Porn-to-Porn Networks?

I'm sure I've posted on this topic before, but I see the GAO is telling Congress [PDF] that "users of peer-to-peer networks can readily access child pornography." Now, I'm sure there's some floating around out there on the p2p networks—I'd be shocked if there wasn't, frankly. But I'm more than a little sceptical when I read this:

In one search, using 12 keywords known to be associated with child pornography on the Internet, GAO identified 1,286 titles and file names, determining that 543 (about 42 percent) were associated with child pornography images.

And then read how they went about it:

Because child pornography cannot be accessed legally other than by law enforcement agencies, we relied on Customs to download and analyze image files. We performed analyses based on titles and file names only…. Our analysis of 1,286 titles and file names identified through KaZaA searches on 12 keywords showed that 543 (about 42 percent) of the images had titles and file names associated with child pornography images.

So, as far as I can tell, we've got an estimate largely based on an analysis of text descriptions, presumably meaning that any time a file purported to contain images of "young teens" or "schoolgirls" some such thing, that was taken at face value. Then there's a smaller sample that a Customs agent examined visually, and "determined" that a comparable proportion were child porn. I'm curious: How? Pigtails? Petiteness? Plaid skirts? Counting the rings?

The report then says that:

Juvenile users of peer-to-peer networks face a significant risk of inadvertent exposure to pornography when searching and downloading images.

Again, what they seem to mean is that (as with Google) searching for an innocuous term may yield file descriptions of pornographic images. I don't know how you'd go about inadvertently downloading them. But this will doubtless serve as fodder for the draconian Protecting Children from Peer-to-Peer Pornography Act.