␈α
that␈α
my␈α
1961␈α
paper␈α
does␈α
not␈α
offer␈α
any␈α
arguments␈α
for␈α
the␈α∞importance␈α
of
␈↓ α∧␈↓on-line␈α∂operation␈α⊂in␈α∂the␈α∂commercial␈α⊂uses␈α∂of␈α∂computers,␈α⊂although␈α∂I␈α∂had␈α⊂had␈α∂the␈α⊂thought.␈α∂ My
␈↓ α∧␈↓interest␈α∞was␈α∞in␈α∞scientific␈α∂computation␈α∞where␈α∞the␈α∞first␈α∞successful␈α∂debugging␈α∞run␈α∞of␈α∞a␈α∂program␈α∞is
␈↓ α∧␈↓often␈α∀the␈α∀only␈α∃production␈α∀run.␈α∀ It␈α∃would␈α∀be␈α∀interesting␈α∀to␈α∃know␈α∀when␈α∀the␈α∃arguments␈α∀for
␈↓ α∧␈↓commercial␈α
on-line␈α
systems␈α
-␈α
other␈α
than␈α
reservation␈α
systems␈α
-␈α
were␈α
first␈α
advanced.␈α
 It␈α
would␈α
be
␈↓ α∧␈↓also␈α
interesting␈α
to␈α
know␈α
if␈α
there␈α
have␈αbeen␈α
social␈α
arguments␈α
in␈α
addition␈α
to␈α
ordinary␈αcommercial
␈↓ α∧␈↓arguments.

␈↓ α∧␈↓A 1965 model world of the future.

␈↓ α∧␈↓␈↓ αTNow␈αlet␈αme␈αturn␈αto␈αa␈α
1965␈αmodel␈αworld␈αof␈αthe␈αfuture.␈α
 I␈αthought␈αit␈αwas␈αPaul␈αBaran's,␈αbut␈α
it
␈↓ α∧␈↓isn't␈αin␈α
the␈αpaper␈α
where␈αI␈α
remembered␈αit␈αto␈α
be.␈α He␈α
remembers␈αit␈α
vaguely␈αbut␈α
isn't␈αsure␈αwhether␈α
it
␈↓ α∧␈↓was his.

␈↓ α∧␈↓␈↓ αTA␈α
clerk␈α
in␈α
Company␈α
A␈α
hears␈αa␈α
beep␈α
and␈α
turns␈α
to␈α
her␈αterminal.␈α
 On␈α
the␈α
screen␈α
there␈α
is␈αa
␈↓ α∧␈↓␈↓ u3


␈↓ α∧␈↓message␈α∞from␈α∞the␈α∂inventory␈α∞control␈α∞computer␈α∂saying:␈α∞"We␈α∞need␈α∂5000␈α∞number␈α∞2␈α∂pencils.␈α∞ Please
␈↓ α∧␈↓order␈αthem␈αfrom␈αCompany␈αB".␈α The␈αclerk␈αturns␈α
to␈αher␈αtypewriter␈αand␈αtypes␈αa␈αpurchase␈αorder␈α
and
␈↓ α∧␈↓mails␈αit.␈α In␈αCompany␈αB,␈αanother␈αclerk␈αreads␈αthe␈αpurchase␈αorder,␈αturns␈αto␈αher␈αterminal␈αand␈αtypes:
␈↓ α∧␈↓"Send Company A 5000 number 3 pencils and a bill".

␈↓ α∧␈↓␈↓ αTAs␈αfar␈αas␈α
I␈αcan␈αremember,␈α
this␈αworld␈αwas␈α
depicted␈αwith␈αa␈α
perfectly␈αstraight␈αface.␈α
 Moreover,
␈↓ α∧␈↓it was an excellent prophecy.  In the main that world has arrived.

␈↓ α∧␈↓1983 model worlds of the future.

␈↓ α∧␈↓␈↓ αTThe␈α
first␈α1983␈α
model␈α
relates␈αto␈α
the␈α
1965␈αmodel.␈α
 How␈α
are␈αthe␈α
two␈α
computers␈αgoing␈α
to␈αtalk␈α
to
␈↓ α∧␈↓each␈α
other␈αdirectly␈α
while␈α
the␈αformer␈α
clerks␈α
loll␈αon␈α
the␈α
beach?␈α They␈α
are␈α
going␈αto␈α
talk␈α
to␈αeach␈α
other
␈↓ α∧␈↓in␈α∞CBCL␈α∂(the␈α∞Common␈α∞Business␈α∂Communication␈α∞Language).␈α∞ See␈α∂(McCarthy␈α∞1982)␈α∞if␈α∂you␈α∞can
␈↓ α∧␈↓find it.

