I have a new obsession hobby. This past weekend, I discovered BitTorrent, a cooperative distribution protocol for sharing large files. In my case, it’s a way to get concert recordings (of artist who approve of that sort of thing) without having to set up a snail mail trade with someone and hope it works out for the best. It’s fast way to distribute shows, too. I was able to download a show yesterday that was recorded the day before!
The key to scalable and robust distribution is cooperation. With BitTorrent, those who get your file tap into their upload capacity to give the file to others at the same time. Those that provide the most to others get the best treatment in return. (“Give and ye shall receive!”)
Cooperative distribution can grow almost without limit, because each new participant brings not only demand, but also supply. Instead of a vicious cycle, popularity creates a virtuous circle. And because each new participant brings new resources to the distribution, you get limitless scalability for a nearly fixed cost.
So, what’s the library world application of this kind of protocol? The obvious one that comes to my mind is distribution of electronic publications. If part of the difficulties of publishing electronically is the cost of maintaining the backfiles, then perhaps publisher should look into a BitTorrent-type distribution model where everyone who is interested in the electronic content shares the burden of distributing it on demand.
I’m not sure if this would work in libraryland, but it certainly makes it a lot easier to share concert recordings.
There’s a problem with the current model of BitTorrent as it would apply to a large database of files for which there would exist only occasional demand. BitTorrent works by sub-dividing files, to quote Marvin the Martian, into billions and billions of microatoms (and that is a lot of microatoms), and then getting the people downloading the complete file to upload the microatoms that they have already got to other users.
Because of this, the model only really works when you have a critical mass of users currently downloading the file (or leaving the client running after they finish it, which is a Nice Thing To Do according to Benny’s Big Book o’ Netiquette). If you have too few people hosting the file, you’re just downloading it from one or two people, which really doesn’t take advantage of the reason you’d want to use BitTorrent in the first place — the program’s magic is in its infinite scalability (the way it’s structured, a torrent’s maximum distribution speed is only limited by the infrastructure of the Internet). With files that are less likely to be in massive demand, it seems like a Napster-style P2P app would make more sense.
As an aside, Anna, it turns out that you are only three degrees of separation from the guy who wrote the BitTorrent program in the first place :). In Kyoto I lived next door to a friend of the guy. Small world, no?