extreme programming?

[Mike Coleman]

david nicol <whatever at davidnicol.com> writes:
> Mike Coleman wrote:
> > pair programming strikes me as a bit nutty.
> 
> I disagree.  It strikes me as a sound cat-herding management practice:
> Not only do you get the programmers focussed on their work better,
> but each in the pair becomes the other's drop-dead replacement.

The "focus" aspect makes it sound a lot like sitting with a baby-sitter or
micro-manager all day long.  I doubt I'd be happy with this.

I actually did a little ad hoc pair programming in a previous job.  It was the
most horrible software development experience of my career, so that may be
coloring my judgement some.  It was kind of a "we don't need no stinking
design or forethought, let's just start writing code" experience.  Awful.

The idea that pairs somehow serve as a replacement for documentation seems
bogus.  I'm generally in favor of self-documenting code and minimal
documentation, but when something needs to be documented (e.g., that heavy
string-matching algorithm) it *needs* to be, and spare pecans are not a
substitute.

> If you can afford it, of course.

The simple cost doubling due to pairs is obvious, though this is disputed by
XP fans.  Don't forget a more subtle cost, though, which is that many talented
people simply won't come work for you if you require XP.  Smaller supply means
higher cost.

I do believe this, and software development in general, is worthy of
scientific study.  This needs to be done in an objective way, though.  Studies
done by XP partisans carry little weight.

--Mike