Last week, I wrote some about prediction markets at Desirable Roasted Coffee. These are markets where hundreds, even thousands, of participants, each armed with “some” knowledge, pool their thinking to make better predictions than pollsters, better decisions than “experts”.
(Note: Wikipedia’s Prediction Market article is a good starting place if you want to learn more, as is James Surowiecki’s The Wisdom of Crowds.
But what applications do prediction markets have for business and PR practitioners? The evidence is thin to date, but last week, Google announced it’s using prediction markets to make better internal decisions:
At Google, we’re constantly trying to find new ways to organize the world’s information, including information relevant to our business. Building on the ideas of Friedrich Hayek and the Iowa Electronic Markets, a few Googlers (Doug Banks, Patri Friedman, Ilya Kirnos, Piaw Na and me, with some help from Hal Varian), set up a predictive market system inside the company.
The markets were designed to forecast product launch dates, new office openings, and many other things of strategic importance to Google. So far, more than a thousand Googlers have bid on 146 events in 43 different subject areas (no payment is required to play).
We designed the market so that the price of an event should, in theory, reflect a consensus probability that the event will occur. To determine accuracy of the market, we looked at the connection between prices of events and the frequency with which they actually occurred. If prices are correct, events priced at 10 cents should occur about 10 percent of the time. (Read more)
Google claims the prediction market is working: prices quickly reflect what’s likely, and entropy declines significantly over time. Just as you would expect in a functioning market.
The next step (and Google doesn’t say if they have or will take it) is to use prediction markets to make better management decisions. To do so would be a significant departure from management doctrine, which is that — no matter how “flat” your organization — most important decisions are made by the CEO/COO/CFO. But if markets, no matter how much the participants are laypeople, make better predictions than experts, then that’s the logical next step.
Looking back on my own tenure as CEO of a 125-person agency, and as Finance Director of a 13,000 member association, I am pretty sure the “market” — had we had them — would have made some different decisions than I and my management colleagues made.
What’s this mean for PR practitioners? Probably little, now, since prediction markets are only now slipping in to management’s minds. But if I were doing PR for Microsoft’s MSN Virtual Earth, I would be concerned about 
” alt=”graph” />, for example. Taken from the Yahoo Research Buzz Game, a prediction market, it shows how poorly the “buzz” around MSN Virtual Earth (red dots, bottom graph) has been compared to GoogleEarth (blue dots). And how the “market” views the two properties (top graph). (Note, the green lines denote NASA’s product).
I’ve had only a few clients where a prediction market could have been deployed effectively (it takes a fairly good-sized pool, I believe), but I’ve no doubt the market would have made some of our marketing decisions easier and faster.
5 comments
10/3/2005 at 1:56 pm
Dana
This article was a little hard for me to understand, but since no one else has commented, I’ll get the ball rolling and risk looking a little dumb. J So correct me where I’m wrong.
(I’ll number my statements to make it easier to reply.
1.Okay. If I understand correctly, this new system, called a prediction market, is to allow a large number of people to rate their opinion on the likelihood of something like stock increasing or decreasing. (Which is a lot like the masses making a prediction.)
2.A point that is brought up is how more or less accurate this unified knowledge (or prediction) is over the knowledge (or prediction) of a professional.
3.SO… it is thought that regular people’s predictions pooled together is more accurate than any few professionals.
4.And… that this should be used as legitimate information to make managerial decisions.
Okay…What I don’t understand is the validity of this whole prediction market assumption. However, I do understand the thought process behind two head are better than one, but it these prediction markets are open to anyone with internet access then their could be large scale sabotaging of companies or just plain uneducated people giving their opinion.
Let me know if I’ve totally misinterpreted the blog post, and correct me where I’ve gone wrong.
10/3/2005 at 3:54 pm
Allan Jenkins
Good points, Dana, pointing out where I haven’t been clear (and since I am in the middle of writing an article about prediction markets, your feedback is very helpful).
Let me try to make it clearer (without promising that I’ll do so!).
James Surowiecki’s example is that of cattle raisers at a fair who are asked to guess the weight of a young bull about to go on auction. All the cattle raisers have expert knowledge: they are around bulls every day, sell and buy them, and have pretty good eyes. None has “perfect” knowledge: none of them owns the bull, and so cannot know what it weighs.
So they place bets — I forget what the prize is, but it’s enough to make them think — on the bull’s weight. Most are fairly close, some more than others. As you would expect. But — and here’s the but — as a collective group, they were within ounces of the right weight.
What’s important in the example are three things: 1) the participants were knowledgeable, 2) there was an incentive to get the answer right, 3) the participants did not influence each other.
To your points:
1) The predicition market idea is based on financial markets. But not stocks and bonds. Instead, most are best on options that expire when the question is resolved, with a payout of either 1 or 0 (players buy at some point between 1 and 0).
2) In most areas of life and commerce, a thousand laymen are smarter than 10 experts.
3) Yes.
4) Yes, certainly in an advisory sense.
Your issues are that anyone could access and anyone could sabotage. Prediction markets for internal decision making, as Google is running, are internal. Only employees would be able to access. The “market” factor would render sabotage irrelevant. And the participants “insider” status nulls out the “ignorant” people argument.
On public prediction markets, sabotage is limited because no player is given enough funds to actually influence the market. And, even with a few hundred players, the market is supple and quick to react.
Here’s an example: I offered to buy Auburn beats So. Caro contracts at 40 last week (that’s basically saying Auburn would lose)… the market laughed at me. Not a single taker. I finally gave in and bought some at 85.
Now, I am a Carolina fan. And showing it in my opening bid. But a) I could not sabotage Auburn, who were either going to whip on field or not no matter my opinion and b) as uneducated as I am, I felt able to muster enough knowledge to make a bet, and c) even though I hoped Auburn would lose (in a vague way, I am not all that partisan), I realized that they were unlikely to, and so I bought Auburn contracts.
The market worked, in that case. And, in fact, there’s enormous evidence that it usually does.
10/11/2005 at 11:29 pm
Jeff Osborne
Interesting post. Please forgive my preaching to the choir, but some readers may not have read Thomas Malone’s book “The Future of Work”. It has a great chapter about using the power of internal predictive markets and similar mechanisms in a management context. It’s exciting and not completely suprising to hear that Google has taken to the idea.
Predictive markets seem bound to pop up increasingly in the near future. I’d be really curious how Google’s managers would evaluate the transactions costs of participants who engage in their “free” predictive markets.
9/25/2007 at 6:41 am
Arthur Kathan
Regarding internal predictive markets, I cannot help but wonder if the wisdom of the bettors is simply an opportunity to express an opinion to management without having to tell management what it wants to hear. If one’s job is not on the line, then those closest to a project (and presumably more numerous than mgmt.) can express their best (not most optimistic, not deadline dependent) guess.
9/25/2007 at 2:32 pm
Allan Jenkins
Arthur, that should be a big part of the appeal for management. “Players” send management objective anonymous signals, instead of subjective “signed” signals.
The technology and the theory has moved forward a lot since I first wrote this post, though I see little evidence companies are picking up on it. That’ll change, though.