The Long Tail 8 - Long Tail Economics

blog archive
Author

Tom Slee

Published

March 11, 2007

Note

This page has been migrated from an earlier version of this site. Links and images may be broken.

This is another part of my critical reader’s companion to The Long Tail, and it discusses Chapter 8 - Long Tail Economics. Part 0 is here. You can find a complete list of the Long Tail pieces here.


By this time in the book we’ve seen variants on the long tail graph a lot of times.

The chapter is a collection of observations about this shape. I don’t have a whole lot to say about this chapter, so this is a short post.

The chapter starts by talking about the Pareto 80/20 rule (that 20% of the people in populations Pareto studied owned 80% of the wealth) and Zipf’s Law (that the frequencies of use of words in the language follow a similar kind of fall-off). These and others are examples of power law distributions that crop up in a bunch of different places. Power laws are called long-tailed curves because “the amplitude … approaches but never reaches zero as the curve stretches out to infinity” [126]. This is true of many kinds of distributions, of course, including exponential and log-normal distributions. But even within the family of power-law curves, of the form

y = ax-k

The area under the long tail depends on the value of the exponent k. A power law distribution itself is not enough to ensure a significant long tail. It’s only if the exponent is a small number (close to 1) that the “tail” is “fat” as John says - has a significant area under the curve.

How Distribution Bottlenecks Distort Markets [127-130]. If you plot a power law on a log-log scale you get a straight line with slope -k. Anderson looks at some markets (movies on movie screens) and notices that they fall off below the line at a small value. He interprets this as a failure of markets to meet demand because of the cost structure of movie theatres. If there’s not enough demand to fill a theatre, films won’t get distribution. There’s some truth to this - there are costs to producing and distributing movies that demand a return on investment, and movies that don’t make the cut will not be distributed at all. In non-theatrical channels, where cost is less, more movies are available and watched. As the title of the section suggests, Anderson sees only the untruncated power law as somehow “natural” [127] and a reflection of consumer demand, and the truncated ones a measure of artificial scarcity caused by distribution bottlenecks, to be removed in the world of abundance. Yet even in the online world there are truncated distributions, they just occur at different places in the distribution chain. The small number of search engines or online auction houses are two examples. The idea that consumer demand is “naturally” a power law shape is unsubstantiated, at least in the pages of the book.

The strangeness of the comparison between theatres and rentals is exemplified by this sentence: “demand keeps on going into niches that were never even considered before – instructional videos, karaoke, Turkish TV, you name it” [130]. I don’t know what he means by “never even considered before” but I’m sure the reason instructional videos were not made available on movie theatre screens was not so much “an artifact of the traditional costs of offering them” but the fact that the number of people who want to watch instructional videos _in movie theatres _is probably vanishingly small no matter how large a geography you spread them over. It’s like comparing the market for banquets to the market for snacks - they’re both food, but the demand pattern is different for one than for the other, and the limited demand for banquets is not “unnatural”.

This section is one of those that come occasionally in the book, containing a scattered set of only loosely related data points, with sentences like “In books, Barnes and Noble found that the bottom 1.2 million titles represent just 1.7 percent of its in-store sales, but a full 10 percent of its online (bn.com) sales. PRX, which licenses a huge library of public radio programming online, reports that the bottom 80 percent of its content now accounts for half of its sales.”[130] The “statistics” come from “direct personal correspondence with their executives”. Is this trustworthy data or is it picked to illustrate a pre-conceived idea? What does PRX do and what is the nature of its business? Who knows - that sentence is the sole reference to them in the book. How does the Barnes and Noble statement fit with Lee Gomes’ claim in the Wall Street Journal that “The head of a major New York publishing operation says that the distribution of his titles is essentially the same in both online and”bricks and mortar” channels.” These isolated, context-free sentences from executives seem to be the best we can do – actual company data being tightly-held secrets.

I have little to say about the remainder of the chapter, The 80/20 Rule [130-135] is a meandering take on whether or not the Rule still applies. The other sections are Does a Longer Tail Mean a Shorter Head? [135-137], Does The Long Tail Increase Demand or Just Shift It? [137-138], Should Prices Rise or Fall Down the Long Tail? [133-139], “Microstructure” in the Long Tail? [139-142], The Long Tail of Time [142-143], and The Tragically Neglected Economics of Abundance [143-146]. As the titles suggest, this is a miscellany of observations about the shape of the graph. There’s not a whole lot to disagree with, but there’s not a whole lot of substance.

I could stop here… or I could just throw in a few things that irritate me, just to be grouchy. So don’t take this too seriously, but:

In the section on “Microstructure” [140] Anderson shows a graph of the average sales for various of the many subgenres that Rhapsody divides music into and claims that this selection shows “a Long Tail”. Yet it doesn’t. The graph does fall off left to right, as it cannot help but do, but it is as close to a linear fall-off as a power law.

“One of the features of powerlaws is that they are ‘fractal’, which is to say that no matter how far you zoom in they still look like powerlaws” [138] Wrong. Straight lines still look like straight lines as you zoom in too, but that’s not enough to make a fractal. A fractal reveals new structure as you zoom in, and that structure is similar to the structure seen from further away.

“Einstein described time as the fourth dimension of space, you can think of it equally as the fourth dimension of the Long Tail” [142]. Wrong. It’s not the “fourth dimension of space”, and the way he plots it is actually a third dimension of the Long Tail graph. Plus, it’s name dropping. This section, by the way, is devoted to the realization that “If you think about it, today’s hit is tomorrow’s niche”, which is hardly shocking, but “Both hits and niches see their sales slow over time; hits may start higher, but they all end up down the Tail eventually. The research to quantify this conclusion is continuing” [142] Count me underwhelmed.

On abundance: “Abundance, like growth itself, is a force that is changing our world in ways that we experience every day, whether we have an equation to describe it or not” [146]. This is stuff that should have been thrown out during editing.

That’s all. More tomorrow, I hope.