Prediction Markets: Tapping the Wisdom of Crowds + Yahoo!’s new currency: Yootles + Yootopia project

Quick random notes on: confab.yahoo on prediction markets: Dec 13, 2006 – 5:30pm

—

#1. It would be great if the conference were available as a MP3, MPEG, or MP4 file —so as to replay it in local. Thanks.

Addendum: The replay of the David Pennock-endorsed event can be viewed at two connection speeds: 100k | 300k

#2. The MicroSoft guy said that what is important for firms is the business value at the end of the day. Two sets of actors should play in sync: the forecasts and the executives who (should) listen to them.

#3. James Surowiecki pointed that accuracy should be understood as the prediction markets versus the other forecasting methods. (He made another point, alluding that people were expecting absolute accuracy from prediction markets, because it’s a brand-new tool, which people don’t know the limitations yet. Not a stupid point. I hope that he’s right on that, and that the media will educate people about relative accuracy.)

#4. The Google guy made a brainy point regarding prediction markets and reputation systems. I will need to replay the video to mind his point. (And who the hell was the village idiot he mentioned in his intro??)

#5. In his intro, James Surowiecki made a point I want to discuss. He said that prediction markets are the world’s 8th marvel in theory, but, in practice, corporations’s senior executives are, as of today, very slow to embrace this new forecasting tool. Why?… he asked. I don’t remember his answer, but I will tell you something. What about taking the problem in reverse? What about blaming the prediction market industry instead of blaming its (prospective or current) customers? Would it be useful to brainstorm about the industry’s marketing approach, and what should be improved? New products, new processes, new designs, new topics, new people, new explainers, whatever. James Surowiecki could be right, but he could also be wrong: maybe internal prediction markets will take off once the vendors have reduced their complexity —or else (any idea?).

#6. Robin Hanson said that prediction markets reveal information “IF YOU GET PEOPLE TO TRADE IT”. Exactly. And later on, he said he favors paying employees for trading internal contracts so that the corporation (the sponsor) gets valuable information. His recommendation: “go for the jugular”, address the biggest issue at hand, so as to raise maximum awareness (and increase accuracy, I suppose).

#7. There has been a discussion on contract statement —what’s a good question, what’s a bad question. Robin Hanson mentioned the NKM contract —it’s not clear whether Robin Hanson blamed the traders for not reading the contract statement correctly or TradeSports for crafting an ambiguous contract statement (or maybe both).

#8. The other speakers were great, too. I will mention them later on. (Sorry. Very late.) [Addendum: See Eric Zitzewitz's pre-event reporting on his presentation: Prediction Markets for the CFO.]

#9. And, as I suspected just before the event, there’s a worldwide conspiracy going on in America and the goal is to downplay BetFair, the world’s #1 prediction exchange. BetFair was not cited a single time.

—

Addendum: Via George Tziralis, this Dan Farber story…

- Yahoo’s new currency–the Yootle

Pennock said that Yahoo Research is working on an entire Yootle economy with prediction markets inside, as well as other forms of auctions, group decision-making and friendly wagers. It’s not hard to imagine Yootles and group decision making mechanisms baked into the Yahoo’s virtual social network further down the road. When you have half a billion connected users as Yahoo does, making friendly wagers can be a compelling social application.

—

Links:

- Yootles.com — Mechanisms for Group Prediction and Decision Making

- Paper on Yootles – Group Decision Making with Yootles – (PDF) – by Daniel M. Reeves, Bethany M. Soule and Tejaswi Kasturi – 2006-11-10

Abstract: Yootles are a currency (scrip system) for group decision making. We briefly introduce the yootles infrastructure being developed at Yahoo! Research and then suggest an array of decision mechanisms for small groups. The mechanisms themselves are currency-agnostic. We describe new and existing mechanisms for choosing among a short list of options, choosing among an effectively innumerable list of options, as in meeting scheduling, allocating shared goods and responsibilities, public good provision, and bilateral trade. We list desirable mechanism properties and describe the tradeoffs that the mechanisms make among them. Finally, we present new equilibrium results for two-player versions of some of the above mechanisms.

Conclusion: We have introduced a new currency (yootles) for facilitating group decision making. We focus here not on the currency but on an array of decision mechanisms for common group decision problems. In particular we introduce a new mechanism (DAUC) for redistributing the VCG surplus to achieve greater fairness while retaining a semblance of incentive compatibility. We also introduce the NFA mechanism for achieving all desirable mechanism properties when agents play altruistically. For many applications this is less useful but we describe applications of NFA to domains where it is a sensible mechanism (bilateral trade and joint purchase). For domains such as meeting scheduling where the submission of full preferences is costly we propose iterative versions of the above mechanisms. Finally, we describe our implementations of these mechanisms in a forthcoming service from Yahoo! Research.

—

Psstt… Spot the new “Currencies” sub-category at the bottom of this blog post.