Google Gets Answers as Staff Bets on GMail, `Star Wars' Success

By Crayton Harrison

July 10 (Bloomberg) -- The day after a company's stock rises, employees are more optimistic about the projects they're working on. You can bet on it.

And at Google Inc., they do.

That's what researchers found in the behavior of employees on an internal trading system the company designed. Using a faux currency called the Gooble, 2,000 workers have wagered on about 370 subjects, from the success of the company's Gmail service to the quality of a new ``Star Wars'' movie.

Academic studies show these so-called prediction markets work as financial modeling tools. Google managers use the results as a reference in strategy meetings and crunch the data to see how employees behave. After finding traders got bullish about meeting goals following a climb in the stock, Google started examining how productivity and optimism are connected.

``It may be that employees also do their jobs differently on that date,'' said Bo Cowgill, the manager at Mountain View, California-based Google who worked on the research. ``They may have other expressions on optimism -- the hours they work or how much stuff they manage to get done, the scores they give to candidates being interviewed that day.''

Because markets of this kind are based on anonymous bets, employees can weigh in without worrying what managers will think, said John Maloney, who helps organize research conferences on the subject.

`Well Served'

``They work quite well,'' said Maloney, whose San Francisco company, Net Intelligence, promotes academic talks. ``If you can find a problem that has a number of stakeholders in it where it's very difficult to figure out the outcome, you'd be well served by a prediction market.''

The University of Iowa's electronic market, started in 1988, lets students invest $5 to $500 to trade on who will win the popular vote in the next U.S. presidential election. On Cantor Fitzgerald LP's Hollywood Stock Exchange, participants wager on the box-office revenue of film releases such as ``The Dark Knight,'' a Batman movie that debuts this month.

``There's a lot of things companies want to know that they have trouble getting through ordinary channels,'' said Robin Hanson, an associate economics professor at George Mason University in Washington. ``Prediction markets are a mechanism for cutting out some of the crap, getting people to be honest.''

Apple, `Star Wars'

Hewlett-Packard Co., Microsoft Corp. and Intel Corp. have tried prediction markets, and companies including retailer Best Buy Co. use such tools to get employees' views on sales forecasts, Hanson said.

The data can be revealing, especially when combined with staff locations, academic backgrounds and social connections, Cowgill found. The links between Google workers were like ``a huge hairball'' to be unwound, he told an audience in February at a New York industry conference.

Google started the project in April 2005, basing it on the Iowa markets. Most of the ``securities'' were bets on what would happen to products, such as how many users the Gmail service would accumulate in a given quarter, or industry events, such as when Apple Inc. would release a new computer model.

The company sprinkled in off-topic questions, such as whether critics would pan the ``Star Wars'' installment ``Episode III: Revenge of the Sith.'' (Workers thought the movie would be a bomb until positive reviews began to pop up, prompting a ``huge swing'' in the market, Cowgill said.)

Location Key

With Justin Wolfers of the University of Pennsylvania and Eric Zitzewitz of Dartmouth College, Cowgill pored through the data to find patterns. In addition to correlating optimism with a stock increase, they noticed that newer employees tended to make trades that were more optimistic about Google's performance.

Google found the trading patterns of employees were closest to those who sat nearby, showing a stronger link than among people who collaborated on a project or had the same boss. The conclusion helped vindicate a Google policy of cramming staff close together, said Cowgill, who shares an office with Hal Varian, the company's chief economist.

Cowgill said he's interested in other ways employees rub off on each other.

``If you have a team of folks who aren't performing as strongly, if you put someone in there who is strong performer, will that bring everyone up or pull everyone down?'' he said.

To contact the reporter on this story: Crayton Harrison in Dallas at tharrison5@bloomberg.net.

Last Updated: July 10, 2008 00:01 EDT