that (based on one data point) a group of
computer scientists and software professionalscan produce poetry according to spec and on time,
but not software.(It must be admitted that the specifications for
neither the poetry nor the software said anythingabout quality.)
The biggest difference that I saw between the processes
was that the process used for the sestina made effectiveuse of parallelism, and the decision to use parallelism
was made quickly and clearly; but the process for the softwaredevolved into a discussion about strategy that never converged,
followed by breaking up into independent teamseach of which attempted to solve the entire problem.
I also noticed that the initial group discussion of the
sestina project focused on structure (choosing six end wordswith little regard for their meanings, then quickly
producing a structural decomposition into stanzas that werethen addressed by independent teams), while the initial group
discussion of the software project focused on metaphor(should heat transfer be modelled as radiation or
convection?). This was exactly the opposite of whatI might have predicted; I would have expected a software
project to focus more on structure and a poetry project onmetaphor!
During the discussion of the "anthropologists' reports"
after TheExercises, I threw together this Triolet andread it to the group:
We hackers can crank out a verse
In thirty minutes flat--Provided that we keep it terse,
We hackers can crank out a verse.Our software isn't any worse:
We pull it from a hat!We hackers can crank out a verse
In thirty minutes flat.