Wednesday, July 13, 2005

Using Markets to Predict Milestone and Release Dates? I wonder if internal prediction markets (http://commerce.net/publications/CN-TR-05-02.pdf) could be used on a project to more accurately determine project milestone and release dates? There are markets for other events, like who will win the presidental election, it seems project dates could be better predicted by some sort of market gestalt derived from everyone involved on a project. We know large project deadlines are pretty much a joke. How do you get more realistic numbers? Can you imagine Microsoft setting up an internal stock market in which people can buy and sell shares on Longhorn milestone and release dates? I bet the result would be pretty accurate. Around release times I imagine a flury of buying and selling as people make their bets on what will happen. How would feature reduction/time boxing fit into the market? People would have to make a bet on if they thought the schedule dates mattered more than features or if features mattered more than dates. Perhaps the internal calculations of all the people would predict what management would do. But the market wouldn't predict what management should do, it would predict what people thought management would do, so using market information to set policy might not be a good idea. That would be the ideal though, figuring out a way to get group input on the most likely release dates. I think it would be interesting anyway. Usually all the scheduling negotiations happen elsewhere and developers don't get much input. A market mechanism would give everyone some input and a way to make a very complex aggregate calculation. comment[] 10:22:07 PM

© Copyright 2006 todd hoff. Last update: 7/11/2006; 12:47:39 PM.