Terence Goggin  <terenceg@nutshellsoftware.com> wrote:
+---------------
| I'm desperately trying to find the source of a story/experiment that
| I was told several times when I was going to school for my CS degree...
| ...divided his class into three groups... [plan, hack, mixture]
+---------------

I don't recall that exact story, but I do recall a report of a formal
experiment that was done to compare the effects of "interactive" versus
"batch" programming environments? IIRC, it was done using a fully-interactive
system, but with the compiler hacked so that different test groups were
"locked out" of the compiler for a specified minimum time after each compile.
The results were something along the lines of...  One of the "middle" groups,
the one that was locked out of the compiler for 2 hrs (???  20 minutes?
somewhere in there) after each compile, finished the task most quickly.

Groups with less "lock-out" time (including the "fully-interactive" group)
spent too much time "churning" the compiler fixing one tiny problem at a
time, while groups with a higher lock-out time didn't get enough compile/edit
cycles. The claim was that the middle groups did better because they were
forced to "back off" long enough for them to look at all (or at least most)
of their errors per cycle, and also because the forced "idleness" allowed
them time to look at bigger issues than just low-level typos.

If you want to find this reference, look in some of the old "programming
methodology" books/articles, by people like the Yourdon Group, Tom DeMarco,
Gerry Weinberg, etc., in the late 1970's and early 1980's. There was a lot
of this kind of thing done back then...


-Rob

-----
Rob Warnock, 7L-551		rpw3@sgi.com   http://reality.sgi.com/rpw3/
Silicon Graphics, Inc.		Phone: 650-933-1673
2011 N. Shoreline Blvd.		FAX: 650-933-4392
Mountain View, CA  94043	PP-ASEL-IA