/FONT#0=BASL30/FONT#1=BASI30/FONT#2=NGR40/FONT#3=NGR25/FONT#4=NGR20/FONT#15=STA200
␈↓␈↓↓␈↓α␈↓β␈↓∧␈↓∂␈β↓↓␈↓∂S␈↓α
␈↓ ↓H
␈β↓Q␈↓ ¬	STANFORD UNIVERSITY
␈↓ ↓H
␈↓β␈↓ ¬_STANFORD, CALIFORNIA 94305
␈↓ ↓H
␈↓∧
␈↓ ↓H

␈↓ ↓H

␈↓ ↓H

␈↓ ↓H
ARTIFICIAL INTELLIGENCE LABORATORY␈↓ 
+   Telephone:
␈↓ ↓H
COMPUTER SCIENCE DEPARTMENT␈↓ 
+415-497-4430
␈↓ ↓H
␈↓␈↓ ¬h2 December 1974
␈↓ ↓H

␈↓ ↓H

␈↓ ↓H
Mr. David L. Bowen
␈↓ ↓H
Vice President and Director of Communications
␈↓ ↓H
The Associated Press
␈↓ ↓H
50 Rockefeller Plaza
␈↓ ↓H
New York, New York 10020
␈↓ ↓H

␈↓ ↓H
Dear Dave:
␈↓ ↓H

␈↓ ↓H
␈↓ α_Here␈αis␈αa␈αwriteup␈αon␈αthe␈αnew␈αversion␈αof␈αthe␈αnews␈αprogram␈αthat␈αincludes␈αthe␈α␈↓↓New␈αYork␈αTimes␈↓
␈↓ ↓H
␈↓ ↓Has␈αwell␈αas␈α
A.P.␈α
It␈α
also␈α
contains␈α
other␈α
improvements␈α
that␈α
we␈α
told␈α
you␈α
we␈α
planned␈α
to␈α
make.␈α
It␈α
can␈α
store
␈↓ ↓H
␈↓ ↓Hnews␈αfor␈αan␈αarbitrary␈αtime,␈αdepending␈αonly␈αon␈αhow␈αmuch␈αdisk␈αspace␈αwe␈αallocate␈αto␈αit,␈αand␈αit␈αallows
␈↓ ↓H
␈↓ ↓Harbitrary␈α
keyword␈α
expressions.
␈↓ ↓H

␈↓ ↓H
␈↓ α_The␈α
following␈α
remarks␈α
can␈α
be␈α
made:
␈↓ ↓H

␈↓ ↓H
␈↓ α_1.␈αWe␈αare␈αstill␈αnot␈α
satisfied␈α
with␈α
the␈α
ease␈α
of␈α
use␈α
for␈α
inexperienced␈α
people.␈α
When␈α
we␈α
can␈α
afford
␈↓ ↓H
␈↓ ↓Hto␈α∞put␈α∞a␈α∞terminal␈α∞in␈α∞a␈α∞Stanford␈α∞library␈α
with␈α
a␈α
placard␈α
explaining␈α
its␈α
use,␈α
we␈α
will␈α
get␈α
more␈α
useful
␈↓ ↓H
␈↓ ↓Hdata.␈α∂(Of␈α∂course,␈α∂we␈α∂think␈α∂our␈α∂system␈α∂has␈α∂always␈α∞been␈α∞easier␈α∞to␈α∞use␈α∞than␈α∞the␈α∞NYT␈α∞Data␈α∞Bank).
␈↓ ↓H
␈↓ ↓HIncidentally,␈α
some␈α
people␈α
from␈α
there␈α
visited␈α
us␈αand␈αwe␈αsuggested␈αthat␈αwe␈αmight␈αjoin␈αin␈αa␈αStanford
␈↓ ↓H
␈↓ ↓Hsubscription␈αand␈αexperiment␈αwith␈αallowing␈α
our␈α
users␈α
to␈α
query␈α
the␈α
system␈α
in␈α
a␈α
uniform␈α
format␈α
and␈α
to
␈↓ ↓H
␈↓ ↓Htranslate␈αthe␈αquery␈αto␈α
their␈α
system␈α
when␈α
the␈α
query␈α
was␈α
appropriate␈α
to␈α
their␈α
data␈α
bank.␈α
The␈α
visitor␈α
(I
␈↓ ↓H
␈↓ ↓Hforget␈α
his␈α
name,␈α
but␈α
not␈α
John␈α
Rothman,␈α
the␈α
proprietor␈α
of␈α
the␈α
system)␈α
was␈α
quite␈α
unenthusiastic␈α
about
␈↓ ↓H
␈↓ ↓Hhaving␈αtheir␈αsystem␈αused␈αin␈αany␈αbut␈αthe␈αabsolutely␈αstandard␈αway␈αeven␈αthough␈αall␈αqueries␈αwould␈αbe
␈↓ ↓H
␈↓ ↓Hproperly␈α
paid␈α
for.
␈↓ ↓H

