S is for superstars

And long tails

The superstar distribution of earnings works like this:

Compare plumbers and classical violinists. In this town there are different plumbers to choose from. Some have a better reputation than others, and they will get more calls, and can probably charge something higher than other plumbers do to fix your leaky pipes, and still get plenty of work. So there is a local earnings distribution of plumbers, with the best one earning the most. If we did a graph of the earnings distribution, with the top earner on the far left and then in descending order as we move right, we get a downward sloping line (obviously), but not too steep. And that is because there is a limit on what your town’s best plumber can earn, since she can only do one job at a time. She earns more than the average plumber, but not a lot more.

Now consider Joshua Bell (also from this town, it turns out). There are different violinists in the world who people in the know would rank of different caliber, and he ranks very highly. But he has the ability to record music and video, which means he can be heard by millions all at once. Aren’t there many really fine violinists out there? Sure. But when someone wants to listen to Brahms’ Violin Concerto, they will typically seek out what is considered the best recording. Even if a Joshua Bell cd costs a few more dollars than other recordings they will buy his anyway (or stream him on Spotify or Apple or whatever), because it is just a few dollars more, and if you are going to devote valuable time to listening at all you might as well get the best. The biggest cost of consuming any “recorded” work - listening to a record, watching a movie, reading a book - or a live broadcast of a concert or a soccer game, is the cost of your time, not the purchase price. As a result, the distribution of earnings for the world’s violinists does not look like the earnings of plumbers at all: it starts with a small number of people earning millions of dollars, and then very steeply declines, with a “long tail” of violinists earning very little from their music performance, and probably having to supplement their income through teaching and other jobs.

We get superstar earnings distributions when it is possible for those who are generally considered the best to sell to millions at a time. Athletes in the top leagues in their sports earn millions, while those in just the next division down (AAA Baseball; Championship League in English Football) earn much less, and this is so even though the differences in talent might be quite small.¹ But people will choose to watch the top leagues on television, so that’s where the advertising and sponsorship revenues are, and the top players can capture some of that in salary negotiations.

---

Chris Anderson published the essay “The Long Tail” in Wired in 2004; it is paywalled, but I’ll try to give the gist. Until the adoption of the internet, the arts had been dominated by the superstar distribution of earnings, especially where recording and live broadcast had been the norm (we would expect less of a superstar effect in ballet, for example, which relies much more on in-person performance in small (i.e. not football stadium-sized) venues). But, said Anderson, the internet would allow niche artists - bands with at least some following, authors with a dedicated but small group of followers - to reach their widespread audience, and be able to earn some decent, though not extravagant, income from that. The very platform you are reading right now caters to that idea. There was a living to be made in the long tail of the earnings distribution.

The essay is mostly famous for getting it wrong. We now have a more skewed distribution of earnings in the arts, and life remains hard in that long tail. It turns out that as consumers, we are still interested in what the bestselling books are, the ones getting reviewed in the elite spaces, and in seeing the movies that have some buzz, and radio aimed at teens still relies on short playlists with known singers. This isn’t completely irrational: what is popular is a signal to us that since so many people find this music / movie / novel worth their time, there must be at least something interesting there (even if we end up a bit disappointed). And sometimes, as with sports for example, the top leagues are indisputably where to find the top talents.

We might have been fooled about the long tail by algorithms used by Amazon, Spotify and the like: “if you liked that book you might also like this.” And I can attest that sometimes the algorithms are benign, and lead me to a band or a book or a movie that I otherwise wouldn’t have heard of and I think “this is actually pretty good: thanks algorithm, for steering me to this niche title!” The problem is that the algorithm is not my personal cultural advisor - it is giving the same sorts of referrals to lots of other people. And so our consumption remains concentrated.

I don’t blame Anderson: when I first read the essay twenty years ago I thought he had a great insight. Oh well.

---

And so today we get lots of stories about struggling artists - “a thousand people listened to my song and Spotify gave me twelve cents”, or, more generally, “the median artist/writer/composer/musician/actor earns well under the poverty wage.” What can be done?

And here I’m afraid I have to put on my dismal science hat: not much can be done.

The superstar distribution is the result of technologies specific to particular sectors: the arts, professional (including college) sports, and, more recently, invention in technology, and management of large enterprises. It does not appear where the worker, even if highly skilled and decently paid, can for the most part only serve a limited number, or even just one, customer at one time - plumbers, dentists, masseuses.

Why are there so many struggling artists - why don’t they just pack it in and go work for a bank or something? Because if they “make it”, the payoff is huge: big income plus doing something you really like doing.² The equilibrium is then a large supply of people hoping for their break, and putting up with low earnings in the meantime. The empathetic response of “well, couldn’t we have an arts policy that helps these struggling artists out a bit?” would only draw even more people into this lottery. But you can’t magically increase what people are willing to pay for cultural products when they have lots of other things they still need to buy, you can’t get people to stop wanting to have the best / the most popular / what all their friends are reading, and you can’t stop writers and singers and actors from trying to make it big. And this means there is no arts policy fix to this very skewed distribution of earnings.

1. In some cases the difference in talent might be non-existent, say when a pop singer gets a lucky break, and some buzz, which then snowballs, even though there were singers of equal talent who did not get that break. This piece by Duncan Watts makes the claim. This is less likely where there are more objective measures of differences in talent, say in sport, or, I would venture, in classical music performance.

2. In the economic development literature, this has been the model for why in poor countries you get so many people moving from the country to the city, only to end up unemployed there. They might be very poor in the city, but if they can land one of those scarce jobs, it is far better than what they had back on the farm.