COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	Prof. J. Moses, Head
C00005 ENDMK
C⊗;
Prof. J. Moses, Head
Dept. of EE and CS
38-403
M.I.T.
Cambridge, MA 02139

Dear Prof. Moses:

	I have been following Professor Robert H. Halstead's work,
because his approach to parallel execution of LISP is similar
to mine, and in many respects he is ahead of us.

	I believe that his Concert Multiprocessor and its use
for implementing his parallel LISP Multilisp is a scientific
and engineering achievement well meriting tenure.  I know
more about the language part than the machine part or the
implementation, and I consider Multilisp a well designed langauge.

	In this narrow field, his is one of three efforts in
the world, our QLISP and Ito's PAILISP at Tohoku University in
Sendai, Japan being the other two.  Halstead started implementation
first and is ahead.  In the larger field of parallel computing, I
suppose he is well respected, but I imagine this depends in part
on whether the other people agree with his approach to parallelism.

	I think his is the right approach to parallel symbolic
computation and that parallelism is required for high performance
symbolic computation.  Therefore, his work will gain in importance
and recognition.

	I don't know about his teaching, but his lectures at meetings
are entirely clear.

	His work is well presented both orally and in writing, and
this gives it a leading position.  For example, BBN chose to implement
Multilisp on their Butterfly machine even though the Butterfly
architecture is only partly suited for it.  Their new project, Monarch,
goes further in Halstead's direction.

	I don't know his former graduate students.

	He seems to have many more ideas than those involved in
Multilisp and the ability and motivation to implement them, and
therefore I expect further professional growth.

	I have no doubts about recommending him for tenure or
even a full professorship.

Sincerely,