COMMENT ⊗   VALID 00005 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	Joleen:
C00004 00003
C00014 00004	Lyall:
C00018 00005
C00019 ENDMK
C⊗;
Joleen:

Here are the  course evaluations.  After the fact,  Larry and  I did  some
detective work and  associated the forms  with some info  we gleaned  from
them the first day (contrary to Lie-all More-ill's impression of reality).

I'm also  enclosing a  rebuttal  I wrote  that I'd  like  to send  to  the
participants if that's consistent with  WICS policy. I've also enclosed  a
personal note to Morrill; I think the guy's really sick.

Both Larry and I would like to do the course again; current thinking is  a
two-week stint: one week LISP, and a second week AI techniques. The second
week being open to those who  have sufficient LISP background (we'll  mail
them an entry test!) --probably several guest shots should be in week two.

Any by the way, next year I want  a hotel room, an ARPA tip, video  taping
equipment, and ...

				Jean-Jacques Allen
				Holiness-at-large

	  An Analysis of the WICS LISP Course Evaluation Results

		     (the personal view of J. Allen)

Essentially, the  evaluations were  disappointing.   Perhaps some  of  the
issues were  cleared up  in our  last  session on  Friday, but  they  bear
repeating:

WICS courses are meant to be graduate-level, not survey courses. As  such,
we expected high-levels of performance.  Simply gaining knowledge of  "how
to program  in LISP"  is not  appropriate performance;  of course,  it  is
necessary  that  one  knows  the  rudiments  of  programming  to  continue
self-education.  That is why (at least) 3 hours per day were dedicated  to
lab time.  We set  up the course with  6 hours of lecture  and 3 hours  of
lab; that's  an  appropriate  partitioning of  mental  and  manual  labor.
Furthermore, there  were an  additional two  hours of  optional time  each
evening, and if lectures were boring no one required that you attend (more
about "boring" lectures later). Basically,  If you didn't attend the  lab,
that is your problem and your loss.

Furthermore, learning "how LISP is implemented" is interesting,  valuable,
but in my mind,  not the appropriate content  of a graduate course  titled
"LISP". Perhaps,  "Implementation of  Complex  Systems", but  not  "LISP".
More importantly, one should never try to understand a computing  language
by trying to implement it (do not learn  to drive by building a car (or  a
cdr)).

What then is the appropriate thrust?   It is:  A through grounding in  the
fundamentals with introductions  to the  applications, and  implementation
issues, with hands-on practice to exercise the fundamentals.  What are the
fundamentals? They are: style and semantics.  That is why we spent  Monday
and Tuesday going through over 100 slides on LISP subsets and their proper
application. Monday and Tuesday were all, programming style,  abstraction,
and representation.  Wednesday 's  examples from  AIP were  all about  the
application of abstraction; here in AI as discrimination nets, data bases,
facts, and theorems represented as list-structure.  Thursday's  evaluators
were examples of abstraction and representation; this time applied to LISP
itself.  And running from Monday through Friday was the continuous example
of polar arithmetic:  changes of  style, changes  of representation.   The
whole issue  of record  packages  is that  of hiding  the  representation:
that's abstraction.  Such  issues of style  and concept are  more in  line
with the content of a graduate class.  If these issues bored you, then you
should re-evaluate your goals concerning LISP work.

You should realize  that there is  much more to  learning a language  than
learning  the  syntax.   Unfortunately,   most  computing  languages   are
"infantile"; LISP is not.  Proficiency in  LISP is an issue in  semantics:
you must  learn  to "say  things  well."  That was  the  issue  underlying
Monday-Wednesday.  If you missed it, re-read your notes.

Unfortunately, the AIP book doesn't practice-what-it-preaches in the early
going. That is why we replaced the initial chapters with our own material.
If  you  believe  that  one  can  a  complete  discussion  of  LISP,  from
fundamentals  to  functionals   in  two  days   in  100+  slides   without
preparation, then I thank you for the compliment: I must have made it look
too easy. If the  issue of dynamic versus  lexical scoping was  repetitive
but incomprehensible, then then I'll prepare that better, for that is  THE
critical issue in the "dynamics" of languages (just as first-class-ness is
the issue in the "statics"). 