␈↓ ↓H
␈↓ α_(Now␈αwe␈αare␈αin␈αthe␈αprocess␈αof␈αtrying␈αto␈αsubscribe,␈αbut␈αwe␈αaren't␈αsure␈αwe␈αcan␈αafford␈αit,␈αand␈αwe
␈↓ ↓H
␈↓ ↓Hdon't␈α
have␈α
their␈α
reaction␈α
yet␈α
to␈α
our␈α
proposed␈α
mode␈α
of␈α
use).
␈↓ ↓H

␈↓ ↓H
␈↓ α_2.␈α⊃We␈α⊃are␈α⊃now␈α⊃planning␈α⊃a␈α⊃collaborative␈α⊃project␈α⊂with␈α⊂the␈α⊂Communications␈α⊂Department␈α⊂at
␈↓ ↓H
␈↓ ↓HStanford␈α⊃which␈α⊃includes␈α⊃journalism.␈α⊃The␈α⊃idea␈α⊃is␈α⊂to␈α⊂create␈α⊂the␈α⊂␈↓↓Stanford␈α⊂Electronic␈α⊂Times␈↓␈α⊂which
␈↓ ↓H
␈↓ ↓Hwould␈αinclude␈αwire␈αservice␈αnews␈αwith␈αan␈α␈↓↓information␈αthread␈↓␈αprovided␈αby␈αan␈αeditor␈αwho␈αwould␈αbe␈αa
␈↓ ↓H
␈↓ ↓Hjournalism␈α∞student.␈α∞There␈α∞would␈α∞also␈α∞be␈α∞local␈α∞news␈α∞provided␈α∞by␈α∞news␈α∞writing␈α
classes␈α
and␈α
a␈α
some
␈↓ ↓H
␈↓ ↓Hslowly␈α⊂changing␈α⊂information␈α⊂about␈α⊂movies,␈α⊂restaurants,␈α⊂Stanford␈α⊂events␈α⊂and␈α⊂facilities,␈α∂etc.␈α∂There
␈↓ ↓H
␈↓ ↓Hwould␈α⊃also␈α⊃be␈α⊃editorials␈α⊂and␈α⊂a␈α⊂letters␈α⊂to␈α⊂the␈α⊂editor␈α⊂system␈α⊂allowing␈α⊂back␈α⊂talk␈α⊂to␈α⊂all␈α⊂stories␈α⊂and
␈↓ ↓H
␈↓ ↓Heditorials.␈α∂It␈α∂would␈α∂be␈α∂possible␈α∂to␈α∂include␈α∞all␈α∞letters␈α∞to␈α∞the␈α∞editor,␈α∞and␈α∞a␈α∞reader␈α∞would␈α∞be␈α∞able␈α∞to
␈↓ ↓H
␈↓ ↓Hrequest␈α∂notification␈α∂of␈α∂all␈α∂comments␈α∂that␈α∂were␈α∂subsequently␈α∂attached␈α∂to␈α∞an␈α∞editorial␈α∞or␈α∞story␈α∞that
␈↓ ↓H
␈↓ ↓Hinterested␈α∞him.␈α∞We␈α∞have␈α∞not␈α∞yet␈α
sent␈α
off␈α
the␈α
proposal␈α
to␈α
NSF,␈α
and␈α
it␈α
is␈α
by␈α
no␈α
means␈α
certain␈α
that
␈↓ ↓H
␈↓ ↓Hthey␈α
will␈α
fund␈α
it␈α
even␈α
if␈α
they␈α
like␈α
it,␈α
because␈α
it␈α
may␈α
not␈α
have␈α
enough␈α
science␈α
in␈α
it␈α
for␈α
them.
␈↓ ↓H

