For a non-Agile conference it was gratifying to see a standing room only crowd.. literally! People where standing along the walls and sitting in the ailse.

The Bob, Ken, & Martin each gave a brief talk and then there was a Q&A panel. The usual issues were raised:

"What about Agile & CMM?"... "No conflict" (actually, Watts Humphreys answered a very similar question yesterday with a very similar answer.

"How do I get my manager to let us pair program?"... "Just do it & show them the results. If they llike statistics, read Laurie Williams' book & use that."

"How do I move to Agile with an existing project?"... "Do new work agile, and gradually add tests to the legacy codebase".

One of the most interesting pieces of information, I thought, was some numbers that Martin & Bob quoted regarding decreased defect rates... Martin talked about a project at Chrysler that have 1 bug in a year. Bob talked about a project that went from 1 bug per 2 programmer-days down to 1 in 20 when they adopted XP.

With numbers like that who can argue?

And on the lighter side, where was a fair bit of good natured ribbing over iteration length. Ken takes the SCRUM line on this with 1 month iterations. Bob pushes 2 week, and Martin talked about Thoughtworks using 2 week for distributed teams and increasingly 1 week for colocated teams. One thing he said was that you should *not* run 1 week iterations Monday-Friday. Teams that have put iteration boundaries in the middle of the week have reported better results.

I can see the logic in this. Your minds can churn away on problems subconciously over the weekend, when otherwise you would be between problems. It would also avoid the Monday-morning & Friday-afternoon impact of iteration ramp up/down.