I recently heard, on a Podcast, Mike Cohn from Mountain Goat Software talk about Planning Poker as a means of estimating software projects. It sounded like a really interesting idea. Anyone tried it?
The idea is that everyone involved in the planning and estimating process gets a deck of cards with values on it (he uses the fibonacci numbers on the deck). Everyone decides on their own what the estimate for a piece of work should be, picks the card that corresponds to that value and puts it face down in front of them. When everyone has picked a card, they are all turned over and the group sees what the values are. Discussions are held on the discrepancies, so that people's assumptions can be examined and then the group does the same exercise again until consensus is achieved. Mike says that consensus is usually achieved after 3 hands or less and that the method is quite accurate.
If the goal of planning is to reach consensus, then this sounds like a viable approach. I've used "voting" methods in the past, rather than "game" methods, with similar intent.
I wonder about the "quite accurate" part, though. I'm not sure if repeated estimate-discuss-estimate-again sessions gain more accuracy, or just more agreement. I'll have to look for an opportunity to try this out.
On occasion, I've been involved in what seemed like "Planning Name That Tune" or even "Planning Russian Roulette". That wasn't as much fun.
I tried it and we passed through 3 stages before making it work:
1. At the beginning it took us ages to agree on the estimate for each feature.
2. After a couple of iterations we started agreeing too fast (and not always hitting the estimation-actual area).
3. Then all of a sudden we went back to not agreeing (maybe due to the fact that consensus had taken us off-course) and then we started working correctly and hitting the target a larger percent of the time.
Bottom line is that it's harder than we though, took more time to get it to work than previously planned, today each time we "sit around the table" we know that it will be a hard working meeting and not a poker game, but at the end of the day it provides good results.
One additional thing is that as we've done the process for some time now we even started speaking our own jargon during the sessions so all of a sudden when a new guy joined it was initially hard for him to join the process.