The Tech - Online EditionMIT's oldest and largest
newspaper & the first
newspaper published
on the web
Boston Weather: 41.0°F | A Few Clouds and Breezy

Google Staff Lay Bets On Company’s Future

By Ian Austen
THE NEW YORK TIMES

Like all search engines, Google helps people sort through information from the past. But a new service, being used inside the company, tries to forecast the future.

Google has created a predictive market system, basically a way for its employees to bet on the likelihood of possible events. Such markets have long been used to predict world events, like election results. Intrade, part of the Trade Exchange Network, allows people to bet on elections and even the weather, for example.

In Google’s system, employees can bet on how the company will perform in the future, forecasting things like product introduction dates and new office openings. It was devised under a program that allows engineers to spend one day a week on a project of their choice.

To help develop the system, Google consulted Hal R. Varian, an economist at the University of California, Berkeley. Varian (who also writes the Economic Scene column for The New York Times) said that the final product was not entirely what he anticipated.

“I was a little surprised,” he said. “I expected this to be accurate because there’s a lot of literature and experience with these systems. But this has been even better than I expected.”

Google has not offered precise data on the system’s accuracy, but a chart posted on the company’s blog last week showed that, in the words of its accompanying entry, prices set for events through employees’ wagering were a “pretty close” indication of the probability of events.

The market is based on the idea that a price established for an event will reflect bettors’ consensus of the likelihood that it will happen. Thus, something priced at 20 cents should happen 20 percent of the time. The system accepts bets in 10-cent increments up to a dollar (no actual money is involved).

On its blog, Google compares the market to its search engine software. “Our search engine works well because it aggregates information dispersed across the Web, and our internal predictive markets are based on the same principle: Googlers from across the company contribute knowledge and opinions which are aggregated into a forecast by the market,” the blog said.

Varian, who has consulted with Google on other projects, attributes the higher-than-expected levels of accuracy to the large number of employees participating. In general, the higher the number of bettors in such systems, the better the predictions.

There is one issue, however, for which Google’s market offers no prediction. “It’s a fun thing,” said Varian. “Now one of the things we’re thinking about is what to do with it.”