The ExtremeTuesdayClub ran an ExtremeHour in London with about 20 people from a Java developers group.
We split them into 3 smaller groups of 6-8 with a minimum of 2 developers, 2 stakeholders, and 2 QA. Everyone seemed to have a good time and to be a bit fired up by it. Of course we ran over (hadn't worked out our ProjectVelocity for the exercise itself), so we compressed the 2nd iteration, but we think we got the point across.
Notes:
- we changed the example to an automatic tea maker (we're Brits after all). This seems to be a nice balance between the mousetrap (infinite possibilities) and the vacuum cleaner (lots of technology).
- we emphasised the storyboard aspect of the drawings, that makes them easier to estimate in terms of the number of expected steps.
- our plan was to finish by having the stakeholders present their product to the other groups. This would focus their minds and reflects the real world. Didn't make it this time, although we did quickly show a couple of solutions. To do this, we had developers draw on overheads, which might make refactoring and reuse easier (draw components on different slides and overlay them to combine).
- we gave a quick 15 minute introduction to XP, but some people didn't know anything about it before, so there was quite a range. It would have been easier with more time to explain the principals.
- For example, some stakeholders had a hard time pitching the level of a UserStory between filling in technical details and saying "I don't care where the water comes from, it should just work." Hmmm, sounds like a good example...
- The move away from LoadFactor to ProjectVelocity seems to work. My team took to the idea that they had X ideal minutes to play with in the next round, rather than doing the arithmetic.
- could you expand on this, please. The people involved found it very easy to grasp the idea that "We completed X minutes' worth of tasks in the last iteration, so let's budget X for the next iteration". It was very simple and intuitive. When running this exercise, we now let people assume that ideal minutes are real in the first iteration and let them discover that they aren't.
- Any other comments, guys?