Science Plugs Into Prediction Markets
Although some set-up hurdles remain, a market in scientific predictions may help us forecast innovation.
Over the weekend, Warner Brothers released a science-fiction thriller serendipitously timed to feed off public fascination with a real-life scientific breakthrough. Biologist Craig Venter announced May 20, to equal praise and alarm, that he had inserted synthetic DNA into a living cell.
“It was an interesting juxtaposition with science and science-fiction,” said David Rejeski, who directs the Woodrow Wilson Center’s Science and Technology Innovation Program. As such, the film’s debut offered the ideal opening experiment for STIP’s latest endeavor: a prediction market for science and technology questions that range from near-term bureaucratic events (when will the FDA approve the next oral antibiotic?) to far-off fantasy (who will make the first breakthrough with real “artificially intelligent” machines?).
In the spirit of the moment, STIP asked its following of science and tech enthusiasts last week to place a bet with virtual dollars in an online prediction market posing this question: “How much money will the film ‘Splice’ gross domestically during its opening weekend?”
By the time the market “closed,” the leading prediction among several hundred bets was $5 million to $10 million.
Alas for Adrien Brody — but to the credit of all the people who bet against him — Splice made a mere $7.5 million at the box office.
STIP’s prediction market is now turning the experiment to more monumental questions. Which Millennium Prize math problem will be solved next? When will the total number of known exoplanets reach 500? And as for that real “artificially intelligent” machine, will the breakthrough come from the entertainment industry, toy makers, airlines, a random private inventor, government, the auto industry or academic teams? (The leading opinion, as of Wednesday: academic teams.)
Prediction markets aggregate evolving opinion in the same way the stock market does (albeit with virtual currency — to use real money would constitute gambling). They’ve been used with success to predict elections and to aid businesses in forecasting future events like stock volumes and new product sales.
The U.S. government has even attempted to apply prediction markets before, although the lone foray was mostly a PR disaster. In 2003, the Pentagon tried to set up a market to predict future terrorist attacks.
“It basically ended up in the news as ‘Pentagon Plans Online Terror Bets,’” Rejeski recalled. That project was scrapped, and government has shied away from the idea ever since.
But no one has ever really applied the model to science and technology questions before. And if the Woodrow Wilson Center perfects the technique, currently with the help of a grant from the National Science Foundation, perhaps these tools could be used one day to inform policymakers.
For now, Rejeski is interested less in the predictions themselves than in how to make the market work.
“The biggest issue for us, and it’s a cutting-edge issue, is can we get people to play in a science and technology market?” Rejeski said. “If we can’t get them in there, the rest of this stuff becomes kind of irrelevant.”
The project is building up to a market that will open later this summer specifically around synthetic biology, one of STIP’s core interests. The emerging discipline draws on experts who don’t typically work together, from biologists to engineers to computer scientists. And so a prediction market may be a good place to corral their expertise.
The field of synthetic biology also lends itself well to the quantifiable questions necessary to run a prediction market: How many genes can scientists put together? Can they create a synthetic organism that will allow us to produce hydrogen fuel? When will that happen?
While people are mulling those questions, Rejeski is trying to answer many of his own: Can you get people interested in a prediction market when there’s no real money on the line? Can you identify the people who seem to be better at this than others? And how do you keep people in the game when the really big questions won’t be answered for months, or even years?
“The goal here is to make it a fun experience,” Rejeski said. “The secondary outcome is you might get some good predictions because people know stuff.”