␈↓ ↓H
␈↓ α_3.␈α∞If␈α∞you␈α∞are␈α∞interested,␈α∞we␈α∞can␈α∞again␈α∞explore␈α∞storing␈α∞all␈α∞the␈α
A.P.␈α
news.␈α
The␈α
programs␈α
now
␈↓ ↓H
␈↓ ↓Hwork␈αfor␈αthis,␈αand␈αwe␈αare␈αnow␈αusing␈αthe␈αmost␈αcost-effective␈αIBM␈αdisk␈αfile␈α-␈αthe␈α3330␈αmodel␈α11.␈αMy
␈↓ ↓H
␈↓ ↓Hold␈α⊂estimate␈α⊂was␈α⊂that␈α⊂it␈α⊂would␈α⊂cost␈α⊂$750␈α⊂per␈α⊂month␈α⊂to␈α⊂store␈α⊂a␈α⊂year's␈α⊂news.␈α⊂If␈α⊂it␈α⊂were␈α⊂done,␈α∂we
␈↓ ↓H
␈↓ ↓Hprobably␈αshould␈αalso␈αget␈αsome␈αother␈αcomputer␈αto␈αsubscribe␈αso␈αthat␈αvery␈αlittle␈α
news␈α
would␈α
be␈α
lost␈α
due
␈↓ ↓H
␈↓ ↓Hto␈α
computer␈α
downtime.
␈↓ ↓H

␈↓ ↓H
␈↓ α_IBM␈αhas␈αjust␈αannounced␈αa␈αnew␈αfile␈αthat␈αis␈α
much␈α
more␈α
cost-effective␈α
for␈α
very␈α
large␈α
amounts␈α
of
␈↓ ↓H
␈↓ ↓Hinformation␈α-␈αa␈αyears's␈αA.P.␈αnews␈αcould␈αbe␈αstored␈αfor␈α$75␈αper␈αmonth␈αmarginal␈α
cost.␈α
The␈α
catch␈α
is␈α
that
␈↓ ↓H
␈↓ ↓Hthe␈α∞minimal␈α∞system␈α∞which␈α∞stores␈α∞35␈α∞billion␈α∞characters␈α∞rents␈α
for␈α
$10,000␈α
per␈α
month␈α
and␈α
requires␈α
a
␈↓ ↓H
␈↓ ↓H3330␈αin␈αaddition␈αso␈αthat␈αwe␈αcouldn't␈α
replace␈α
our␈α
present␈α
equipment.␈α
I␈α
don't␈α
suppose␈α
A.P.␈α
would␈α
want
␈↓ ↓H
␈↓ ↓Hto␈αacquire␈αsuch␈αa␈αsystem,␈αbut␈α
we␈α
are␈α
looking␈α
into␈α
whether␈α
a␈α
combination␈α
of␈α
applications␈α
would␈α
allow
␈↓ ↓H
␈↓ ↓Hus␈αto␈αsupport␈αsuch␈αa␈αsystem.␈αIf␈αwe␈αcan␈αfind␈αother␈αinterest,␈αwe␈αmay␈αask␈αif␈αyou␈αare␈α
interested.␈α
The␈α
full
␈↓ ↓H
␈↓ ↓Hsize␈α
IBM␈α
system␈α
stores␈α
472␈α
billion␈α
characters␈α
so␈α
there␈α
is␈α
lots␈α
of␈α
room␈α
for␈α
growth.␈α
As␈α
of␈αDecember,
␈↓ ↓H
␈↓ ↓Hthere␈αis␈αlittle␈αprospect␈αof␈αa␈αsuitable␈αcombination␈αto␈αget␈αthis␈α
monster␈α
file,␈α
but␈α
there␈α
may␈α
be␈α
one␈α
on␈α
the
␈↓ ↓H
␈↓ ↓HARPA␈αnet␈αthat␈αcould␈αbe␈αused.␈αIf␈αthis␈αturns␈αout␈αto␈αbe␈αtrue,␈αand␈αthe␈αtechnical␈αpossibilities␈αlook␈αgood,
␈↓ ↓H
␈↓ ↓Hand␈αyou␈αare␈αinterested,␈αI␈αwill␈αtry␈α
to␈α
convince␈α
you␈α
and␈α
Mr.␈α
Gallagher␈α
that␈α
his␈α
worry␈α
about␈α
the␈α
ARPA
␈↓ ↓H
␈↓ ↓Hnet␈α
is␈α
not␈α
warranted.
␈↓ ↓H

