COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	.require "let.pub" source
C00006 ENDMK
C⊗;
.require "let.pub" source
∂AIL Professor Jack Schwartz↓Courant Institute
↓251 Mercer Street↓New York, N.Y.  10012∞

Dear Jack:

	Enclosed finally is the ancient memorandum I referred to.  I
couldn't find it in my own files, which were disrupted by the move
of the AI Lab, but fortunately Don Knuth was able to retrieve a copy.
He had also written to Strachey, asking about the relation between
his 1959 ICIP paper and multi-terminal time sharing, and he sent
me a copy of the reply, a copy of which is enclosed.  In order not
to lose the memo again, I have had it typed into a file.  Also the
printout is more legible than a xerox of a ditto.

	Here are a few comments on the memo.

	1. Someone thought the date should be January 1960, but I
forget why.  It was certainly shortly after it became known that
IBM would produce a transistorized version of the 709 but probably
before its naming as the 7090.

	2. I don't know why I said the ideas weren't especially new,
except possibly that I had been advocating them since I came to M.I.T.
in 1957.  Also I might have interpreted the presence of boundary registers
in the STRETCH architecture as evidence of an IBM time-sharing project.

	3. As Strachey's letter mentioned, the word time-sharing
was "in the air", and I think I must have misidentified what I meant
by time-sharing with what became known as multi-programming.  This
is why I complained about the lack of a comprehensive treatment
rather than claiming novelty.  I was probably trying to avoid
claims of novelty in order to try to persuade Morse to commit the
Computation Center to time-sharing.

	4. I didn't believe that boundary registers could be added
to an existing machine design without slowing it more than IBM was
likely to agree to.  This proved mistaken, and when we asked IBM
for modifications in support of time-sharing, they had no objections
to Teager's proposed form of boundary registers.

	5. My memo also doesn't mention inhibition of input-output
instructions in user mode.  I forget when this was identified as
part of the required package, but STRETCH didn't have it, although
it did have boundary registers.

	It occurs to me that there may be an M.I.T. proposal to
NSF for time-sharing research submitted about 1960.  Perhaps
the NSF people could find it.  The people involved would have
been Morse and/or Teager and/or Dean Arden and/or me.

.reg

cc: Jack Blackburn