why eval is evil

["Andrew L. M. Shalit" <cambridge.apple.com!alms@bloom-beacon.mit.edu>]

OK, we all know eval is evil (notice that I didn't cross-post to
comp.lang.lisp, where probably only about 90% of the people know
eval is evil).

I'm looking for a paper which explains *why* eval is evil, and explains
how things like closures and backquote and such can be used instead.
I'm trying to re-educate some Lisp programmers who were raised in
the 1970's (some on InterLisp) and think that

         <the power of Lisp> = <the power of eval>

Has anyone ever taken the time to write this stuff down?  Or is it
just 'general culture'?

Send responses to alms@cambridge.apple.com.  I'll post a summary if
there's interest.

  thanks!

    -andrew