COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	Well, this is certainly unplesant to write, but should have been done
C00016 ENDMK
C⊗;
Well, this is certainly unplesant to write, but should have been done
long ago.   It is my fault for not  being more persistent sooner. But
if we are to  work together we had better  get things straight.   I'm
tired of this annual snit; so are you; and I know Paulette is. 

This has been going  on since 1969 and perhaps stems  from me being a
student of yours.  The paternalism required for graduate students has
not changed.  When I came  up here I expect to work "with"  not "for"
you. That is how my letter of appointment was worded and that is what
I expected to do. It has not occurred. 

There are at least two possible reasons for this development.  Either
you do not believe that I am capable of independent judgements or the
ideas which interest  me are too far afield from those which you wish
to pursue. If  either of those  conditions pertain  then there is  no
point in continuing. 

This current stalemate  is not just a result  of "how's whileassem?".
It  has been  developing  for a  long time,  starting from  our joint
paper, through UCLA and continuing up here. 

The current  difficulties run through  a long  string of rather  dull
programming  and unkept promises  that things  will get  better soon.
Programming can  be a  very  destructive art.    Unless the  work  is
interesting, it is deadly; there is no middle  ground.  My first year
on theorem-proving was interesting we we building a new system; and I
did not object. At UCLA things started getting bogged down.  You were
more interested in  working with Steve and JRB; I  was more interested
in teaching. Round 1: Dec 1970. 

The next year at UCLA, we had the same problems. I felt you were much
more interested in other things (that kind of feeling is infectious). 
For me the  two years an UCLA were not a  waste. I developed the LISP
notes which I believe is still by far the best approach to the field.
Round 2: phone-call 1971. 

I came to  Stanford with the  understanding that we would  freeze the
prover and  that in the Fall I could work  on the LISP notes. Neither
happened instead of freezing it,  more was requested. The LISP  notes
sat untouched for a year and a  half, till the following Fall.  Round
3:Oct 1972. 

That intervenining  summer I spent going through JRB's stuff.  It was
a total mess and I  expected to begin a new system from  scratch that
fall.  We even had  meetings with JJM to discuss the structure of the
new language.  I looking  into Conniver as a possible  implementation
language.  Instead, you said patch the old one.  Programming help was
promised again; none arrived. Proposal for language lab was "begun". 
Round 4:Oct 1973. 

I asked the secretaries to type in LISP notes when I started teaching
JJM about LISP. Three months later, two pages were in the machine.  I
did them myself from then on and began modification. I wrote McKeeman
and volunteered to teach one  week.  I was trying to interest  you in
some of  my ideas on abstract  data structures.  I was  trying to run
experiments on unification,  debug JRB's crap  for unify and  pumping
problems.  April 1974:Round 5. 


This fall  it started again;  a new proposal  for language lab.   You
asked for  input which I gave  and as far as I  can see, ignored. The
proposal seems to be simply more of what Nori has done. I can  see my
next job:  make a few  changes to Nori's program.   That just  is not
acceptable to me. 

Besides this  difficult to understand one another, there appear to be
stong differences about the  substance of research.  Clearly  the two
areas are related, but not identical.  My feeling is that the goal of
automatic programming and verification  is to be  a "real" system  to
help "real" programmers write "real" programs.   I'm sure that's what
Congress expects.   I feel that Jack's stuff and approaches like that
are toys; similarly verification  techniques are missing many of  the
essential ingredients. 
It's  not that they're just  first steps, I feel  that they are wrong
steps.  My intuition says  that they are similar to  the developments
of Mechanical Translation, or  application of TP to robotology. There
are large conceptual flaws in the approaches.  The argument the there
will be "spin-off" to other areas just doesn't wash. 

I attempted  to  percipitate discussion  of these  points by  writing
PLL.PRO. It didn't work. I want to pursue a system based on the ideas
in that outline. If I can't do it here, I'll find somewhere else. 

So the difficulties stem  from at least two areas.   One I've got  to
learn to  talk (and  yell if  necessary) and you've  got to  learn to
listen.   This of course  is assuming that we still  have a basis for
communication.  Second, part of my yelling has got to become reality.
I'm very tired of working on projects which do not interest me. 

I am tired  of going outside of my job  to find things which interest
me and in which  I can take  pride. That is  a symptom of  something
being very much  out of joint.   I go to  Santa Cruz and San  Jose to
teach partly because I enjoy teaching but in large part because I get
a feeling of accomplishing something  which is using my talents.   My
background and training is much more than just programming and a want
to be able to exercise that experience. 

I  want to publish the book on LISP  for many reaons; it will help me
establish a name which I must do since I don't have a degree; it is a
good book and will get better before publishing; but mostly I want to
publish I because I believe strongly  that it's the best approach  to
introductory C.S. Someone else has published LISP books but they miss
too  of the important  ideas.  People  in SIGACT  are finally talking
about the ideas  in pedagogy which  I was espousing  at UCLA 5  years
ago.  A recent IBM Reserach Report did the same. The student memebers
of the  curriculum committe at Stanford are finally attacking the use
of MIX for data structures. Very soon someone will also think of LISP
or a LISP-like language as the obvious tool for data structures. 

Your comment on my LISP efforts have not been at encouraging; perhaps
that's to be expected since  you see it as  simply a dillution of  my
programming time. That should have been clear when I came here.  This
fall  was no  improvement. Publishers  want to know  when it  will be
finished I've told them June. I taught at Santa Cruz to work out some
of the bugs and I'll teach a San Jose to clear some lower level bugs.
Now as I understand a Research appointment, some of the duties are to
be teaching and  research. Granted I go  to other institutions to  do
the teaching; I played the  politics game last spring and this summer
to teach CS206, but to no avail. I know damn well I can teach as well
as the local talent. 

I have perhaps an overdose of pride, but I  take little pride in what
JRB did.  (You may chaulk this off to jealosy since he got his degree
but I  didn't.   That  might be  valid,  but I  doubt it.)  There  is
something wrong with the approach. 

What would interest me  here?  That's easy. A full  scale attack on a
system to  help construct correct programs.  Many of the ideas are in
PLL.PRO. I saw no influence of it in the presentation to ARPA. As far
as  I could your  mind's made  up: a verification  system based  on a
Pascal compiler.  We  were supposed to begin  such a proposal a  year
ago and we were supposed  to begin meeting last April on  my ideas on
program construction (after the last blow-up).They didn't happen. 

I firmly believe that  there is a market for such a system, and it is
quite realizable in three years. 


I do not want to be protected; if my ideas are crack-pot then  I want
to be condemed on that basis. If  McCarthy is after my scalp them you
should have told me. 


Where do  we go from here? Well I feel that there is room for a solid
proposal  to  NSF or  whatever.  But  it  must  reflect  some  of  my
interestes.  If we can't do that then I must try somewhere else. 

Well  I am  emotionally exhausted  from this  business; I'm  tired of
fighting and bickering. If you have read this far then perhaps we can
work something out. I will not be in  this afternoon, and perhaps not
tomorrow. I want to think about just what is going on.