If you visit the Yahoo! Research Labs / O’Reilly Buzz Game site today you’ll find the following notice where last time there was (and later today there will again be) a market:
AUCTIONEER, n. The man who proclaims with a hammer that he has picked a pocket with his tongue. — Ambrose Bierce, The Devil’s Dictionary
Here’s the scoop: We discovered that our dynamic pari-mutuel market (DPM) allowed players to exploit a market when a player is permitted to simultaneously invest in competing instruments. As a result, you will now be required to sell all of your holdings in a given market before you are allowed to purchase a competing instrument. As an example, if you hold “IE” but desire to purchase “Firefox”, you must first sell all shares of “IE”. One of our goals for this R&D project was to see how the DPM behaved in the wild. A note of thanks to the alpha geeks who have helped us pound on this research. You’ve helped validate the motto: Release early, and release often. That’s the spirit of Open Source, that’s the spirit of R&D, and that’s the spirit of Yahoo! Research Labs. We’ll be resuming the game tomorrow (3/29/05) at 3pm PT (with portfolios set back to values prior to our findings on Saturday).
Here’s a rather balanced take on the outage:
Looking at it one way, it’s an encouraging sign of the accuracy of the Tech Buzz simulation if someone really has successfully hacked it to improve their standing, or to make their company, brand or technology look better in this very public forum. But if your glass is half-empty, then the Yahoo! Tech Buzz game was first hacked only two weeks after it launched, and will be hacked again and again, distorting the underlying opinion market and the final results until either Doctor Flake packs up his bat and ball and goes home, or the winning technologies/brands/memes are declared as “Hacker networks” and “Linux”.
There are going to be growing pains in any such attempt–particularly so with the audience we chose as our starting point and subject matter. While we could have chosen something far less interesting (think “sweeps week”) and gleaned safety in sheer lack of self-interest, the manipulation suggests we’re on to something useful, even if by being useful in the way it is played it ends up being less useful in what it manages to predict.