>>It is not essential to even know how to program, because that is
>>merely a skill which is easily learned by people comfortable with
>>computers, ...
Keyuan Jiang writes:
>Developing software involves developing algorithm to the problem in question
>and implementing the algorithm. Developing algorithm is NOT a skill but an
>ART of problem solving.
I agree that developing *good* algorithms is an art, and I highly respect
those who are good at it. But it is easy to learn enough to do it poorly,
and it takes years of practice to do it well. However, this doesn't alter
my main point. A biologist writing a program for his or her own use is
under no constraint to do it well, as long as it does what is intended.
Later when hordes of other biologists are pounding on his or her doors the
biologist can hand it (and lots of money) to a computer scientist and say
"I want a program that works like this, um, 'prototype' [ ;-) ], but I want
it to work faster (or in three-dimensions or with more data or on a
different computer or whatever)."
Spreadsheets revolutionized the practice of accounting, and the algorithms
for the arithmetic in them did not need to be beautiful or fast, merely
accurate. It was the way the numbers were displayed that made spreadsheets
useful.
ON THE OTHER HAND, I have thought about my assertion that computer
scientists can't contribute much to biology without being trained in
biology, and I want to partially retract it. I have seen a computer whiz
watch something a biologist was doing and and say, "Oh, you could really
use a copy of Whatchamacallit Software to do that." And the lab was never
the same again. But they still had to understand what the biologist was
trying to do.
Ben Jones BioQUEST / Department of Biology
jonesbb at beloit.edu Beloit College, Beloit, Wisconsin
