31 October 2003

Amazon PR Bullshit
This is starting to look like a bad, bad 1980s movie about corporate greed. Possibly one starring Pauley Shore. Yes, that bad.

   We'll start with this gem, as reported on Publisher's Lunch (a newsletter on the publishing industry):

Despite all the ink given to those authors who have qualms about the program, Amazon’s vp of the North American media group says only 15 authors have asked to have their books removed program. As indicated informally in our previous report, Kessel says, "We're definitely cooperating with that."

Let's just hope that this quotation came from Monday. If it came as late as Tuesday, it's an outright lie, in two respects. First, I am personally aware of more than 15 authors who had asked for removal by the close of business on Tuesday, and I'm certain there are others I don't know about. Second, the "definite cooperation" involves violating 17 U.S.C. § 512(c)(2), among other problems. Amazon is basically insisting that authors go through their publishers to get material removed; cutting through such bullshit is exactly what § 512 was supposed to do.

   Next, we'll go into the realm of scientific implausibility. Again quoting from Publisher's Lunch, although I've seen variants of this particular assertion in several other sources:

Amazon dropped a press release yesterday saying that in the first five days of their Search Inside the Book program, titles in the program showed sales growth of 9 percent versus non-participating titles. The e- tailer says that another 37 publishers have contacted them asking to participate in the program. Jeff Bezos says, "We're truly excited by the positive response from customers."

Can you say "inadequate statistical sample?" See? I knew you could. Even if the sample was adequate, it does not account for multiple alternative explanations—such as the spike effect that any new feature at any Internet retailer has on sales. Look at a full month's sales in, say, February, By then, the statistical sample will be scientifically valid, and the newness variable will no longer complicate analysis. If at that time there is a difference—given that participation in the program is self-selecting in a way that skews the analysis to start with—I suspect that comparative sales growth will be under 3%, or in other words less than or equal to the inherent margin of error from this kind of analysis! The math isn't all that hard to do, particularly computer-assisted; it's just so seldom done by people who should know better.

   Of course, the real problem is that trumpeting the purported successes of the program does nothing to address the needless harms or the authors' individual rights. Some proponents made the same assertions that Napster et al. would result in increased sales of recorded music. As it happens, I do not trust recording industry accounting any more than I trust GAO contract analysis (which is to say not at all). However, there is little or no evidence of an overall increase in sales for any artist who did not specifically authorize release of MP3s.