COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	\\M0BASL30\M1BASI30\M2NGR40\M3NGR25\M4NGR20\MFSTA200\M5ngr30\.
C00009 ENDMK
C⊗;
\\M0BASL30;\M1BASI30;\M2NGR40;\M3NGR25;\M4NGR20;\MFSTA200;\M5ngr30;\.
\'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\C25 November 1974


Mr. John Gabor
The Information Bank
The New York Times
229 West 43 Street
New York, New York 10036

Dear Mr. Gabor:

\J	Our eventual scientific goal is to make a system which will
answer questions by going automatically to appropriate sources on the
basis of its knowledge of what information these sources posess and how
to get it.  These sources will include information in our own computer
and also information in others.  As a step toward that we plan to develop
a uniform question asking language and use programs that reformulate the
queries in the language of particular data bases.  The utility of this
system will be compared to direct use of the data bases in which the
user has to know the language of the particular data base.

	Our intended use of the New York Times Data Bank is as follows:

	1. Our computer will emulate one of the display terminals and
will look like such a terminal to your system.  The HAZELTINE looks
appropriate at this time.

	2. A user at one of the 64 display terminals connected to our
system will formulate inquiries in one of several ways: The initial way
offered will be a direct connection through the computer's telephone
dial-out facility, and the user will engage in a dialog with your
information bank in your usual way.  The second way is that he will
formulate his query, and a program will carry out the dialog with
the information bank.

	3. The user who receives an abstract on his screen will be able
to read it and print it using the computer printer that printed this
letter.

	4. \F5Abstracts transmitted will not be put into any
secondary data bank in our own system.  If another user makes a query
that would retrieve the same abstract a new inquiry to the New York
Times Data Bank will be made.\F0

	5. The data will not be manipulated since the user will receive
only complete abstracts.

	6. We would like an individual user to be able to put a copy of
an abstract as a file in his disk area, since this is our general way
of storing personal information.  For example, my copy of this letter is
stored as a file in my personal disk area.  There would be no index
to files stored individually, so secondary information retrieval will
not be possible.  Of course, a user can print a copy of a file
for someone else just as a piece of paper can be xeroxed.
\F5We will reluctantly forego this if you insist.\F0

	7. We will subscribe on a transactional basis.  The access codes
will be stored in our computer system and will not be available to our
users who have passwords and identifications for logging in to our
computer.

	8. Besides the terminals in our laboratory, we also have one
in the Computer Science Library.  The librarians in the main library
and the undergraduate library might also want access through our system.
This would depend on an arrangement whereby they would pay for their accesses.

	9. While this usage is somewhat different from that of your other
customers, I believe it is in compliance with your \F1General Conditions\F0,
and gives you the same revenue per user look as you get from other users.

	Let me again emphasize the experimental nature of this work.
Should we wish to use the Information Bank in a different way from that
outlined above, we will tell you what we propose to do.

	I hope this information is sufficient to answer your questions,
but if you need more information, I will be glad to supply it to the
extent I can - i.e. our plans are not in a precise state.\.

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





John McCarthy
Director, Artificial Intelligence Laboratory
Computer Science Department
\←S\→L

P.S. We will not be ready to start until we are able to imitate the HAZELTINE.

JMC:pw
newyor.le1[let,jmc]:SU-AI