henry@utzoo.UUCP (Henry Spencer) (10/19/85)
Whenever I hear AI people or dataflow enthusiasts or functional-programming
converts sounding off about how *of course* their languages run slowly on
existing machines, because we all know that Von Neumann machines are crocks
that will be utterly and completely obsolete once the AI/dataflow/FP machines
really get going (that is to say Real Soon Now), I remember a comment made by
a perceptive friend of mine, Hugh Redelmeier. He was doing work related to
functional programming, and was giving a seminar on it. When he started
discussing how his latest ideas could be implemented efficiently on
conventional machines (you know, the ones that work and can be purchased),
another friend from the audience asked him why he cared. His answer:
"Having two revolutions simultaneously isn't easy."
[Caveat: Hugh is a gentle soul and is not responsible for the, uh, skeptical
tone of the above. That's mine. Heh heh.]
--
Henry Spencer @ U of Toronto Zoology
{allegra,ihnp4,linus,decvax}!utzoo!henry