Can you persuade people to like a product by telling them that it's popular? Do teenagers like Taylor Swift because she's good or because everyone else they know likes her, so, hey, she must be good, right?
Sociologist Robert Merton dubbed this tendency to base what we think on what other people are doing the "self-fulfilling prophecy" in 1949, and since then social scientists have tried to measure how powerful it actually is. Now, based on some studies conducted with the help of the internet, it seems clear that we're often just sheep.
A few years ago, Duncan Watts, a network-theory pioneer and scientist at Yahoo and Columbia University, wanted to test the strength of self-fulfilling prophecies in pop culture. The problem, he realized, was that to really explore the phenomenon you'd have to rewind history. For example, I could argue that Madonna is famous because she's uniquely talented. You could counterargue that she's just lucky; she got picked up by the right label at the right time, and enough people glommed onto her. But what if you could replay history with different conditions? If Madonna becomes famous each time, then her success is due to raw talent. If not, it's just luck.
You can't rewind history, of course. But Watts devised a clever way to simulate the effect. He and his collaborator, Matthew Salganik, created a music-downloading website. They uploaded 48 songs by unknown bands and got people to log on to the site, listen to the songs, then rate and download them. Users could see one another's rankings, and they were influenced in roughly the same way self-fulfilling prophecies are supposed to work. That meant some tunes could become hits—and others duds—partly because of social pressure.
Watts and Salganik ran the experiment over and over, each time with a new group of people, until they'd gotten 12,900 participants. In essence, they rewound history each time. Every new group started fresh, listened to the same 48 songs, and made up their collective mind.
The result? Different songs were hits with different groups. A few songs frequently, but not always, hovered near the top, and a few at the bottom. But for most of the tracks, success—or failure—seemed random.
Or as Watts concluded, about half of a song's movement could be attributed to intrinsic appeal. The rest was luck. Rerun history, and Madonna, it seems, could be working as a waitress.
So what about advertising, marketing and hucksterism? Can you browbeat people into thinking something is popular when it isn't?
To figure that out, Watts and Salganik ran a deliciously devious experiment. They took the song ratings of one group and inverted them so bottom-ranked music was now at the top. Then they gave these rankings to a fresh set of listeners. In essence, they lied to the new group; they told them that songs that weren't popular with previous listeners actually were.
The new listeners dutifully took their social cues from the bogus popularity rankings—they ranked the fake high ones high, even downloading them, while snubbing the fake low ones.
Apparently, flat-out lying works.
But only sometimes. Eventually, some of the previously top-ranked songs began to creep back up, and previously bottom-ranked ones slid down. And people in the upside-down world downloaded fewer songs overall.
Maybe the participants sensed that the ratings somehow weren't accurate and started to wonder about the entire system. If so, this strikes a small but happy blow for quality. It also offers a cautionary tale to marketers: If you lie about the merits of your product, you might suppress demand across your entire sector.
It will be exciting to see how this "multiple worlds" paradigm gets used to test self-fulfilling prophesies in other areas—like financial markets. Learning how to avoid a bank run or currency collapse could come in handy these days.
Clive Thompson is a contributing writer at the 'New York Times' magazine. In 2002, he was a Knight Science-Journalism Fellow at MIT.
Open Mic is a weekly feature in the Bohemian. We welcome your contribution. To have your topical essay of 700 words considered for publication, write email@example.com.