September 10, 2005
The Cutter Technology Council has weighed in with their opinion on Extreme Programming (XP). The “Assertion” is that XP has “come of age”. I recall that, during the initial years, many sniffed at it with only mild curiosity. A common phrase was that the approach was “good for ‘pond-bridges’, but maybe not for ‘large-scale suspension structures.'” Most SEI-centric organizations still saw moving up the CMM scale as the way to reduce cycle time, costs, and defects. You can see a PDF of the council opinion at www.qsma.com.
The Council is like an IT Supreme Court, with opinions of concurrence with an assertion, or dissent. They are among the leading thinkers of our industry.
Therefore I was impressed with the fact that all of the members cast their opinion on the “concurrence” side of the assertion. Fascinating that Josh Kerievsky, contributor along with myself on this opinion, gave a “partial conurrence.” That can be read as a partial dissent. This is notable since Joshua is the creator of “Industrial XP”, that is, XP for large-scale applications.
As far as the metrics are concerned, my findings of “before versus after XP” on time, cost, and defects at a major medical instrument company were significant: 25 percent improvement in speed, with a four-fold reduction in defects. Time will tell if this is repeatable, as we get more data.
But what really got my attention was that defects fell so dramatically in spite of relatively large sized teams. Normally this doesn’t happen. More people on a project always results in more defects. It didn’t happen here. I think it’s because of co-located client/developer teams along with paired programming. I’d be curious to hear what other people think as more data comes in…
As I said in the text of the opinion, “stay tuned…”