COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	\\M0BDR25\M1BDI25\M2NGR30\M3XMAS25\M4NGB25\.
C00008 ENDMK
C⊗;
\\M0BDR25;\M1BDI25;\M2NGR30;\M3XMAS25;\M4NGB25;\.
\F2\CARTIFICIAL INTELLIGENCE LABORATORY
\CCOMPUTER SCIENCE DEPARTMENT
\CSTANFORD UNIVERSITY
\CSTANFORD, CALIFORNIA 94305
\F0

\C24 April 1973




Mr. Richard McQuillin, President
Composition Technology Incorporated
639 Massachusetts Avenue
Cambridge, Massachusetts 02139

Dear Dick:

\J	Here are some facts about the situation we are in.  About two years ago,
Larry Tesler, in our laboratory, developed a system for transforming a \F1manuscript\F0
into a \F1document\F0.  This system is embodied in a program called \F3PUB\F0
which has been used extensively here and at Carnegie-Mellon University for producing
reports.  Carnegie-Mellon even used it for producing a book by Bell and Newell
on register transfer modules that was published by D.E.C.  \F3PUB\F0 includes
a limited facility for producing mathematical formulas, but it is oriented towards
the line printer.  I believe Lowell has a manual for \F3PUB\F0.

	We have recently acquired a Xerox \F4XGP\F0 printer and are interested in
making good use of its ability to print arbitrary character sets and so have decided
to make a complete revision of \F3PUB\F0 in collaboration with Carnegie-Mellon and
Xerox where Tesler now works.  This revision will include a great improvement in
its ability to produce mathematical formulas, but the improvement will be limited
by the effort we are willing to invest which is approximately Brian Harvey's time
between now and when he starts school in September.  Other people will devote
similar efforts to improving the other aspects of \F3PUB\F0.  In the unlikely
event that there is a legal obstacle to Brian's working on mathematical formulas,
he will work on some other aspect of \F3PUB\F0, and someone else will do the
formulas.

	Let me state our goals.

	1. The system is to be optimized for an author working directly at a 
display terminal, preparing his manuscript using our text editor, and getting
drafts of the document on the \F4XGP\F0 printer.  The input must therefore be
something the author is willing to write himself so we will probably use something
like \F1ALGOL\F0 as input notation for formulas.  Remember that our authors are
programmers.

	2. The final output will be either also on the \F4XGP\F0 printer or will
be on something like the FR-80 having similar ability to accept arbitrary
characters and drawings.

	3. My goal as to quality is to exceed that of mathematics prepared on
a typewriter such as the books of the Benjamin Company, and I am quite sure
we can do that with a little effort.  We do not plan to put a large effort
into attaining book quality mathematics.

	Brian thinks that we would be able to use very little of your code even
if we had it, because of our differing front end and tail end and the limited
goals we have set.

	If there were to be a substantial project, ARPA would have to fund it
separately, because all the investment we plan to put in is the six man months
of Brian's time.  The other work will be done by Xerox and Carnegie-Mellon.
I am sending a copy of your letter and this letter and the preliminary plans
for the project to Larry Roberts in case something strikes him.

	In principle, I would very much like to collaborate, but I can't think
of a way to do so at present.  However, I don't see that our efforts are
likely to cut into your business until display terminals become generally used
by mathematicians.  Then you will have to make a product suited to that situation,
and you may find our programs of use.\.

							Sincerely yours,



							John McCarthy
							Professor of Computer Science
							Director, Artificial Intelligence Laboratory

cc. Dr. L. G. Roberts