Can politics be predicted?
- By Patrick Marshall
- Jun 08, 2018
Many political prognosticators were sorely humbled by the 2016 election of Donald Trump. Was it the hidden hand of Russia that tipped the scales and proved false the widespread expectation of a Clinton victory? Or were the prognosticators’ tools themselves faulty?
In this new climate of uncertainty about global politics, the office of the Director of National Intelligence has issued a blunt challenge: Can you do better?
Intelligence Advanced Research Projects Activity in March launched the Geopolitical Forecasting Challenge. The challenge is to “develop solutions that produce probabilistic forecasts in response to numerous closed-ended forecasting questions that concern specific, objectively verifiable geopolitical events containing timeframes with deadlines and locations.” In other words, come up with a program that can accurately predict political events.
Those who haven’t registered for the challenge can still do so even though the first milestone was completed April 30, with 10 awards of $750 dispersed to the top solvers. But the challenge is accepting rolling registrations, so solvers can register at any point. The challenge wraps up Sept. 7, so there’s still time to jump in to get a piece of the total $200,000 in prize money.
In the first round, solvers were challenged with 14 individual forecasting problems, including:
- Will Turkish President Recep Tayyip Erdoğan experience a significant leadership disruption by Aug. 31, 2018?
- What will be the maximum sea ice extent on the Barents Sea between Jan. 1 and April 10, 2018?
Clearly, some of the questions are heavy on scientific analysis while others require chops heavy on factoring social, cultural, economic and political issues.
In developing their solutions, solvers will have access to a continuously updated stream of data and forecasts produced by a crowd of human forecasters. In addition, solvers are welcome to develop or access other data sources, IARPA said.
Finally, relying solely on intuition to make predictions won’t cut it. To be eligible for a prize, the solver must provide a document detailing the solution method employed. That method can, however, include human judgments. According to IARPA, it is “interested in any method that advances the science of forecasting. This may include investigating how a combination of computer models and human judgments can be used to make accurate forecasts.”
Solvers will be judged against each other, as well as against the results of a platform developed by an earlier challenge --the Hybrid Forecasting Competition. The HFC platform relies more heavily on human input and judgments than the methods being nurtured by Geopolitical Forecasting Challenge. HFC, by the way, is still accepting new registrations as well, but analysts cannot participate in both challenges.
Full details of the competition are available at here.
Patrick Marshall is a freelance technology writer for GCN.