This morning I woke up to find that the torrent had died. Someone - no one knows who - had put enough pressure onto the operators of Suprnova.org and TorrentBits.com to shut them down. SuprNova.org was amazing, the Wal-Mart of torrents, a great big marketplace of piracy, all neatly dished up and aiming to please. You want this new Hollywood release? Here's a recording from someone who smuggled a camcorder into a screening. - How about the latest episode of that hit HBO series? There you go, and no subscription fees to pay. Just fire up your favorite BitTorrent client - BitTornado, Azureus, Tomato, or that good old-fashioned Bram Cohen code. Click on the torrent, and you're up and downloading, sharing what you're getting with hundreds of others. Share and share alike. What could be more friendly? …
If you're just one person with one recording of one show, and it's a popular show, your computer's internet connection is going to get swamped with requests for the show; eventually your computer will crash or you'll take the show off the Internet, just so you can read your email. And in the early days of peer-to-peer, that's how it was. Someone would find a computer with a copy of the song they wanted to listen to, connect to that computer, and download the data. It worked, but anything that got very popular was likely to disappear almost immediately. Popularity was a problem in first-generation peer-to-peer networks.
In November 2002, an unemployed programmer named Bram Cohen decided there had to be a better way, so he spent a few weeks writing an improved version of the protocols used to create peer-to-peer networks, and came up with BitTorrent. BitTorrent is a radical advance over the peer-to-peer systems which preceded it. Cohen realized that popularity is a good thing, and designed BitTorrent to take advantage of it. When a file (movie, music, computer program, it's all just bits) is published on BitTorrent, everyone who wants the file is required to share what they have with everyone else. As you're downloading the file, those parts you've already downloaded are available to other people looking to download the file. This means that you're not just "leeching" the file, taking without giving back; you're also sharing the file with anyone else who wants it. As more people download the file, they offer up what they've downloaded, and so on. As this process rolls on, there are always more and more computers to download the file from. If a file gets very popular, you might be getting bits of it from hundreds of different computers, all over the Internet - simultaneously. This is a very important point, because it means that as BitTorrent files grow in popularity, they become progressively faster to download. Popularity isn't a scourge in BitTorrent - it's a blessing.
It's such a blessing that, as of November, 35% of all traffic on the Internet was BitTorrent-related. Unfortunately, that blessing looks more like a curse if you're the head of a Hollywood studio, trying to fill seats in megaplexes or move millions of units of your latest DVDs releases. And, although BitTorrent is efficient, it isn't designed to make data piracy easy; BitTorrent relies on a lot of information which can be used to trace the location of every single user downloading a file, and, more significantly, it also relies on a centralized "tracker" - a computer program which registers the requests for the file, and tells a requester how to hook up to the tens or hundreds of other computers offering pieces of the file for download.
As any good network engineer knows (and I was a network engineer for over a decade), a single point of failure (a single computer offering a single torrent tracker) is a Bad Thing to have in a network. It's the one shortcoming in Cohen's design for BitTorrent: kill the tracker and you've killed the torrent. But network engineers know better than to design systems with single points of failure: that's one of the reasons the Internet is still around, despite the best efforts of hackers around the world to kill it. Failure in any one part of the Internet is expected and dealt with in short order. Various parts of the Internet fail all the time and you only very rarely notice.
Back to today, when the hammer came down. SuprNova.org and TorrentBits.com each played host to thousands of BitTorrent trackers. When these sites went down the torrents went Poof!, as if they'd never existed. This evening the members of the MPAA must be feeling quite satisfied with themselves - they see this danger as passed; never again will BitTorrent threaten the revenues of the Hollywood studios. Nothing could be further from the truth.
As Hollywood is so fond of sequels, it seems perfectly fitting that today's suppression of the leading BitTorrent sites bears an uncanny resemblance to an event which took place in July of 2000. Facing a rising sea of lawsuits and numerous court orders demanding an immediate shutdown, the archetypal peer-to-peer service, Napster, pulled the plug on its own servers, silencing the millions of users who used the service as a central exchange to locate songs to download. That should have been the end of that. But it wasn't. Instead, the number of songs traded on the Internet today dwarfs the number traded in Napster's heyday. The suppression of Napster led to a profusion of alternatives - Gnutella, Kazaa, and BitTorrent.
Gnutella is a particularly telling example of how the suppression of a seductive technology (and peer-to-peer file trading is very seductive - ask anyone who's done it) only results in an improved technology taking its place. Instead of relying on a centralized server - a fault that both Napster and BitTorrent share - Gnutella uses a process of discovery to let peers share information with each other about what's available where. The peers in a Gnutella peer-to-peer network self-organize into an occasionally unreliable but undeniably expansive network of content. Because of its distributed nature, shutting down any one Gnutella peer has only a very limited effect on the overall network. One individual's collection of music might evaporate, but there are still tens of thousands of others to pick from. This network of Gnutella peers (and its offspring, such as Kazaa, BearShare, and Acquisition) has been growing since its introduction in 2001, mostly invisibly, but ever more pervasively.
If Napster hadn't been run out of business by the RIAA, it's unlikely that any need for Gnutella would have arisen; if the RIAA hadn't attacked that single point of failure, there'd have been no need to develop a solution which, by design, has no single point to failure. It's as though both sides in the war over piracy and file sharing are engaged in an evolutionary struggle: every time one side comes up with a new strategy, the other side evolves a response to it. This isn't just a cat-and-mouse game; each attack by the RIAA, generates a response of increasing sophistication. And, today, the MPAA has blundered into this arms race. This was, as will soon be seen, a Very Bad Idea.
Pointing up the single greatest weakness of BitTorrent take down the tracker and the torrent dies - has only served to energize, inspire and mobilize the resources of an entire global ecology of software developers, network engineers and hackers-at-large who want nothing so much, at this moment, as to make the MPAA pay for their insolence. Imagine a parent reaching into a child's room and ripping a TV set out of the wall while the child is watching it. That child would feel anger and begin plotting his revenge. And that scene has been multiplied at least hundred thousand times today, all around the world. It is quite likely that, as I type these words, somewhere in the world a roomful of college CS students, fueled by coke and pizza and righteous indignation, are banging out some code which will fix the inherent weakness of BitTorrent - removing the need for a single tracker. If they're smart enough, they'll work out a system of dynamic trackers, which could quickly pass control back and forth among a cloud of peers, so that no one peer holds the hot potato long enough to be noticed. They'll take the best of Gnutella and cross-breed it with the best of BitTorrent. And that will be the MPAA's worst nightmare.
Hey, Hollywood! Can you feel the future slipping through your fingers? Do you understand how badly you've screwed up? You took a perfectly serviceable situation - a nice, centralized system for the distribution of media, and, through your own greed and shortsightedness, are giving birth to a system of digital distribution that you'll never, ever be able to defeat. In your avarice and arrogance you ignored the obvious: you should have cut a deal with SuprNova.org. In partnership you could have found a way to manage the disruptive change that's already well underway. Instead, you have repeated the mistakes made by the recording industry, chapter and verse. And thus you have spelled your own doom.
It's said that the best sequels are just like the original, only bigger and louder. Ladies and gentlemen, prepare yourselves for one hell of a crash. This baby is now fully out of control.