Do large groups of people make more intelligent decisions than a selection of experts?
One of the remarkable stories surrounding the mysterious disappearance of the aircraft of Malaysian Airlines flight MH370 came from the American company DigitalGlobe. It recruited a large number of Internet volunteers to assist with the search for the missing aircraft from their desktops, using its satellite imagery (which powers Google Maps). Each volunteer scrutinizes a piece of the ocean. The more who sign up, the greater the chance that the aircraft will finally be found.
Or at least that's the theory.
Unfortunately, so many volunteers signed up so quickly that the DigitalGlobe website crashed. (It's since been restored.)
Yet the idea is not so crazy after all. The deployment of the masses to help solve a problem, or crowdsourcing, relies on the principle of "the wisdom of crowds." The idea is that large groups of people make more intelligent decisions than individuals, even though they may not be experts.
The theory is based on the work of Sir Francis Galton, a cousin of Charles Darwin. Galton was a staunch supporter of Darwin's theory of evolution theory, which implied that intelligence is heritable. Galton argued that it would be better for society if the most intelligent people – that is, the elites – made the most important political decisions, as the masses could not be trusted.
To demonstrate this, he conducted several experiments. At a cattle market, Galton asked various experts – such as farmers and butchers – to guess the weight of the bull. Then he asked the crowd at the market the same question. When he analyzed all the answers, many from the crowd were far off. But when he calculated the average estimate of the crowd, he found it to be quite close to the actual weight of the bull – and closer than the average judgment of the experts.
The principle was later described in an excellent book by journalist James Surowiecki, who reported on the success of the masses in predicting the outcome of political elections, the future share prices of companies, and the outcome of the sports tournaments, as long as enough people were polled.
I demonstrate the effect in a lecture in which my students must guess how many paper clips are contained in a glass. The average estimate often turns out to be pretty close.
And TV producers have also been inspired by the wisdom of crowds. A stumped “Who wants to be a Millionaire” contestant, for example, has the option of consulting the audience, which nearly always collectively picks the right answer.
Yet other research shows that the masses are not always so intelligent. The judgments must be made independently, they must be decentralized and there must be diversity in opinion. If someone during my lecture were to shout, "There are 500 paper clips in the glass!" people would adjust their estimates and the effect would disappear. Similarly, when someone is appointed as the leader of the crowd, to make a decision on behalf of the group the effect does not appear.
Another crucial condition is that the problem being addressed by the crowd should not be too complex. A recent German study asked people complex questions such as, "How often should you flip a coin to ensure that the probability that you get heads each time is as small as the probability that you win the lottery?" * With such questions you are better off hiring a trained statistician than relying on the average guess of a crowd.
In discussions about the benefits of Big Data the wisdom of crowds effect is highly relevant. The argument is that by collecting large amounts of information we may be able to discover the causes of certain diseases, find patterns that point to match fixing in sports, or prevent acts of terrorism. But Big Data also requires Big Judgments. Ultimately, it takes experts to select from the large pool of information which data should be analyzed and what should be ignored. So even in the era of Big Data, experts are more necessary than ever. They can turn a mystery into a solvable question.
When a problem is not too complex, yet a large amount of information is needed to solve it, the wisdom of crowds can useful. So we might well benefit from consulting the masses on what they think is the most likely explanation for the disappearance of the Malaysian Airlines jet. Yet we would first need the experts to provide us, the masses, with a list of plausible scenarios to choose from.
* The chances of throwing heads 24 times consecutively is about 1 in 35 million, which is approximately equivalent to winning the lottery in Germany, where this study was conducted.
Follow me on Twitter: @ProfMarkvugt
If you want to participate in a "wisdom of crowd"-study to search for the missing aircraft, please click on the following link: http://fppvu.qualtrics.com/SE/?SID=SV_0HwvG6Ehdc9Lsih