␈↓ α∧␈↓␈↓ αTCBCL␈α
is␈αintended␈α
as␈α
a␈αlanguage␈α
in␈αwhich␈α
a␈α
computers␈αbelonging␈α
to␈αCompanies␈α
A␈α
and␈αB
␈↓ α∧␈↓can conduct such dialogs as:

␈↓ α∧␈↓␈↓ αTA: What is your price for number 2 pencils?

␈↓ α∧␈↓␈↓ αTB:␈α
$1.13␈α
per␈α
gross␈α
plain␈α
and␈α
$1.40␈α
per␈α
gross␈α
with␈α
your␈α
message␈α
provide␈α
at␈α
least␈α∞100␈α
gross
␈↓ α∧␈↓are ordered.

␈↓ α∧␈↓␈↓ αTA:␈αI␈αhereby␈α
order␈α(authorization␈α30621)␈α
1000␈αgross␈αwith␈α
the␈αmessage␈α"A␈α
fool␈αand␈αhis␈α
money
␈↓ α∧␈↓are soon parted".  We propose standard commercial contract 361.

␈↓ α∧␈↓␈↓ αTB: How do you want them delivered?

␈↓ α∧␈↓␈↓ αTA: 100 gross per month starting in 1990 January.

␈↓ α∧␈↓␈↓ αTB: I accept your order authorization 30621.  My authorization is 77522.

␈↓ α∧␈↓␈↓ αTThis␈α⊂dialog␈α⊂is␈α⊂more␈α⊃elaborate␈α⊂than␈α⊂most␈α⊂transactions␈α⊂will␈α⊃require,␈α⊂but␈α⊂it␈α⊂is␈α⊃designed␈α⊂to
␈↓ α∧␈↓illustrate␈α
some␈α
of␈α∞the␈α
problems␈α
and␈α∞issues.␈α
 At␈α
first␈α∞I␈α
thought␈α
the␈α∞CBCL␈α
problem␈α
was␈α∞a␈α
simple
␈↓ α∧␈↓matter␈α∃of␈α∃standardization␈α∃like␈α∃the␈α∃universal␈α∃product␈α∃code.␈α∃ A␈α∃committee␈α∃of␈α∃the␈α∀American
␈↓ α∧␈↓Standards␈αAssociation␈αcould␈αhold␈αmeetings␈αfor␈αthree␈αor␈αfour␈αyears,␈αproduce␈αa␈αdocument,␈αand␈αthen
␈↓ α∧␈↓manufacturers␈αcould␈αimplement␈αsystems␈αthat␈αwould␈αcommunicate␈αin␈αaccordance␈αwith␈αthe␈α
protocols.
␈↓ α∧␈↓This␈α∪was␈α∀a␈α∪mistake.␈α∀ The␈α∪problem␈α∀is␈α∪rather␈α∀to␈α∪define␈α∀the␈α∪semantics␈α∀of␈α∪a␈α∀substantial␈α∪and
␈↓ α∧␈↓interesting subset of English.  Here are some of the points that arose.

␈↓ α∧␈↓␈↓ αT1.␈αSince␈αthe␈αcommunication␈αis␈αbetween␈αcomputers,␈αthere␈αis␈αno␈αreason␈αto␈αuse␈αEnglish␈αsyntax.
␈↓ α∧␈↓In␈α
fact␈α
I␈α
believe␈α
that␈α
a␈αLISP␈α
type␈α
syntax␈α
is␈α
suitable.␈α
 Each␈αnon-atomic␈α
part␈α
of␈α
the␈α
message␈α
is␈αa
␈↓ α∧␈↓list,␈α∞and␈α∞the␈α∂first␈α∞element␈α∞of␈α∂the␈α∞list␈α∞identifies␈α∞the␈α∂overall␈α∞meaning␈α∞of␈α∂the␈α∞part.␈α∞ We␈α∂thus␈α∞have
␈↓ α∧␈↓(PENCIL␈α∂(TYPE:␈α∂"NUMBER␈α∂2"))␈α∂as␈α∂an␈α∂item␈α∂and␈α∂perhaps␈α∂(PENCIL␈α∂(TYPE:␈α∂"NUMBER␈α∂2")
␈↓ α∧␈↓(WITH (MESSAGE "A fool and his money are soon parted"))).

