In On the Heavens, Aristotle imagined a man idling midway between food and water. He is equally hungry and thirsty, but without a good reason to pick one option over the other, he dies of hunger and thirst. It’s a thought experiment, still unresolved, that appears in the writings of Michele de Montaigne, Baruch Spinoza, and, most famously, the medieval philosopher Jean Buridan.
It remains unresolved because it raises a fundamental question about how people decide. If humans are rational and always opt for the best option, what happens if two options are equally good? In a world awash with choice—I’m thinking about the toothpaste aisle at my pharmacy—this arcane philosophical paradox feels contemporary. Even when we know what we need, simple decisions paralyze the mind.
I can’t tell you how to decide better. Maybe you should ask a friend to decide for you or avoid situations with choice overload. In his stunning profile of Barack Obama in Vanity Fair, Michael Lewis writes that the President only wears blue or gray suites. “I’m trying to pare down decisions,” Obama told Lewis. “I don’t want to make decisions about what I’m eating or wearing. Because I have too many other decisions to make.”
I can recommend one rule of thumb. It’s a deceptively simple and overlooked idea, one of many scattered throughout Paul Ormerod’s brilliant book Positive Linking: How Networks Can Revolutionize the World.
Ormerod mentions a paper, “On the Rate of Gain in Information,” written by William Hick and published in 1952. Hick devised a clever experiment involving semaphore lamps and corresponding Morse code keys. He discovered the time it took participants to choose the corresponding key after he lit a lamp increased exponentially as the amount of options increased. That is, if it takes ten seconds to select between two options, it won’t take one minute to select between ten options. It will take about three and a half minutes.
What happens when there are simply too many options? Google has brilliantly answered this question by relying on networks effects. Its algorithms don’t analyze the contents of each webpage. They list websites with the most inbound links, assuming, correctly, that people naturally link to the most useful and credible sources. The somewhat disturbing implication is not that the millions of people who also searched for “good Mexican food in Manhattan” helped me find what I am looking for. It’s that their searches and selections shape my preference.
There are very few aspects of life that are not influenced by networks. If you’re associated with overweight people, even indirectly, it’s more likely that you’re overweight. If someone in your community commits suicide, you’re more likely to commit suicide. James Fowler and Nicholas Christakis write that getting a $10,000 raise is less likely to make you happy than having a happy friend. If you look at the data, a $10,000 raise is actually less likely to make you happy than having a friend of a friend of a friend who is happy.
Other people regularly help us decide, even though we don’t realize it. That’s the good news. Copying is a helpful decision-making aid because it reduces time spent choosing. We cannot gather and analyze information like the standard economic model predicts, so we rely on fast and frugal heuristics, which usually means trusting the wisdom of the crowds.
Several years ago Duncan Watts and Matt Salganik famously demonstrated how this strategy changes what we prefer. They created an artificial music market comprised of 48 unknown songs from unknown bands. The 14,341 participants, recruited from a teen-interest website, were divided into two groups. Watts and Salganik gave participants in the first group the name of each track and band and asked them to rate, from one star (hate it) to five stars (love it), each song. After rating a song, they had the opportunity to download it. Participants in the second group received the same instructions but with one difference: they saw how many times a song had been downloaded.
This tiny change had huge effects. The distribution of downloads in the first group looks like a steady downward slope. The most popular song was downloaded one and a half times the average—the least only half the average. The distribution in the second group was heavily swayed in favor of a few songs on the top. For this group, the top three songs represented about 80 percent of total downloads.
The most popular song in each group was not the same. There was a relationship between songs in each condition—the best songs rarely did poorly while the worst rarely did well—yet with just one extra piece of information, the participants preferences shifted dramatically.
One implication of Watts and Salganik’s study is that winner take all effects exist in many areas of life, from the salaries of athletes to album sales. We’re living in a world where the rich get richer…. because they were just a little bit richer to begin with. The second, more interesting conclusion, is that making consumer decisions about, say, what toothpaste to buy would be much harder if we couldn’t see what other people have selected. Would you like to blindly demo songs on iTunes until you find one you like? Or would you rather read the reviews, glance at the charts, and winnow your selection down from there?
Thankfully, we’re living in the second world. We might not make the best decisions collectively, but we’re pretty good at helping each other find stuff that’s good enough—so we don’t die idling in the toothpaste aisle.