George Mason University is leading a team of researchers in the search for a way to harness the “wisdom of the crowd” in order to allow the U.S. government to better predict and prepare for future world events.

Headed by Mason research professor Charles R. Twardy, the Decomposition-Based Elicitation and Aggregation team will attempt to develop a system that would improve an intelligence analyst’s predictions of events pertinent to U.S. interests.
These would include events that occur in the political, economic and medical spheres.

“The goal is for the U.S. government to have a heads up that major events, like the Arab Spring, are coming before the events even happen,” Professor Kathryn B. Laskey said.

Laskey is the co-leader of the project. Both Laskey and Twardy are with Mason’s Center of Excellence in Command, Control, Communications, Computing and Intelligence.

“Our research could help businesses and the government to make better forecasts about world events,” Laskey said. “It would be good for our economy, it would be good for our national security. … It would improve our quality of life as citizens.”

The DAGGRE approach to better predictions will include crowdsourcing questions about the possibility of certain events happening and breaking down the questions so people can choose to answer only the parts that they know about, Twardy said.

“We are trying to improve estimates and forecasts by taking the judgment of many people, each of whom has expertise in different areas, and combining them mathematically to make a better overall, or aggregate, forecast,” Twardy said.

With crowdsourcing the team can take advantage of the research-proven wisdom of the crowd, Twardy said. “Asking the best expert is not as good as getting the opinions of many informed people.”

However, the DAGGRE team wants to go beyond forecasting based on the average prediction of a crowd, a method Twardy said has been hard to beat so far.

“Everyone has this intuition that if you could put weight on the right people, for each question, you could do better [at predicting future events],” Twardy said. “The trick is picking the right people for that question ahead of time.”

After the first year of the project, the team hopes to have found several techniques that improve prediction accuracy over the predictions based on the “unweighted average” opinions of a crowd.

“What is particularly interesting about our [method] is that we are going to ask people to make conditional estimates,” Twardy said.

This means people will be able to include assumptions or “ifs” into their predictions concerning world events.

For example, to answer a question, a forecaster can say, “Under this assumption I think the probability is ‘this’, under that assumption I think the probability is ‘that,’” Laskey said.

“As far as we know no one else is doing the same thing we are doing,” Lasky said.

The DAGGRE team is still looking for participants to join the team and make predictions. Twardy said ideal candidates would be people who keep up with current events and have an area of expertise, or those who want to learn more about the world, and have an interest in forecasting and improving their forecasting ability.

Those interested in participating in the project can sign up anonymously at daggre.org. More information can also be found on their blog, blog.daggre.org, or by signing up for their Twitter feed @DAGGRE_.

The DAGGRE team has a $2.2 million contract from Intelligence Advanced Research Projects Activity for the first year of their research, with the possibility of receiving $8.2 million over a four-year period. The funding comes from IARPA’s Aggregative Contingent Estimation program.

Other members of the DAGGRE team include Mercyhurst College, James Madison University, the Australian Centre of Excellence for Risk Analysis at the University of Melbourne, Defense Research and Development Canada — Toronto, as well as the two companies nemoSibi Ltd. and Fairfax-based KaDSci LLC.