␈↓ α∧␈↓␈↓ αT2.␈α∞Most␈α∞work␈α∞on␈α∞making␈α∞computers␈α∂use␈α∞English␈α∞has␈α∞involved␈α∞making␈α∞the␈α∂computer␈α∞send
␈↓ α∧␈↓and␈α⊃receive␈α∩the␈α⊃usual␈α∩kinds␈α⊃of␈α∩messages␈α⊃people␈α∩receive␈α⊃from␈α∩and␈α⊃send␈α∩to␈α⊃computers␈α∩-␈α⊃only
␈↓ α∧␈↓expressing␈α∩them␈α∩in␈α∩English.␈α∪ The␈α∩business␈α∩communication␈α∩problem␈α∩involves␈α∪messages␈α∩whose
␈↓ α∧␈↓␈↓ u4


␈↓ α∧␈↓content␈α⊃is␈α⊃presently␈α⊃only␈α⊃expressed␈α∩in␈α⊃English␈α⊃but␈α⊃where␈α⊃we␈α∩are␈α⊃willing␈α⊃to␈α⊃use␈α⊃any␈α∩kind␈α⊃of
␈↓ α∧␈↓formalism␈α∂convenient␈α∂to␈α∂program.␈α∂ In␈α∂my␈α∂view,␈α⊂contrary␈α∂to␈α∂what␈α∂I␈α∂take␈α∂to␈α∂be␈α⊂the␈α∂Chomskyan
␈↓ α∧␈↓emphasis␈αon␈αsyntax,␈αthe␈αsecond␈αproblem␈αis␈αscientifically␈αmore␈αinteresting,␈αthough␈αadmittedly␈αmore
␈↓ α∧␈↓difficult.  My fellow AI people have also found it more convenient to concentrate on syntax.

␈↓ α∧␈↓␈↓ αTIn␈α⊃summary,␈α⊂broadening␈α⊃the␈α⊂usefulness␈α⊃of␈α⊂computers␈α⊃in␈α⊂the␈α⊃next␈α⊂22␈α⊃years,␈α⊃will␈α⊂require
␈↓ α∧␈↓making␈α∂them␈α∂communicate␈α∞substantively␈α∂with␈α∂each␈α∂other.␈α∞ This␈α∂will␈α∂require␈α∂the␈α∞understanding
␈↓ α∧␈↓and the formalization of an interesting part of the semantics (not the syntax) of natural language.

␈↓ α∧␈↓Another (though related) world of the future.

␈↓ α∧␈↓␈↓ αTMore␈α∩and␈α∩more,␈α∩the␈α∩general␈α∩public␈α⊃will␈α∩find␈α∩itself␈α∩interacting␈α∩directly␈α∩with␈α⊃computers.
␈↓ α∧␈↓There␈α∂will␈α∂be␈α∂personal␈α∂computers,␈α∂but␈α∂there␈α∂will␈α∂also␈α∂be␈α∂computers␈α∂belonging␈α∂to␈α∂the␈α⊂sellers␈α∂of
␈↓ α∧␈↓products␈α⊃and␈α⊃services,␈α⊃to␈α⊃various␈α⊃levels␈α⊃of␈α⊃government␈α⊃and␈α⊃to␈α⊃educational␈α⊃and␈α⊃other␈α⊃private
␈↓ α∧␈↓organizations.␈α∩ Much␈α⊃useful␈α∩interaction␈α⊃can␈α∩be␈α⊃carried␈α∩out␈α⊃with␈α∩a␈α⊃level␈α∩of␈α∩understanding␈α⊃of
␈↓ α∧␈↓computers␈α∀corresponding␈α∀to␈α∀the␈α∪understanding␈α∀of␈α∀electricity␈α∀required␈α∪to␈α∀use␈α∀a␈α∀light␈α∪switch.
␈↓ α∧␈↓However,␈α
many␈α
useful␈αservices␈α
that␈α
computers␈αcan␈α
perform␈α
are␈αcomplex␈α
in␈α
their␈α
interaction␈αand
␈↓ α∧␈↓what a user must understand about their internal state is complex.