␈↓ ↓H
␈↓ α_What␈α∀we␈α∀have␈α∀done␈α∀with␈α∀your␈α∀money␈α∀is␈α∀to␈α∀have␈α∪Martin␈α∪Frost␈α∪continue␈α∪his␈α∪work␈α∪on
␈↓ ↓H
␈↓ ↓Himproving␈α⊂the␈α⊂news␈α∂service␈α∂program.␈α∂Besides␈α∂that,␈α∂I␈α∂hired␈α∂a␈α∂guy␈α∂to␈α∂help␈α∂work␈α∂on␈α∂ideas␈α∂for␈α∂the
␈↓ ↓H
␈↓ ↓H␈↓↓Stanford␈αElectronic␈αTimes␈↓␈αand␈αhome␈αterminals␈α
in␈α
general.␈α
Unfortunately,␈α
he␈α
didn't␈α
accomplish␈α
much,
␈↓ ↓H
␈↓ ↓Hand␈α
it␈α
took␈α
3␈α
months␈α
to␈α
find␈α
that␈α
out.
␈↓ ↓H

␈↓ ↓H
␈↓ α_If␈αAssociated␈αPress␈αis␈αin␈αa␈αposition␈αto␈αsupport␈αthe␈αwork␈αfurther,␈αMartin␈αwould␈αcontinue,␈αbut␈α
it
␈↓ ↓H
␈↓ ↓Hwould␈α∞be␈α∞very␈α∞good␈α∞if␈α∞we␈α
could␈α
put␈α
a␈α
terminal␈α
in␈α
some␈α
public␈α
place␈α
in␈α
Stanford␈α
so␈α
as␈α
to␈α
explore
␈↓ ↓H
␈↓ ↓Hfurther␈αhow␈αto␈αmake␈αit␈αuseful␈αfor␈αthe␈αinexperienced.␈αThis␈αwill␈αbe␈αdone␈αby␈αrecording␈α
the␈α
interactions
␈↓ ↓H
␈↓ ↓Hand␈α
by␈α
giving␈α
the␈α
user␈α
a␈α
chance␈α
to␈α
describe␈α
his␈α
difficulties␈α
at␈α
the␈α
end␈α
of␈α
each␈α
session.
␈↓ ↓H

␈↓ ↓H
␈↓ α_I␈αam␈αsorry␈αto␈αhave␈αbeen␈αso␈αlong␈αin␈αreporting␈α
to␈α
you,␈α
and␈α
it␈α
may␈α
be␈α
too␈α
late␈α
for␈α
you␈α
to␈α
consider
␈↓ ↓H
␈↓ ↓Hcontinuing␈αsupport␈αnext␈αyear.␈αMost␈αof␈αthis␈αletter␈αwas␈αwritten␈αin␈αOctober,␈αbut␈αsome␈αemergency␈αput␈αit
␈↓ ↓H
␈↓ ↓Hon␈α
the␈α
back␈α
burner␈α
and␈α
I␈α
didn't␈α
get␈α
back␈α
to␈α
you.
␈↓ ↓H

␈↓ ↓H

␈↓ ↓H

␈↓ εW
Sincerely yours,
␈↓ εW

␈↓ εW

␈↓ εW

␈↓ εW

␈↓ εW

␈↓ εW
John McCarthy
␈↓ εW
Director, Artificial Intelligence Laboratory
␈↓ εW
Computer Science Department
␈↓ εW

␈↓ ↓H
JMC:pw
␈↓ ↓H
bowen.le2[let,jmc]:SU-AI
␈↓ ↓H