by Jon Orwant
I just spent some time discovering what's on Freenet, the peer-to-peer system that makes it possible to publish data anonymously and in such a way that it effectively cannot be removed. The software has a solid architecture, but feels like little more than the foundation of a castle. You can see what it might be someday, but at the moment it's not very useful: the installation process is clunky, and there aren't any polished clients. It's hard to find out what other nodes are out there; when you've found a node it's hard to find out what content you can retrieve from it; once you've found content you have no way to discern its quality.
Freenet has physics lectures by the late great Richard Feynman, the confessions of Jean-Jacques Rousseau, and the text of NAFTA. It also has an entire Sinead O'Connor album, the text of one O'Reilly book, kiddie porn, and pictures of naked people playing darts.
Also read Slashdot user comments about this article.
It would be unfair to infer the tastes of the average Freenet user from what's been uploaded (and more to the point, from the keys of the uploaded files -- there's no reason why something called constitution.txt can't actually be an MPEG movie of exotic clown dancing, sausages being made, or Florida polling booth webcam footage). But if we were to indulge ourselves and construct a demographic of the average Freenet user from Freenet content, he'd be a crypto-anarchist Perl hacker with a taste for the classics of literature, political screeds, 1980s pop music, Adobe software, and lots of porn.
I tabulated all the items I could find on Freenet (1075 items) and categorized them based on their key name.
but seemingly meaningful
indexes of other Freenet content
These stats tend to overreport more storage intensive media: for instance, one Ruth Rendell audio book (which was counted as audio, not as text) was split into four uploads, and so is represented by four distinct keys.
Also, several dozen items were uploaded under separate names but
almost certainly contain identical content, such as
Further observations within each medium:
19.1% were seemingly books and 80.9% were "mere" documents, although it was often hard to draw a fine line between the two.
Breaking down the text:
most of them contraband in the U.S.
including Freenet docs
There was one O'Reilly book included: the PDF of Open Sources. That was the only computer book, unless you count the Jargon file. A Harry Potter novel was there as well, as was George Orwell's 1984.
There was also information about the Florida ballot recount currently underway.
all of them Jesus Christ Superstar
Tom Lehrer, Weird Al Yankovic, and Monty Python
There were entire albums by some bands, including Genesis, Eurythmics, Alphaville, Stereolab, Information Society, and Sinead O'Connor.
Note that while Freenet music is definitely overwhelmingly rock, these by-key statistics will tend to over-report rock compared to, say, classical music, since the average rock album has more tracks than the average classical album.
The porn images include four images whose names imply that they are kiddie porn.
(And before you start snickering, let me take this opportunity to remind you that I'm categorizing these based only on key name. I haven't seen any of the images.)
an Arnold Schwarzenegger ad
The two movies were "The Matrix" and "Scary Movie."
including two C programs, and no other positively identifiable source code
Of the proprietary software:
Of the proprietary games, 81.2% were for Nintendo consoles.
All Media on Freenet
Overall porn: 15.6%
Overall sex, drugs, and rock and roll: 53.8%
Jon Orwant is a well-known member of the Perl community, founded The Perl Journal and co-authored O’Reilly’s bestseller, Programming Perl, 3rd Edition.
Return to the P2P DevCenter.