You can be wiser than a crowd

It’s a well-documented fact that if you ask a large group of people to guess how many jelly beans there are in a jar, they might all be wrong — but the average of their guesses will be very close to right.

The group requires no special understanding of jars or jelly beans for the trick to work. Instead, a crowd of uninformed people collectively — and somewhat magically — arrives at the right answer by being wrong in different but error-cancelling ways. 

This is the “wisdom of crowds” — a term popularized by James Surowiecki, whose book on the subject suggests that the magic works far beyond beans in jars. 

“Under the right circumstances, groups are remarkably intelligent, and are often smarter than the smartest people in them,” he wrote in 2004. Even if “most of the people within the group are not especially well-informed or rational.”

It’s a comforting thought: No matter how wrong we are individually, collectively we can be right.

“When our imperfect judgments are aggregated in the right way,” Surowiecki says, “our collective intelligence is often excellent.” 

Unfortunately, subsequent studies suggest this only works for things like counting jelly beans.

When researchers at MIT asked a crowd of people whether Philadelphia is the capital of Pennsylvania, the crowd proved not very wise — their collective answer was “yes.” 

But the researchers also asked each respondent what they thought everyone else would answer. 

A majority of the respondents who knew the correct answer — “no” (because Harrisburg is the capital of Pennsylvania) — also knew that most of the others would get it wrong and say “yes” (because it does seem like the great city of Philadelphia should be the capital).

This gap — between what informed people answered and what they expected others to answer — created a signal that pointed toward the truth of the matter.

By detecting and amplifying that signal, MIT’s researchers cut prediction errors by 21% relative to just following the crowd. 

This suggests that group prediction is not a magical process of uninformed guesses averaging out. Just the opposite: It’s simply a matter of how well-informed the group is.

It’s true that a large number of uninformed guesses averages out closer to the truth than a small number of them would do. One study even found that uninformed people improve their accuracy by averaging out a bunch of their own guesses (forming their own personal crowd, as it were).

But the study also found that informed people needn’t bother, because they give essentially the same answer every time.

In short, uninformed people are better in a crowd. But — contra Surowiecki — crowds are better when they have fewer uninformed people. 

Prediction markets seem to confirm this less magical, more intuitive result.

In the Iowa Electronic Markets, for example, Cass Sunstein found that only a small subset of traders set the prices: “It turns out that 85 percent of the traders aren’t so smart. They hold onto their shares for a long period and then just accept someone else’s prices. The market’s predictions appear to be driven by the other 15 percent — frequent traders who post their offers rather than accepting those made by other people.”

This makes predicting the outcome of an Iowa caucus different from guessing how many jelly beans are in a jar. 

Prediction markets “do not simply rely on the median or average judgment of a randomly selected group of people,” Sunstein explains. “They are genuine markets, in which people voluntarily choose to participate, presumably because they think they know something.”

The operative word here is “think,” because new data shows that prediction-market bettors don’t seem to know as much as they think they do.

“We find that even the largest and most visible markets frequently behaved in ways inconsistent with rational, information-driven updating,” an academic study reports. 

The authors of the study measured the efficiency of prediction markets in part by looking at how the same bets traded on different venues. 

They do not seem efficient: “There were almost no days in which the prices of similar contracts moved together in rationalizable ways.” 

No days!

Instead, the same political event produced different price movements on different platforms, which is not how it’s supposed to work.

The study’s findings suggest that political prediction markets are just less efficient versions of regular markets: “Prices were shaped more by the actions of traders in each individual market than changes in the overall political landscape.”

So what do the researchers think of the collective wisdom of these traders? “Our results cast doubt on whether political prediction markets necessarily aggregate private information into accurate and efficient collective forecasts.”

This is excellent news.

If the wisdom of crowds worked on predictions in the same way it does on jelly beans — by averaging out a multitude of uninformed guesses — there wouldn’t be any reason to participate in prediction markets; the prices would always be correct.

But prediction markets do not work by wisdom-of-crowds magic; they’re only accurate when well-informed traders dominate. 

Which means there must be money to be made!

You only have to pick the correct crowd to bet against.

— Byron Gilliam

_{Brought to you by:}

AI hits a power bottleneck. Bitcoin miners have the solution: plug-in ready energy grids.

Miners are pivoting to High-Performance Computing, leasing grids to AI hyperscalers. VanEck Onchain Economy ETF (NODE) captures this convergence, holding the infrastructure bridging digital assets and the AI grid.

Crypto’s premier institutional conference is back this March 24–26 in NYC.

Don’t miss SEC Chairman Paul S. Atkins’ keynote on Day 1.