␈↓ α∧␈↓␈↓ αTFortunately,␈α⊃we␈α∩humans␈α⊃already␈α∩have␈α⊃language␈α⊃suited␈α∩to␈α⊃expressing␈α∩our␈α⊃understanding
␈↓ α∧␈↓much␈α∩about␈α∪the␈α∩internal␈α∪states␈α∩of␈α∪computer␈α∩based␈α∪systems.␈α∩ This␈α∪is␈α∩the␈α∪naive␈α∩psychological
␈↓ α∧␈↓language␈α∂used␈α∂to␈α∂express␈α⊂what␈α∂people␈α∂know␈α∂about␈α∂each␈α⊂other.␈α∂ I␈α∂begin␈α∂with␈α∂an␈α⊂overly␈α∂simple
␈↓ α∧␈↓illustrative example:

␈↓ α∧␈↓␈↓↓"Place␈αthe␈αcontrol␈αnear␈αthe␈αbed␈αin␈αa␈αplace␈αthat␈αis␈αneither␈αhotter␈αnor␈αcolder␈αthan␈αthe␈αroom␈α
itself.␈α If
␈↓ α∧␈↓↓the␈α
control␈α
is␈αplaced␈α
on␈α
a␈αradiator␈α
or␈α
radiant␈αheated␈α
floors,␈α
it␈αwill␈α
"think"␈α
the␈αentire␈α
room␈α
is␈αhot␈α
and
␈↓ α∧␈↓↓will␈α∂lower␈α⊂your␈α∂blanket␈α∂temperature,␈α⊂making␈α∂your␈α⊂bed␈α∂too␈α∂cold.␈α⊂ If␈α∂the␈α∂control␈α⊂is␈α∂placed␈α⊂on␈α∂the
␈↓ α∧␈↓↓window␈α
sill␈αin␈α
a␈αcold␈α
draft,␈αit␈α
will␈α"think"␈α
the␈αentire␈α
room␈αis␈α
cold␈αand␈α
will␈αheat␈α
up␈αyour␈α
bed␈α
so␈αit
␈↓ α∧␈↓↓will be too hot."␈↓ - from the instructions to an electric blanket.

␈↓ α∧␈↓␈↓ αTThe␈α
manufacturers␈α
of␈α
the␈αblanket␈α
imagined␈α
that␈α
some␈αof␈α
their␈α
customers␈α
would␈α
know␈αless
␈↓ α∧␈↓and␈αwould␈αbe␈αhelped␈αby␈αthe␈αpsychological␈αlanguage.␈α We␈αwho␈αunderstand␈αthermostats␈αdon't␈αneed
␈↓ α∧␈↓to␈α∃refer␈α∃to␈α∃what␈α∀the␈α∃thermostat␈α∃"thinks",␈α∃because␈α∀the␈α∃language␈α∃of␈α∃machinery␈α∃is␈α∀adequate.
␈↓ α∧␈↓However,␈αI␈αwill␈αmaintain␈αthat␈αthe␈αusage␈αis␈αlegitimate␈αand␈αmay␈αbe␈αuseful␈αin␈αrelation␈αto␈αthe␈αblanket
␈↓ α∧␈↓and␈αthat␈αit␈αwill␈αbe␈αalmost␈αessential␈α
to␈αexpress␈αwhat␈αpeople␈αneed␈αto␈αknow␈αabout␈α
certain␈αimportant
␈↓ α∧␈↓computer based systems.

␈↓ α∧␈↓␈↓ αTThe␈α∞issue␈α
isn't␈α∞whether␈α
the␈α∞thermostat␈α
"really␈α∞thinks".␈α
 We␈α∞merely␈α
observe␈α∞that␈α
there␈α∞is␈α
a
␈↓ α∧␈↓useful␈α⊗correspondence␈α↔between␈α⊗a␈α⊗person␈α↔being␈α⊗misled␈α⊗by␈α↔observation␈α⊗and␈α↔the␈α⊗thermostat
␈↓ α∧␈↓measuring␈αthe␈αtemperature␈α
of␈αthe␈αradiator␈α
when␈αit␈αshould␈α
have␈αbeen␈αmeasuring␈α
the␈αtemperature
␈↓ α∧␈↓of the room.

