COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	\\M0BASL30\M1BASI30\M2BASB30\MENGR40\M3NGR25\M4NGR20\MFSTA200\
C00010 ENDMK
C⊗;
\\M0BASL30;\M1BASI30;\M2BASB30;\MENGR40;\M3NGR25;\M4NGR20;\MFSTA200;\;
\'3;↓↓\FFS\FE
\'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\C15 January 1975



Professor Andrei Ershov
Computing Center
Novosibirsk 630090
U.S.S.R.

Dear Andrei:

\J	I had been looking forward to seeing you at the conference
in Los Angeles in April.  Unfortunately for that, though fortunately
for other reasons, I will be in Japan at Kyoto University between
March 25 and June 30.  However, I hope you will be able to visit
Stanford anyway, and if your other friends here haven't already
undertaken to arrange it, I will be glad to help.

	While the Foonly project collapsed after the design was
completed, it had the good result that Digital Equipment Corporation
used a number of the ideas from it, and in appreciation have
agreed to give us a KL-10 processor which we shall receive towards
the end of next year.  Besides that we are getting 256K more
memory, keeping the KA-10 processor, and building a mapping
multiplexor to connect i-o devices to memory and to map
the i-o devices into memory in the paging environment of the
PDP-10.

	I have been working quite steadily on expressing the
Scott logic of continuous functions within first order logic
in a form that can be used with our first order logic proof checker
FOL.  I think I have solved all the conceptual problems and
will have my first proofs in a few weeks.  Besides this, I
am working on a general concept of pattern and am thinking
that research in pattern verification is more cogent at
present than pattern recognition.  When you come to Stanford,
you can talk to Richard Weyhrauch, Michael Gordon, and possibly
Arthur Thomas about these ideas.

	I trust that Vadim Kotov's visit was fruitful for him;
we enjoyed having him.

	Anyway, I hope to see you in Tblisi in August and
maybe in Novosibirsk too, but I am not yet ready to solicit
an invitation.\.

\←L\→S\←R\-L\/'2;\+L\→L
Best Regards,





John McCarthy
Director, Artificial Intelligence Laboratory
Professor of Computer Science
\←S\→L
JMC:pw
andrei.le4[let,jmc]:SU-AI

\JP.S. You may be interested in a new file that IBM announced
a few months ago.  It consists of wide spools of magnetic tape
with a mechanical tape handler.  It can access up to 472 billion
bytes in about a second and is intended to be used as a
virtual memory for an IBM 3330, i.e. to make the 3330 appear
to the computer as though it were much larger.  IBM already
has orders for more than 500 of them.  This makes putting
large libraries on line really feasible since storing a book
would have a capital cost of less than $5.00 which is less
than keeping it on a library shelf, counting the building.
The real problem now is getting the books into the computer.

	We can't afford to buy a 3850 by ourselves, because
the cheapest version storing a mere 35 billion bytes costs
about a million.  Therefore, we are trying to form a co-op
of local PDP-10s to rent one jointly and connect it to all
our computers.  It is too soon to say whether this is
technically or administratively feasible.\.

\←L\→S\←R\-L\/'2;\+L\→L
Best Regards,





John McCarthy
Director, Artificial Intelligence Laboratory
Professor of Computer Science
\←S\→L
JMC:pw
andrei.le4[let,jmc]:SU-AI

\JP.S. You may be interested in a new file that IBM announced
a few months ago.  It consists of wide spools of magnetic tape
with a mechanical tape handler.  It can access up to 472 billion
bytes in about a second and is intended to be used as a
virtual memory for an IBM 3330, i.e. to make the 3330 appear
to the computer as though it were much larger.  IBM already
has orders for more than 500 of them.  This makes putting
large libraries on line really feasible since storing a book
would have a capital cost of less than $5.00 which is less
than keeping it on a library shelf, counting the building.
The real problem now is getting the books into the computer.

	We can't afford to buy a 3850 by ourselves, because
the cheapest version storing a mere 35 billion bytes costs
about a million.  Therefore, we are trying to form a co-op
of local PDP-10s to rent one jointly and connect it to all
our computers.  It is too soon to say whether this is
technically or administratively feasible.\.