Copyright (C) 1980 Byte Magazine
Excerpted from Jerry Pournelle's article in the December 80 Byte

   	"LISP is, however, a peculiar language. It was written in 
the 1950's by Dr. John McCarthy, now Director of the
Stanford Artificial Intelligence Laboratories (SAIL), nad
it's extensively used at Standford and MIT (where McCarthy wrote it).
	"LISP does bit-by-bit arithmetic, meaing that there is 
no theoretical limit to the precision you can obtain; if you
want an exact numerical expansion of, say, 2 to the 55th 
power, or 87 factorial, you can get it from LISP, and with
only about three lines of code for a program -- and you'll
get the answer faster than you think.  LISP is one of the 
fastest languages I know of, often approaching assembly-
language programs in speed of operation.
	"LISP programs are very tight;  it's almost impossible to
write unstructured code in LISP.  It's also very nearly im-
possible to understand a LISP program, even if you wrote
it; at least that's been my experience.  You can strain like a 
gearbox and produce code that runs, and which you
understand just at that moment;  but hours later it's gib-
berish.  The only thing less comprehensible than a LISP
program is one in APL -- APL doesn't even use normal 
letters, but instead requires a special keyboard that can
generate strangley bent arrows and other weird symbols.
Both LISP and APL programmers delight in writing a
whole page of instructions into one line (and you can do 
it, too, because both languages allow functions to call 
themselves).  They also like to baffle fellow professionals
by showing a line of code and challenging anyone to say
what it does.
	"It's very hard to comment a LISP program -- but that's 
all right, because it isn't traditional for LISP program-
mers to comment their programs anyway.
	"In other words, I am not a wild enthusiast for LISP as a 
"standard" microcomputer language.   It's true that one or
another LISP variant is used by just about everyone in
the artificial intelligence field; for certain purposes there's
nothing better.  But for general-purpose programming,
LISP and APL are, in my judgement, simply too obscure."
	
 . . .

	"The best way to learn LISP is to attend Stanford or
MIT and get tutorial instruction from someone already
proficient.  The next best way is to get access to the MIT
Macsyma Consortium computer and run the TEACH-
LISP programs.  There are also a couple of MIT docu-
ments which are pretty good introductions.  I wish I knew
of a good commercial textbook, but I don't.  If you want
to learn LISP, you've no choice but to play about with it;
..."




[user-accounts people -- is this enough evidence to deny him
access to our machines?  Obviously he isn't learning anything
from them.]

Well, Jerry, you can expect uninhibited replies from the M.I.T.
people to uninhibited remarks.  LISP is well taught at many other
places than Stanford and M.I.T. - CMU, Texas, Yale, and Indiana
come immediately to mind.  Winston's Addison-Wesley text is ok,
and I hope mine with Carolyn Talcott will be better, although it
will emphasize proving programs correct, which requires some
mathematical logic.