Returning to AIP, their later chapters on AI examples are quite good; that
is why  we  gave  only  a  few slides  on  Chapters  14-16,  those  either
highlighted the  code (reading  "the literature",  and these  slides  were
direct copies of material  in your book)  or supplied background  material
not covered in the text.  I do  not believe in "reading the book", not  to
undergraduates, and definitely  not to graduate  students.  The notes  for
Wednesday (which you had since Sunday) explicitly stated that Chapters 11,
13, and 14 were to be covered that day; it was your responsibility to read
that material before class.

To finish off  the unpleasant business,  there are two  issues. First,  my
copy of Winston's AI  book is missing; I'd  really like it back.   Second,
one evaluation was particularly irritating, being a personal rather than a
professional criticism.   It should  be clear  that such  behavior is  not
appropriate.

With all that said, I still believe that the effort was worthwhile, and do
plan to  offer the  course  again. I  think we  all  learned alot.  It  is
important to nuture the  initial seed: read the  AIP book's LISP  sections
now that  you have  some  structure on  the  language facilities  of  LISP
--applicative, imperative,  and  modification  subsets;  read  other  LISP
books. Read  the  AIP book's  applications  and  see how  the  notions  of
abstraction make their code more  readily understandable; write your  own,
and see how  abstraction can aid  in the formulative  phases too: less  to
write down,  easier to  modify, ...  Get LISP  on your  machine; write  an
evaluator in LISP and see how LISP works; if you're ambitious, try writing
a LISP system in  a low-level language and  note that the same  techniques
--abstraction (perhaps through macros) and modularization (perhaps through
data-driven dispatch  code) are  still available  to you  as  INTELLECTUAL
TOOLS. By all means  continue; there was no  way to instill all  knowledge
and insight in one week. Things take time; but if you pursue and  persist,
lights WILL go on.

If you wish to respond, please do.

John Allen
The LISP Co.
POB 487
Redwood Estates, 95044
Lyall:

I found your evaluation of the  WICS LISP course disturbing; it seemed  to
me that we attended two separate courses.

In mine, we had six hours of  lectures per day; in yours, "eight hours  of
tedious lectures."

In mine we had the participants fill out questionaires Monday,  describing
their background in LISP, programming, education, interest in and  planned
use of LISP; in yours "the instructors evidently did not know -and made no
effort to discover- what backgrounds and goals we as students had."

In mine, one participant (named  Lyall Morrill --a relation?)   explicitly
asked for a description of the λ-calculus, a theoretical area not directly
related to LISP  and not scheduled  for discussion.  I  made note of  that
request and responded to  it in a  question-and-answer session (after  all
other questions  had been  answered); in  yours, "[we]  were subjected  to
erudite monlogues on arcane theoretical issues."

In mine, 106 slides were presented in  two days, about 150 slides in  five
days, about 130 were given to the students before the course began and the
remainder were passed out  during the week; in  yours, "the lectures  were
sloppy and ill-prepared" and you were subjected to "slides which had  been
drawn at the last minute."

In mine, question-and-answer sessions were  provided during the days:   an
hour on  Tuesday and  the last  sessions (1-3/4  hours) on  Wednesday  and
Thursday;  in  yours,  the  instructors  "merely  flashed  the   solutions
monentarily on the screen during lecture."

In mine, either Larry or I was available during the lab time either in the
lab or across the hall, and the lab was open at least until midnight every
night but Wednesday; in yours, "the limited lab time was further  devalued
because so  little coaching  was  offered" and  "in  a class  this  small,
individual help could have been more freely provided."

Your perception of reality is troublesome. If you had difficulty with  the
course material, you could have  asked for help. Generally, your  comments
were  obnoxious,  non-constructive,  and  uncalled  for,  though  somewhat
amusing.  But Lyall, I tire of it. I don't think you're cut out for LISP.