The Industry Standard taps crowd wisdom

Are crowds really wise? A growing number of businesses are betting they are. The latest venture is a new incarnation of The Industry Standard, the poster child for the rise — and precipitous fall — of the dot-com boom.

You might think that bets placed with Monopoly money wouldn’t be meaningful. You’d be wrong. A significant body of academic research supports the idea that predictive markets — such as Iowa’s Electronic Markets, which features small wagers in real money on national politics — actually outperform conventional polls, whether or not real money is involved. Moreover, the data generated by those markets has real value.

Hewlett-Packard, for example, has been experimenting with an internal predictive market for at least seven years. In one early trial, sales people were encouraged to bet on the number of printers the company would sell. The result? The bettors’ predictions were closer to the actual sales numbers than HP’s own internal forecast, said Bernardo Huberman, director of information dynamics at HP Labs.

Emile Servan-Schreiber, a cognitive scientist and artificial intelligence engineer who co-founded NewsFutures, a five-year-old, online predictive market, says that Google, Microsoft, and perhaps 100 other companies are also experimenting with internal predictive markets. And HP, said Huberman, plans to monetize its work by selling predictive market software and possibly running markets for other companies.

Indeed, the global equities markets are, in effect, predictive markets, said Stanford University’s William Sharpe, a Nobel Prize-winning economist. Prices, he said, are very efficient accumulators of information. What about predictive markets for politics and other arenas? “I’d much rather look at the Iowa market than any of the polls. It’s more accurate,” he said.

Betting on the White House — and more It’s generally illegal in the United States to bet real money in a predictive market. But the Iowa Electronic Markets was granted an exemption because it is widely used for academic research.

It works like this: A given candidate – say, Hillary Clinton – is represented by shares of virtual stock. In the event that she actually wins the nomination, each share will be worth $1 in real money. If she fails, the stock becomes worthless. Until the day of the actual nomination, traders can buy and sell — and even short — her stock.

On the evening of Feb. 1, her shares were selling for a bit more than 59 cents, while rival Barack Obama’s were valued at a bit below 48 cents. On the Republican side, shares of John McCain had soared to 82 cents, well above his closest competitor, Mitt Romney, at 12 cents

The money at stake in the Iowa market isn’t large; the maximum investment is just $500. But it attracts legions of savvy political junkies, whose opinions are remarkably prescient. How prescient? The University of Iowa’s business school recently audited the market and found that, since 1988, it beat the conventional polls’ accuracy by an average of 74 percent. Bet2give, part of NewsFutures, also uses real money, but any winnings go to the charity of the trader’s choice.

Readers of the new Industry Standard will be able to bet fake money on a variety of topics related to online business. For example, a prediction might state, “Apple will ship 10 million iPhones by the end of 2008” or “High-tech venture funding will decrease by 15 percent in Q2 2008.”

The predictive market is just one leg of the new publication’s offerings. It will also feature news stories and columns by influential industry thinkers such as Guy Kawasaki, an entrepreneur most famous for his former role as Apple’s “chief evangelist,” and Mark Anderson, CEO of the Strategic News Service.

Optimum conditions for wise crowds Remember the old IT dictum “garbage in, garbage out”? The same applies to predictive markets. After all, “the problem with the global village is all the global village idiots,” a saying that blogger Paul Ginsparg popularized.

The New Yorker columnist James Surowiecki, whose book “The Wisdom of Crowds” is widely credited with igniting the boom in predictive markets, said crowds can be smart if they address four conditions: diversity, independence, decentralization, and aggregation.

Why decentralize? Good decisions (or good bets) are made by individuals based on their own local and specific knowledge. The open source software development process is an example of decentralization in action.

Diversity is achieved largely through the absence of hierarchy (markets don’t have vice presidents), which ensures that no single person has too much influence and that diverse viewpoints don’t get shut out, said Surowiecki.

That may be why the new Industry Standard won’t have is an editor in chief. “No single voice will dominate the discussion, which is why we decided to forgo the somewhat print-centric idea of an editor in chief, despite talking to some great people for the position,” said Managing Editor Ian Lamont. “We want readers to get viewpoints from the widest range of contributors possible, with the common theme being that these contributors are all people who believe that the Web is a major paradigm shift in business.”

Turning crowd predictions into revenues HP has refined Surowiecki’s ideas, developing a methodology that may improve the quality of predictions while using a group, instead of a crowd. Assuming that more knowledge equals a better prediction, HP’s Huberman and researchers Leslie Fine and Kay-Ute Chen select participants with knowledge of a specific area — supply chains, for example — and then have them play a stock-market-based game.

The results of the game let the researchers grade traders on their willingness to take — or avoid — risk. They then participate in a market, and their predictions are adjusted by their risk scores. The players win real money, and their predictions are of real use to the company, Huberman said.

One of the most successful predictive markets, the Hollywood Stock Exchange, takes play-money bets on weekly movie grosses, Oscar nominations, and the like. Traders’ predictions of opening-weekend returns are more accurate than the movie industry’s forecasts, and the exchange has done a good job of foreseeing nominations as well. In 2002, its traders correctly predicted 35 of the 40 Oscar nominees in the top eight categories, Surowiecki wrote in The New Yorker several years ago.

The exchange monetizes those predictions by slicing and dicing the data collected on the site and selling (in very real money) it to movie studios and other interested parties, said NewsFutures’ Servan-Schreiber. The exchange is private, so there’s no way to estimate how much cash it generates, but it’s successful enough to have been purchased by Kohlberg Kravis Roberts, a major private equity firm.

Taking a leaf from that playbook, the new Industry Standard expects to sell slices of the data it collects on the predictive market via IDC, IDG’s market research arm, said Derek Butcher, The Industry Standard’s vice president and general manager.