The Internet is changing the entertainment business from one that is driven by hits to one that will make most of its money from misses. This is good news for consumers because it means more choice, and we all like things that will never make the bestseller lists for CDs, books or movies. And although it might sound strange, this "new economics of abundance" is already the basis of the Net's most successful companies, such as Amazon, eBay and Google. Internet start-ups are now asking one another: "Are you long tail?"
The phrase was popularized by Wired magazine editor in chief Chris Anderson, who started the ball rolling last year with a series of talks and a long article called "The Long Tail." That caught the imagination of bloggers, who circulated and applied the meme. "The day I knew I was on to something was when people saw the applicability in places I hadn't anticipated," says Anderson. "It's become, in some sense, an open-source meme. They're doing a lot of research and feeding it back to me. That's why I started the Long Tail blog and, unbelievably, it works."
The long tail is named after the type of power law curve you get when you plot the sales of CDs, computer games and other products, or the popularity of Web sites, or the frequency of word use in a language. This is where George Kingsley Zipf, professor of linguistics at Harvard, observed it.
To put it crudely, there are a small number of words that appear very frequently: the, of, to, and, a, in, etc. After that, there is a steep decline, followed by tens of thousands of words that appear relatively rarely: palimpsest, lapidary, rodomontade, epalpebrate and so on. When you graph what is now often called the Zipf distribution, these rare words form the long tail that tapers off to the right.
The Zipf curve turns out to be at least as common on the Internet as it is elsewhere. You might suppose, for example, that since all Web sites are more or less equally accessible, visitors would be evenly distributed. That is not the case. Research by Lada Adamic and Bernardo Huberman showed that, in fact, "the distribution of visitors per site follows a universal power law, characteristic of winner-take-all markets." In other words, a small number of sites -- Google, Yahoo, MSN, etc. -- have tens of millions of visitors each month, while millions of sites attract only a handful each. Counterintuitively, the more choices there are, the more extreme the curve becomes. Life isn't fair. This knocks most naive ideas of competition on the head.
What interested Anderson, however, was what happens when something moves from the physical world of atoms to the digital world of bits. That's where the "long tail" effect starts to make a real impact. Anderson points out that in the real world, there are physical limits on how many titles a shop can stock or a cinema can show. These are the economics of shortages. For example, an average American movie theater can only show films that will attract 1,500 people over two weeks, and almost all of those will live within a 10-mile radius. It is hardly surprising that relatively few movies ever get a showing, and cinema distribution has almost no tail at all: You go straight from hero to zero.
It is not that cinemas wouldn't like to serve these markets, just that they have no economic way of doing so. Sell movies on DVD, however, and the economics are different. A movie that wouldn't fill a local theater can still sell 1,500 copies across America. And instead of making a couple of hundred movies available, like a cinema, a typical Blockbuster can stock 3,000 titles. Most sales will still go to the big Hollywood hits, but at least we can see the start of a tail.
Sell movies online and the equation changes again. According to Anderson's figures, for example, Netflix offers 25,000 DVDs. The 24,000 or so titles in the tail may not sell many copies each, but they add up. (Most of Amazon's sales are of titles the average bookstore just doesn't have room to stock.)
With the move to digital downloads, however, Anderson argues that the misses can be as important as the hits. "With no shelf space to pay for, no manufacturing costs and hardly any distribution fees, a miss sold is just another sale, with the same margins as a hit. Suddenly, popularity no longer has a monopoly on profitability," he wrote in Wired.
Where a retailer might offer, say, 30,000 songs on CD, an online music store can offer a million. The hits will still sell far more than the misses, but since almost every download finds someone willing to give it a try, the long tail could provide the majority of sales. "Now we could look to the right [of the graph], which could be as economically valuable as the left," says Anderson.
The entertainment industries provide an accessible illustration of the long tail, and Anderson reckons these will take up about half his forthcoming book. However, he also cites Google as a long-tail business, because of its contextual advertising. The first big Internet sites aimed to attract millions of users so they could sell banner ads to major brands. Google's AdWords, by contrast, are so cheap that almost anyone can buy them, and they can be targeted at a handful of users. Google's financial success therefore comes from aggregating lots of small advertisements shown to small audiences -- a long-tail effect.
EBay is another long-tail business. It is not about auctioning a few old masters for 20 million pounds apiece; it's about providing a market where huge numbers of people can sell almost anything for a couple of quid.
None of this means Hollywood studios will stop trying to make hit movies and make misses instead. However, Anderson refers to the long tail as "a million billion niches," and niche products can be profitable, if they can find their audience -- or vice versa.
Mass-market advertising is clearly out of the question for niche products, but Internet retailers have already developed ways to make them visible, particularly Amazon. One example is the social software that does "collaborative filtering." Other people who bought X also bought Y: Your taste may be similar. Getting buyers to write reviews and create their own hit lists and wish lists also helps buyers to find people with similar tastes, and perhaps to try their recommendations.
But the process only works if a Web site offers a full range of products. "Hits still matter," says Anderson, because they are needed to attract mainstream buyers. After that, recommendations can encourage them to try niche products, driving them down the long tail. Offering only hits, of course, does nothing to expand the market and change consumer behavior.
Another question is: "What does the tail do to the shape of the head?" This is one of Anderson's research projects. At the moment, on the Pareto principle, the top 20 percent of titles probably bring in 80 percent of the revenues. Will there be a leveling of the graph if more demand moves to the long tail? Anderson thinks so.
Finally, there's an apparent conflict between the economics of the long tail and the progress made over the past few years by Creative Commons, a copyright reform movement started by Lawrence Lessig, a Stanford University law professor (and Wired columnist). This has been driven by the argument that, for example, out-of-print books are not worth anything, so copyright owners might as well give them to the community for reuse.
The long tail argument, in contrast, suggests that copyright owners should make everything available online, as cheaply as possible, because it will still find buyers, and the aggregate sales could be significant.
Anderson tackles this problem on his Long Tail blog, where he points out that the people who usually find new commercial outlets for the back catalog are the ones who repurpose and/or remix it, not the original copyright holders.
With both Anderson and Lessig speaking at the O'Reilly Emerging Technology 2005 conference in San Diego last week, it would have been interesting to hear them discuss the problem. Alas, it didn't happen.
However, the issues are closely related. At the moment, the major film studios and record companies are pricing their online, digital products (if any) to defend their retail sales, not to exploit the low cost of doing business over the Net. The obvious and perhaps inevitable result is rampant piracy.
At ETech, Lessig said the big copyright holders were trying to blind the technology to conform to 18th century law, rather than reforming the law. Maybe they'll read Anderson's long-tail book, when it comes out, and see the light. If Anderson is right about a change from the old economics of shortages to a new economics of abundance, will they have a choice?
Shares