In his article in Sunday’s New York Times Magazine section (4.15.07), “Is Justin Timberlake a Product of Cumulative Advantage? A new theory of the hit record,” Duncan Watts makes two startling claims about movie, book, and music hits: they don’t really reflect what most people want (i.e., their actual preferences) and they are inherently unpredictable. This is bad news for those who pour millions into marketing products in these industries, but Watts goes further, suggesting that business products in general, from software to organic food, suffer from the same problem.
The reason prediction is supposedly hard (if not impossible) is that market success can be shown to depend not primarily on perceived quality by a majority of consumers, but rather on social influence, which quickly overwhelms such perceptions. In this age of excessive choice, people are rarely able to make up their minds about what they like on the basis of their own judgment. Instead, they allow themselves to be swayed by what other people say, i.e., the buzz. The result is that “the rich get richer.” Like the famous “butterfly effect” from chaos theory, over time, even a small, virtually random variation in initial buzz can have a huge impact, translating into the difference between a megahit and being located somewhere far down the infamous long tail of non-hits.
Watts’s claims clearly offend common sense: Was Harry Potter primarily the outcome of initial successful buzz? Could the Beatles have remained stuck playing in the Cavern Club in Liverpool but for some chance advantage in word of mouth about “I Want to Hold Your Hand”? Did the Apple iPod get over 70% market share based mostly on the word of early adopters? Watts roundly dismisses our touching belief in the idea that markets largely reflect our preferences as consisting of stories we make up afterwards to satisfy “our desire to believe in an orderly universe.” But in advancing his theory, Watts is not just ignoring our own intuitions as consumers, but network science itself.
The rich-get-richer effect Watts bases his argument on was in fact first formulated (as Watts acknowledges in Six Degrees) by Albert-László Barabási as the outcome of a law of preferential attachment: we tend to choose things that are already popular—in network terms, connect to nodes that are already highly linked. However, as Barabási points out in Linked: The New Science of Networks, the effect, while undoubtedly having some truth to it, fails in many cases, being superceded by a different effect: the fit get rich. That explains why, for example, a late-entry search-engine like Google could still trounce its competitors long after the rich-get-richer effect should have locked the market up. It was simply a much better—fitter—product.
Similarly (for details, see my Smart World, Chapter 5), the rich-get-richer effect explains the initial success of the MITS Altair as the first real microcomputer, but it was the Apple II’s superior fitness (i.e., match to emerging customer preferences) that enabled it to sweep all before it. Most real hits, whether it’s the Beatles or Harry Potter or the iPod, are the result of better market fit with consumer wants, not initial richness of buzz, and to that extent are partly predictable. Watts’s claims regarding the impact of buzz and the consequent unpredictability of hits work in markets where (as in the case of his carefully controlled experiment on producing hit records) competing products are qualitatively hard to distinguish from one another—a hardly surprising result. But plenty of market competitions aren’t like that, and then another network law comes into play.
That’s not the end of the story, by the way. We can further refine our conception of how hits happen with another effect—the fit get fitter, a sort of Darwinian law of quality. But I’ll save that for another blog…