COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	\\M0BASL30\M1BASI30\M2NGR40\M3NGR25\M4NGR20\MFSTA200\
C00011 ENDMK
C⊗;
\\M0BASL30;\M1BASI30;\M2NGR40;\M3NGR25;\M4NGR20;\MFSTA200;\;
\'3;↓↓\FFS\F2
\'3;↓Q\CSTANFORD UNIVERSITY
\F3\CSTANFORD, CALIFORNIA 94305
\F4



ARTIFICIAL INTELLIGENCE LABORATORY\←L\-R\/'7;\+R\→.\→S   Telephone:
COMPUTER SCIENCE DEPARTMENT\←S\→.415-497-4430
\F0\C2 December 1974


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

Dear Dave:

\J	Here is a writeup on the new version of the news program that includes
the \F1New York Times\F0 as well as A.P.  It also contains other improvements
that we told you we planned to make.  It can store news for an arbitrary time,
depending only on how much disk space we allocate to it, and it allows arbitrary
keyword expressions.

	The following remarks can be made:

	1. We are still not satisfied with the ease of use for inexperienced
people.  When we can afford to put a terminal in a Stanford library with
a placard explaining its use, we will get more useful data.  (Of course, we
think our system has always been easier to use than the NYT Data Bank).
Incidentally, some people from there visited us and we suggested that we
might join in a Stanford subscription and experiment with allowing our
users to query the system in a uniform format and to translate the query
to their system when the query was appropriate to their data bank.  The
visitor (I forget his name, but not John Rothman, the proprietor of the system)
was quite unenthusiastic about having their system used in any but the
absolutely standard way even though all queries would be properly paid for.

	(Now we are in the process of trying to subscribe, but we aren't sure
we can afford it, and we don't have their reaction yet to our proposed mode
of use).

	2. We are now planning a collaborative project with the Communications
Department at Stanford which includes journalism.  The idea is to create
the \F1Stanford Electronic Times\F0 which would include wire service news
with an \F1information thread\F0 provided by an editor who would be a
journalism student.  There would also be local news provided by news writing
classes and a some slowly changing information about movies, restaurants,
Stanford events and facilities, etc.  There would also be editorials and
a letters to the editor system allowing back talk to all stories and
editorials.  It would be possible to include all letters to the editor, and
a reader would be able to request notification of all comments that
were subsequently attached to an editorial or story that interested him.
We have not yet sent off the proposal to NSF, and it is by no means certain
that they will fund it even if they like it, because it may not have enough
science in it for them.

	3. If you are interested, we can again explore storing all the A.P.
news.  The programs now work for this, and we are now using the most
cost-effective IBM disk file - the 3330 model 11.  My old estimate was
that it would cost $750 per month to store a year's news.  If it were done,
we probably should also get some other computer to subscribe so that
very little news would be lost due to computer downtime.

	IBM has just announced a new file that is much more cost-effective
for very large amounts of information - a years's A.P. news could be
stored for $75 per month marginal cost.  The catch is that the minimal
system which stores 35 billion characters rents for $10,000 per month and
requires a 3330 in addition so that we couldn't replace our present
equipment.  I don't suppose A.P. would want to acquire such a system, but
we are looking into whether a combination of applications would allow
us to support such a system.  If we can find other interest, we may ask
if you are interested.  The full size IBM system stores 472 billion
characters so there is lots of room for growth.
As of December, there is little prospect of a suitable combination to get
this monster file, but there may be one on the ARPA net that could be
used.  If this turns out to be true, and the technical possibilities look
good, and you are interested, I will try to convince you and Mr. Gallagher
that his worry about the ARPA net is not warranted.

	What we have done with your money is to have Martin Frost continue
his work on improving the news service program.  Besides that, I
hired a guy to help work on ideas for the \F1Stanford Electronic
Times\F0 and home terminals in general.  Unfortunately, he didn't
accomplish much, and it took 3 months to find that out.

	If Associated Press is in a position to support the work further,
Martin would continue, but it would be very good if we could put a
terminal in some public place in Stanford so as to explore further how
to make it useful for the inexperienced.  This will be done by recording
the interactions and by giving the user a chance to describe his
difficulties at the end of each session.

	I am sorry to have been so long in reporting to you, and it may
be too late for you to consider continuing support next year. 
Most of this letter was written in October, but some emergency put it on
the back burner and I didn't get back to you.\.


\←L\→S\←R\-L\/'2;\+L\→L
Sincerely yours,





John McCarthy
Director, Artificial Intelligence Laboratory
Computer Science Department
\←S\→L
JMC:pw
bowen.le2[let,jmc]:SU-AI