The Excellence Market

Sociologist Duncan J. Watts, in his book Everything Is Obvious (Once You Know the Answer), does something economists once agreed was impossible: he recreates the free market under laboratory conditions. Watts, a lead researcher for Microsoft Research, uses a little-known Amazon service, the Mechanical Turk, to reproduce the music market in exact detail six times. This tests whether markets really reward the best competitors in each field.

Watts describes how he recreates the market under blind conditions. Members of each music market had no communication with users in other markets. They had limited resources to purchase a selection of download files from independent and unsigned artists, mimicking real-world market scarcity. And because the Mechanical Turk service pays participants pennies per momentary service provided, it was cost-effective to recruit thousands of participants, recreating realistic market scale.

This experiment produced, according to Watts, six very different market outcomes. No two music markets produced equal hit parades. Watts writes that “the very best songs never did terribly, while the very worst songs never actually won.” Nevertheless, a song that tanked on five charts could do modestly well (without ever overwhelming anybody) on the sixth. The marked inconsistency of outcomes provides an important challenge to received market economics.

(I should note that, though Watts describes this experiment in his book, he did not perform the experiment solo. The original research outcomes, published in Science magazine in early 2011, were co-authored by Matthew J. Salganik and Peter Sheridan Dodds, both of Columbia University. Unfortunately, in experimental science as elsewhere in culture, an idea’s spokesperson often receives more credit than that idea’s originator.)

Duncan J. Watts

Basically, Watts demonstrates that market outcomes aren’t a function solely of merit. Songs didn’t succeed exclusively because they were good. Though lousy songs, as determined by consistent chart performance, never received more than a muted reception, and usually failed, hit status didn’t automatically accrue to good music. To put it another way, if you suck, you’ll probably fail, but if you’re awesome, success is contingent, and never guaranteed.

For long-term outcomes, the principal outcome modifier apparently wasn’t individual merit, but word-of-mouth scuttlebutt. In markets where participants communicated openly, sales would initially travel among people sharing their stories. Later, when songs moved enough units to chart, that measurable success produced even more success: people wanted to hear what they believed their friends and peers were hearing. First friends wanted to share experiences, then success came to justify more success.

This was echoed in market conditions where participants didn’t communicate. Where people didn’t know what peers and friends already liked, they didn’t feel compelled to spend scarce resources on music they didn’t know. They tended to reproduce their individual comfort zones branching out little. They spent less money, enjoyed what they had, but also felt disconnected from the market. Charts, under these conditions, tended to fizzle rather than soar.

We should stress here that Watts et al. didn’t reproduce an exact real-world analog here. We can use his example to speculate informedly on runaway successes, from Justin Bieber to Coca-Cola to Donald Trump. However, we cannot rerun real-world markets under laboratory conditions, controlling for just one variable. In the non-laboratory world, events simply happen, and we can at best conjecture. Watts’ results let us, at best, improve our spitballing.

Within that stipulation, though, we see something important happening. People who communicate about the things they love, about art or food or politics, show a clustering tendency. Political candidates who create something for voters to talk about, like Bernie Sanders or Donald Trump, naturally create buzz, and therefore following. Candidates who talk policy, like Martin O’Malley or Jim Gilmore, fizzle quickly and vanish, because nobody talks about them.

Just as important, discussion drives outcome—and outcome drives discussion. Committed players who talk good games dominate social consequences. And players who talk poorly, but can purchase their way into discussions (see Citizens United), can also dominate outcomes. That’s why music companies buy saturation advertizing for mediocre artists. It’s also why artists, politicians, and manufacturers who tell us what we already think we know, succeed even when they’re boring.

This also raises an important question: what ideas does our talk advance? Are we building up discussions that improve our shared outcomes? Simply tearing down what we oppose actually reinforces what we oppose, because it keeps these undesirable outcomes centered in our discussion. The “Lesser of Two Evils” approach rewards evil. Because in the marketplace of ideas, mere merit isn’t good enough. Real-world success is more nuanced than that.