Predictify helps you understand other voters

I am not a fan of the prediction market business. I know that many game and social theorists think they're valuable predictors of crowd behavior, but I've seen too many prediction start-ups turn into intellectual wastelands, with a few people controlling the "price" of opinions that are either pointless on the face of it, or bizarrely tilted in one direction or the other.

So I took Parker Barrie's recent pitch for a new feature on Predictify with a grain of salt. On Wednesday, the company is launching a prediction market for the U.S. presidential election, called the Election Showdown. Users will be able to enter their predictions for the election outcomes in 14 battleground states, and the most accurate users will, after the election results are in, divide a pot of $100,000.

Upon hearing this, my hackles rose. Users may predict one thing but vote another, I said, especially when money enters the process. To which Barrie had an answer that rescued the concept for me: "When you ask people to predict what will happen, you separate them from their personal biases and make them consider additional information."

The Election Showdown feature is not, I'm led to believe, so much about predicting the election results as it is about making people think about predicting the election results. "We provide a forum to research, discuss, and predict what you think will happen," Barrie said. And anyway, "We don't position Predictify as precisely accurate as to what will happen." There are polls for that, I guess.

Predict a state, win some money. But mostly, learn something.

(Credit: Predictify)

Over the past year, since I first covered the company, Predictify has in fact shifted its focus from providing precise predictions to creating a platform for community engagement. That works for me. I still don't like prediction markets. But I do like sites that encourage people to think in new ways.

Other political prediction markets (thanks, Peter Norvig): Intrade, Iowa Markets, Rasmussen Markets, Betfair, Newsfutures, Inkling.

See also: The Journal of Prediction Markets.

[nigeleccles]

This is really cool. I've just made my picks. I used our electoral forecast to predict Obama to win by 334 electoral votes.

Rafe - I understand your objections to prediction markets. There have been a lot of "prediction market" start-ups that haven't gained liquidity and therefore their forecasts are meaningless. However that isn't the case with the ones you linked to above. If you take the presidential contract the markets you mention above, our own market Hubdub and the UK betting markets, the prices have moved in sync. Millions of dollars are being traded on the UK betting markets and all the research has shown them to be accurate.

Here is our current market and links to the other major exchanges: http://www.hubdub.com/m17795/Who_will_win_the_2008_US_Presidential_Election

[amsiegel]

There is a fundamental contradiction in this article. First a claim is made that prediction markets don't work because only a few people participate or the prices are "bizarrely tilted" one way or another. This is true, and one could make the same claim of any social network, news aggregator, etc. If you have too few people participating, you get bias.

However, the article goes on to say that what changed the author's mind is hearing that Predictify is based on what people think is going to happen, not what they want to happen. But that's not unique to Predictify - that's a fact about any prediction market that the author just got done disregarding. Because the goal is to be right, traders behave differently. We see this in our corporate marketplaces all the time: people trading against their own projects, "field" employees being pessimistic about risks and milestones when the "official line" is optimism or vice versa.

Prediction markets work in many contexts, not just politics. They just need the appropriate care and feeding as does any crowd driven application.