Wednesday, March 31, 2004

Don't Download This Post! It's Available In Stores For Only $16.99!

The New Scientist reports that file-sharing does not appear to harm legitimate CD sales.
Oberholzer-Gee and Strumpf monitored 680 albums, chosen from a range of musical genres, downloaded over 17 weeks in the second half of 2002. They used computer programs to automatically monitor downloads and compared this data to changes in album sales over the same period to see if a link could be established.

The most heavily downloaded songs showed no decrease in CD sales as a result of increasing downloads. In fact, albums that sold more than 600,000 copies during this period appeared to sell better when downloaded more heavily. For these albums each increase of 150 downloads corresponded to another legitimate album sale.
It's nice to see that people are actually attempting to study this scientifically, rather than relying on ideological arguments from both sides, but I'm not sure this study really does the trick. I haven't read the actual paper, but it seems to me that the RIAA isn't going to be satisfied with the claim that "no decrease in sales" is equivalent to "no economic damage." They want sales to increase (especially sales of these top albums), and they think that downloading slows the rate of increase in sales figures. That's hard to prove or disprove. This study shows a tiny increase in sales attributable to increased downloads; what's to prevent the RIAA from arguing that the increase in sales would've been much steeper without downloads?

I expect to see that "increase of 150 downloads corresponded to another legitimate album sale" translated, in RIAA pressure materials, to "150 people downloaded our song for every one who bought it, so instead of a zillion dollars in profit we should've made $150 zillion. Downloaders cost us $149 zillion." Which would be an utterly ridiculous claim, of course, but it's not as though that's stopped them before. Look at the studies they claim support their position:
For example, a series of surveys conducted by Houston-based company Voter Consumer Research have indicated that those who download more songs illegally are less likely to buy music from legitimate retailers.
The first thing you learn in an introductory statistics or research methods class is that correlation does not imply causation. (Then you spend the rest of your life railing, futilely, against people who use correlational research to support a causal argument - none of whom seem to have taken an introductory statistics or research methods class.)

Heavy downloaders buy fewer albums, perhaps, but that doesn't prove that they'd buy more albums if they weren't downloading. Maybe they don't have much disposable income, and if they didn't download songs they'd borrow CDs from their friends or check them out of the library. Maybe they're really hard to please, and have to download an awful lot of music they don't like in order to find the one thing they do want to buy. Maybe they don't even like music, and only download songs in a spiteful effort to bring RIAA to their knees. (Okay, maybe not. I just couldn't resist.) The point is that the data is equally supportive of all of these interpretations. Correlation - two things tending to happen together - does not prove that one of those things causes the other.

So, lots of credit to Oberholzer-Gee and Strumpf for what looks like a fairly impressive effort to bring actual data into the hysterical debate over file-sharing. I don't think this is the study that ends the debate, but it's miles better than what we've been offered so far.