A mini-interview with Dr. Robert Hanson, crowd prediction data expert
Reporter's Notebook
Assignment 
Steve Peterson conducted a mini-interview with Dr. Robin Hanson in April 2006 about crowd predictions. Here's what he got from an e-mail discussion with him:
Q: This morning [April 18,2006] as the price of oil was rising quicker than gas pump price and gallon meters when I fill up my tank, I was wondering if investors' concern that the United States will exercise military action against Iran is a valid predictor. Spurred by my interest, I e-mailed predictive/information market guru Dr. Robin Hanson at George Mason University.
Do you think that the record high price of oil in world markets indicates that the United States will probably use military action against Iran? The world financial markets seem to act like information markets at times. Futher, since military action in Iran would greatly affect investors and brokerages, are these players reliable enough to gauge the probability of such action?
A short time later – to my surprise – he responded with the following answer:
Answer: They are reliable to give estimates to the questions they are asked. But oil markets are not asked about whether US will attack Iran, and so oil prices do not answer that question. Tradesports.com has some markets more directly on a US attack on Iran.
After that, I decided to visit Tradesports.com, and on the homepage was a market devote to the question, "Air strike against Iran by the end of the year?" Apparently, Tradesports users were less concerned about potential military action against Iran since they collectively wagered a 20.5% chance that an air strike would occur by the end of this year. The assertion that predictive/information markets answer specific questions makes sense, but I still had another question for Dr. Hanson:
Question: It makes sense that information markets work best when they have a specific question to answer. However, I came across something interesting recently that might show that this precondition is not always required. I just read James Surowiecki's The Wisdom of Crowds in which he related how investors on Wall Street were able to determine which of the main contractor's parts caused the Space Shuttle Challenger to tragically explode within hours of the accident. How does this fit into information market theory? Was Wall Street an information market in this case? Does a group of people have to focus on a specific [question] to form an information market?
In response Dr. Hanson said:
Answer: In the day or two after the Challenger accident there was little else to influence the value of those stocks, so those stock price changes have a clean interpretation. That is an unusual case.
Fair enough, Surowiecki's example was a special case. However, I wonder how much we could learn from Wall Street (and other major international financial markets) from such analysis...
Dr. Robin Hanson, an economics professor at George Mason University, is a great resource for crowd prediction data. In fact, he tried to get the government set up a major prediction market that would have experts buy contracts about general geopolitical trends, but several senators shut down his project with the Department of Defense's Defense Advanced Research Project Agency (DARPA) since they thought it market would predict calamities like terrorists attacks (http://money.cnn.com/magazines/fortune/fortune_archive/2003/09/15/349149...). Dr. Hanson's information is at http://hanson.gmu.edu/.
Background
Crowdsourcing Predictions
The power of turning to a large group to answer a factual question is fairly obvious.
But what about asking a crowd something that cannot be known? Question like:
- Who will win the next presidential election?
- What will be the next breakout technology?
- When will Osama bin Laden be captured?
Financial markets have been getting crowds to predict the future for centuries. Now that concept is being widened and deepened. Using the same basic concept -- put your money where your mouth is -- people using services like [[http://www.intrade.com|Intrade]] are offering a new way to detect trends and outcomes. And they can be [[http://query.nytimes.com/gst/fullpage.html?res=9401E4D7163EF937A25751C0A9619C8B63|remarkably prescient]].
The Financial Times this month [[http://www.ftpredict.com|announced a contest]] with a $25,000 prize for a four-month contest. "FTPredict is a fantasy stock league for people who care about elections, finance, business and geopolitics", [[http://www.intrade.com//?request_operation=main&request_type=action&checkHomePage=true#|said Robin Johnson]], President of the FT in the Americas.
Is this for real, or "for entertainment purposes only?" Who takes part? Can this be gamed?
These are some of the questions the writer of the main story will tackle.
Filed Reporting
The Ability to Trust Crowd Predictions
Reporting page:
A quick exchange with a predictive market guru
Steve Petersen interviews Dr. Robin Hanson via email, April 14th-17th, 2007
This morning [April 18,2006] as the price of oil was rising quicker than gas pump price and gallon meters when I fill up my tank, I was wondering if investors' concern that the United States will exercise military action against Iran is a valid predictor. Spurred by my interest, I e-mailed predictive/information market guru Dr. Robin Hanson at George Mason University.
Steve Petersen: Do you think that the record high price of oil in world markets indicates that the United States will probably use military action against Iran? The world financial markets seem to act like information markets at times. Futher, since military action in Iran would greatly affect investors and brokerages, are these players reliable enough to gauge the probability of such action?
Robin Hanson: They are reliable to give estimates to the questions they are asked. But oil markets are not asked about whether US will attack Iran, and so oil prices do not answer that question. Tradesports.com has some markets more directly on a US attack on Iran.
After that, I decided to visit Tradesports.com, and on the homepage was a market devoted to the question, "Air strike against Iran by the end of the year?" Apparently, Tradesports users were less concerned about potential military action against Iran since they collectively wagered a 20.5% chance that an air strike would occur by the end of this year. The assertion that predictive/information markets answer specific questions makes sense, but I still had another question for Dr. Hanson.
Q: I just read James Surowiecki's The Wisdom of Crowds in which he related how investors on Wall Street were able to determine which of the main contractor's parts caused the Space Shuttle Challenger to tragically explode within hours of the accident. How does this fit into information market theory? Was Wall Street an information market in this case? Do a group of people have to focus on a specific [question] to form an information market?
A: In the day or two after the Challenger accident there was little else to influence the value of those stocks, so those stock price changes have a clean interpretation. That is an unusual case.
---
Fair enough, Surowiecki's example was a special case. However, I wonder how much we could learn from Wall Street (and other major international financial markets) from such analysis... As a follow up I had a more in-depth interview with Ken Kam from Marketocracy
4/15/07Google Prediction Markets Group Members
Reporting page:
There is a prediction markets community at Google Groups -- see http://groups.google.com/group/Prediction-Markets?hl=en. Perhaps some of the group members would have some interesting insight into crowd predictions.
4/5/07
Dr. Robin Hanson
Reporting page:
Dr. Robin Hanson, an economics professor at George Mason University, is a great resource for crowd prediction data. In fact, he tried to get the government set up a major prediction market that would have experts buy contracts about general geopolitical trends, but several senators shut down his project with the Department of Defense's Defense Advanced Research Project Agency (DARPA) since they thought it market would predict calamities like terrorists attacks (http://money.cnn.com/magazines/fortune/fortune_archive/2003/09/15/349149...). Dr. Hanson's information is at http://hanson.gmu.edu/.
4/4/07
This is unedited content. What's that?
Related Assignments
My Assignments
