Wednesday, September 24, 2008

Guest Post: To Track Ratios, Or Not To Track Ratios

I'm a little short on time tonight, so I'm going to share a brief essay, courtesy of a friend of mine. He's quite knowledgeable about the BitTorrent protocol, and fairly critical of the status quo.

Talk about things that make you go hmmm...when TorrentFries was a private tracker, only allowing members to upload and download, they collectively reported uploading 100x more than they reported downloading.

The fact is it's impossible to track ratios accurately--the BitTorrent protocol wasn't designed with statistics in mind and doesn't support them reliably. Ratio updates are not infrequently lost and it's easy to fake ratio, undetectably, using any client.

Given its unreliability it is counter-productive to use ratio as a basis for rewards/punishments. You encourage cheating and end up rewarding the dishonest and punishing (by way of not rewarding) the virtuous.

You can try to identify cheats, but someone with a sky high ratio may have a seedbox and an inclination to support your community. At least until you mistakenly accuse them. And, of course, the smart cheats don't do so outrageously--so they never stand out in any way.

If you don't track ratios? The Pirate Bay continues to grow in spite of not doing so. Certainly their seed:peer ratio is lower than that of any ratio tracker, but their torrents survive. Download speed isn't everything--if people can wait weeks or months for a DVDRip to come out they can survive a day or two of downloading.

In spite of all that, I recommend you track ratios. They remind people that sharing is what torrenting is all about, and they are mostly accurate. Just don't make a big deal of them and, when people do complain, admit their failings and point out that they shouldn't be taken too seriously. Don't accede to request to "fix" people's ratios though--you open yourself up to a lot of work with no real benefit.

While I'm on the subject, I'd like to debunk a common myth--that it's impossible for everyone to have a ratio > 1. The theory goes that since everything which is uploaded has to be downloaded, the average ratio must equal 1. So, for everyone who has a ratio > 1 there must be someone who has a ratio < 1. So, people with low ratios are victims of bastards who seed too much, who should be pitied rather than punished.

Nonsense. The theory only works if you assume that nobody ever leaves--totally unrealistic, particularly if your policy is to ban people with ratios < 1. Even if nobody ever left, so long as momentum is maintained (ie. so long as people upload within a reasonable period of downloading, thus restoring their ratios) it would continue to fly.

Green Salad
Not an administrator. Not always wrong.

One thing I'd take exception to is his claim that the global ratio does not always have to be 1.0. It does when you factor in users that have left or been banned. If something is uploaded on a private tracker, another user must be downloading it. In my mind, it's also a little incongruous with his arguments against rewards or penalties based on ratio, since his argument relies on users being banned for having poor ratios. Since the global ratio must inevitably be 1.0, someone will inevitably end up banned. Allowing people to register with the expectation that some will receive summary bans seems pretty ridiculous.

That's all for now, folks.


yogi613 said...

dont forget about exciting things like freeleech to help those that have a difficult time keeping their ratios up.

OnionRings said...

Feeling a need to implement freeleaches is, in my mind, a sign that your user pool/community is unhealthy. That is an inherent problem for high-private trackers. I'm sure Curly has a better formulated opinion than I.

CurlyFries said...

Basically, if it's necessary to implement periodic exceptions to the normal operation of the tracker, you're not doing your job right. A working tracker works for the downloaders all the time, not just a few times a month.

The fundamental point is that people join trackers to download. They may have stuff to share, but they do so in order to justify downloading more. Encouraging them to download responsibly is one thing, but implementing policies that actually prevent or discourage people from downloading what they're after is contrary to the purpose of a tracker.

gipsy_scholar said...

This situation of downloading more than we upload is only going to worsen with time as consumers do not usually have access to large amounts of material as the big multinational copyright holders do. A good way of getting around this is by organising buying groups. I hear that the groups over at The Place do this very well. Is this a popular activity on other trackers.
I also like the Bounty Hunter policy that is operated over at the best art house movie site, that is used to encourage uploaders. (Are we allowed to mention trackers by name here?)

OnionRings said...

You can mention tracker names. We just don't want to mention ours.

c96aes said...

The thing about ratio is that someone must always be the last to download something everybody else already have. That's why I'm in favour of other measurements, like seeding time * rarity or something like that. MVGroup seems to have an "adoption" scheme, for instance.

Laffin said...

But Ratios are kinda stoopid. Think of other ways to reward the users.
Such as seeding time, quality of torrents. etc etc.

there are many ways to reward your users than by adjusting their ratios.

when ratios arent no longer a formula to user rewards. U will see less cheating.

CurlyFries said...

Seeding time sounds neat in theory, but don't forget that you can throttle your client to 0.1 kB/s and still be credited the same as someone running a seedbox.

Clicky Web Analytics