Into The Unknown

Back in 2007, ThoughtYard.com published a maths-heavy essay analysing the problems associated with torrent trackers, a surfeit of generous seeders and the big problems faced by new users (i.e., leechers) trying to improve upon their low upload:download ratios...

In the file-sharing community, balance is everything. If everyone contributes, the system works, and everyone is happy. Community generosity leads to high-speed transfers and high availability. But not everyone is generous. Some people are downright grinchy when it comes to sharing. And that really pisses people off. So people make private communities to keep out the grinches. But how do you decide who to let in?

The primary method of weeding out the trash for most torrent sites has almost universally been the invocation of a minimum ratio requirement. The reason is obvious: everyone should share the responsibility of uploading. The problem with ratios is that they don't really solve the problem, and are actually contrary to the idea of file-sharing.

In fact, ratio systems can actually create problems and invite leeching. On this page I'm going show mathematically why it is that ratios are a bad idea.

Then I'm going to talk about alternative solutions that actually keep out the bad guys.

The article goes into an immense amount of depth, but it's still understandable even if you haven't lived, breathed and eaten Internets for the past decade. If you've ever wondered what the biggest problems are, as faced by some of the most popular private trackers, this is a recommended read.

As an update, since 2007 a few new trackers have sprung up or gained popularity (sites like TVTorrents, What.CD etc) - What relies on a traditional ratio scheme, but has fairly generous grace periods (based on the amount downloaded) for people to attain a higher ratio through either uploading new material or sharing sufficiently to improve it. TVTorrents has a credits system with no immediately observable ratio - you can run your credits down to 0 Megabytes, and then seed until you have an almost infinite amount of Terabytes in credit. If you dip under the 0 limit sanctions can be imposed on your downloads, but it's a much more lenient system - not as much pressure from keeping your ratio up around 1.0 - and I think it works rather well.

BitTorrent and its successor will be used eventually for content distribution on a mass scale, and there may even be economic models based on who is/are the most prolific sharers amongst the community of their peers. How the models adapt to take into account the inherent inequalities in Internet connections and contrasting abilities to share will be something very interesting to observe.

The (now-orphaned) article is thankfully still available on the Internet Archive: http://web.archive.org/web/20070910054159/thoughtyard.com/twiki/bin/view/Main/TheProblemWithRatios , and CodingHorror has some related discussion.