␈↓ α∧␈↓␈↓ αTWe␈α
will␈α
find␈α∞ourselves␈α
referring␈α
to␈α
what␈α∞the␈α
computer␈α
␈↓↓knows,␈↓␈α
␈↓↓believes,␈↓␈α∞␈↓↓intends,␈↓␈α
␈↓↓expects,␈↓
␈↓ α∧␈↓␈↓↓promises␈↓␈αand␈α
␈↓↓wants.␈αThe␈↓␈α
finicky␈αmay␈α
begin␈αby␈αputting␈α
␈↓↓pseudo-␈↓␈αin␈α
front␈αof␈α
each␈αof␈α
these␈αwords,
␈↓ α∧␈↓but␈α
they'll␈αsoon␈α
get␈α
tired␈αof␈α
it.␈α
 Such␈αusage␈α
will␈α
help␈αus␈α
form␈α
and␈αexpress␈α
our␈α
expectations␈αof␈α
what
␈↓ α∧␈↓use␈αwe␈αcan␈αget␈αout␈αof␈αa␈αparticular␈αcomputer␈αsystem.␈α Computers␈αwill␈αturn␈αout␈αto␈αhave␈αvaried␈αlittle
␈↓ α∧␈↓minds,␈αand␈α
wholesale␈αascription␈αof␈α
mental␈αqualities␈αwill␈α
be␈αunwarranted.␈α The␈α
thermostat␈αcannot
␈↓ α∧␈↓usefully␈α∂be␈α∂said␈α∞to␈α∂know␈α∂that␈α∞it␈α∂believes␈α∂the␈α∞room␈α∂is␈α∂too␈α∞hot␈α∂and␈α∂it␈α∞cannot␈α∂␈↓↓realize␈↓␈α∂that␈α∂it␈α∞was
␈↓ α∧␈↓mistaken;␈α
it␈αmerely␈α
comes␈αto␈α
have␈αa␈α
different␈αbelief.␈α
 I␈αdon't␈α
think␈αwe␈α
will␈αfind␈α
it␈αin␈α
our␈αinterest␈α
to
␈↓ α∧␈↓build machines that can be reasonably said to ␈↓↓suffer,␈↓ ␈↓↓love,␈↓ or ␈↓↓have␈↓ ␈↓↓rights.␈↓
␈↓ α∧␈↓␈↓ u5


␈↓ α∧␈↓␈↓ αTHere is a more significant example:

␈↓ α∧␈↓␈↓ αT␈↓↓I␈α∞asked␈α∞the␈α∞computer␈α
whether␈α∞it␈α∞could␈α∞run␈α∞my␈α
job␈α∞at␈α∞4pm.␈α∞ It␈α
said␈α∞it␈α∞could␈α∞but␈α∞it␈α
probably
␈↓ α∧␈↓↓wouldn't.␈α∂ At␈α∂that␈α∂point␈α∂it␈α∂crashed␈α∂so␈α⊂I␈α∂couldn't␈α∂ask␈α∂more␈α∂questions.␈α∂ I␈α∂don't␈α∂know␈α⊂whether␈α∂it
␈↓ α∧␈↓↓thought␈α
I␈α
was␈α
unauthorized␈α
or␈α
had␈α
exceeded␈α∞my␈α
quota␈α
of␈α
resources.␈α
 I␈α
also␈α
don't␈α
know␈α∞whether␈α
it
␈↓ α∧␈↓↓intends to run the job later␈↓.

␈↓ α∧␈↓␈↓ αTSome of these philosophically controversial issues are discussed in (McCarthy 1979).

␈↓ α∧␈↓References:

␈↓ α∧␈↓␈↓αMcCarthy,␈αJohn␈α
(1962)␈↓:␈α"Time-Sharing␈αComputing␈α
Systems,"␈αin␈α␈↓↓Management␈α
and␈αthe␈αComputer␈α
of
␈↓ α∧␈↓↓the Future␈↓, Martin Greenberger (ed.), MIT Press.

␈↓ α∧␈↓␈↓αMcCarthy,␈αJohn␈α(1979)␈↓:␈α
"Ascribing␈αMental␈αQualities␈αto␈α
Machines"␈αin␈α␈↓↓Philosophical␈αPerspectives␈α
in
␈↓ α∧␈↓↓Artificial Intelligence␈↓, Ringle, Martin (ed.), Harvester Press, July 1979.

␈↓ α∧␈↓␈↓αMcCarthy,␈α
John␈α
(1982)␈↓:␈α
"Common␈α
Business␈α
Communication␈α
Language",␈α
in␈α∞␈↓↓Textverarbeitung␈α
und
␈↓ α∧␈↓↓B␈↓
:␈↓↓urosysteme␈↓,␈α
Albert␈αEndres␈α
and␈αJ␈↓
:␈↓urgen␈α
Reetz,␈αeds.␈α
 R.␈αOldenbourg␈α
Verlag,␈αMunich␈α
and␈αVienna
␈↓ α∧␈↓1982.

␈↓ α∧␈↓John McCarthy
␈↓ α∧␈↓Computer Science Department
␈↓ α∧␈↓Stanford University
␈↓ α∧␈↓Stanford, California 94305