Prediction Markets: My Google 20% project

We've finally made a Google Blog post about my 20% project - an internal prediction market system. My role was in analyzing its performance, I hope to be able to post more details about that later (measuring predictiveness is a little tricky). For now, check out the post, which has two of my graphs: accuracy and decisiveness.

Share this

Thanks!

Thanks!

http://www.chrisfmasse.com/3/

http://www.chrisfmasse.com/3/3/software/ has good links to open source projects and other info that would be useful to those trying to set up their own markets.

Oh, and there's more

Oh, and there's more interesting background in the Wikipedia article on Prediction markets:

http://en.wikipedia.org/wiki/Prediction_market

slashdot linked to the

slashdot linked to the google blog entry here:

http://slashdot.org/articles/05/09/22/135211.shtml?tid=217

from the slashdot comments, one thing that was brought up is owise.com , which apparently tries to use prestige and prizes as rewards for good predicitions instead of using the standard "futures market" paradigm. They claim it's easier for non-finance types to grasp.

Would it be possible to go

Would it be possible to go into the details of how the system was actually implemented? Some questions:

- Were the projects being estimate company wide or smaller, team based things? If the latter, were people who may have been in completely different parts of the company allowed to bid on those projects they might have known nothing about?

- What was the physical set up of the system? An internal website? Was there a time limit on the bidding? How much money eventually got paid out?

- Was there an internal administrative request to participate, or did people play through word of mouth?

I don't want to pry at Google secrets, but I'd like to see if something like this could be set up at my workplace. :)

Anyone in the company was

Anyone in the company was allowed to bid on any project. The outcomes were usually about specific, large projects like Gmail or Orkut. The system ran on an internal website, with markets closing at the end of the quarter. I don't think there was administrative pressure to participate, but we did publicize the system extensively.