Private BitTorrent trackers are markets

By John Backus

Private torrent communities are markets, and the paper “Economics of BitTorrent Communities” does a great job exploring this with data:

  • Upload/download ratio requirements are a currency system in disguise. Users are just trying to maintain a positive balance
  • Private trackers extend loans to new users to give them time to maintain a good ratio
  • Users with slow internet connections have lower “earning potential” and they work more hours to match that
  • Central ratio requirements price every file equally, distorting the market and suppressing supply and demand
  • Status and altruism motivate excess earning

I enjoyed the paper, so I figured I would share and summarize some keys points in a quick blog post.

Private torrent trackers are invite-only file sharing communities. Paying for an invite will get you and your inviter banned forever. Private trackers require you to maintain an upload/download ratio; if you only download and don’t give back to the community then you’ll be banned within a week.

The best private torrent trackers felt very special. Trent Reznor said OiNK was like “the world’s greatest record store. Pretty much anything you could ever imagine, it was there, and it was there in the format you wanted.”

You could find almost anything on (OiNK’s successor)

This summary doesn’t do them justice; I’m just sharing the bare minimum so I can talk about the economics.

Private trackers require users to maintain upload/download ratios in order to keep their account. The exact requirement depends: some websites require a 1:1 ratio while just require you upload at least 25% of the amount you download.

Kash, Ian A., et al studied DIME, a private tracker for concert recordings. DIME requires users maintain a ratio of 0.25 so users have to upload at least a quarter of what they download. The paper reframes the ratio requirement as “credits”:

We define the amount of credit or wealth each user has on DIME as:
Credit = 4 × upload − download
which is the amount a user can download (in bytes) without uploading and still satisfy DIME’s share ratio requirement.

So, if I’ve uploaded 100MB to others and downloaded 150MB then I’ve got 250MB worth of credits.

Private trackers don’t ban users immediately if their ratio goes below 0.25. Almost all new users will start by downloading content and they need time to upload to other users. DIME doesn’t ban users for having a bad ratio until they’ve downloaded at least 5GB worth of content.

One can view the enforcement cycles as a form of loaning; set at every 5GB, they allow a user to have a negative amount of credit as long as the balance is positive by the next enforcement cycle.

A file’s price is defined by its size. You’re spending credits for every byte you download. All files on a private tracker have the same price per byte.

Files vary in their “resale value”, meaning the credits a user can earn by seeding the file:

The resale value of a file is the amount of credit earned by an uploader, which is four times the amount he uploads (in bytes). This resale value depends on the upload rate achieved (the rate of return), which depends on the uploader’s bandwidth and may also change over time as seeders and leechers join and leave a torrent, and on the duration of seeding, which is up to the user.

The price and resale value of files are central to the private tracker economy. Users need to keep both in mind in order to maintain a ratio and keep their account on the website.

To the extent that users are constrained by their ability to earn credit or simply want to maintain higher ratios, the resale value of a file is important and can influence user decisions.
Snapshot showing all users’ upload and download amounts. Users marked in green donated money to the site; users marked in red (including those covered by the thick green) did not.

The ratio requirement on DIME is 0.25 but many users go above and beyond. Why?

DIME and other private communities promote [altruistic seeding] by encouraging users to upload at least as much as they download, and by issuing social rewards to users with high ratio. For example, users earn special badges for attaining specific levels of activity, are often more respected in the community, and are given additional privileges on the site. These factors inspire many users to upload more than what is required by the minimum share ratio, and suggest that even users with ratios significantly above 0.25 may care about the resale value of a file.

Users with higher ratios are basically marked as the upper class of the community and they’re rewarded with additional perks. Users also grow their wealth within the system for peace of mind; you never know if the files you download in the future will be popular, so why turn down seeding opportunities today?

DIME’s incentives work. User ratios spike around the minimum ratio and recommended ratio.

We observe distinct increases around ratios 0.25 and 1. The spike at 0.25 is consistent with a group of users performing the minimum amount of work required to remain active in the system due to share ratio enforcement. The bump around 1 shows some users attempting to contribute at least as much as they receive from the system, which is consistent with what DIME recommends that users do.
Histogram of user ratios on DIME

Brand new files are the best bet for a user that wants to make sure they can upload to other users and preserve their upload buffer.

The average upload rate on a torrent is extremely high in the hours immediately following its posting, and that there is a severe drop in rate of return over the course of the first few days. After five days, the decrease in upload rate slows, but continues for the lifetime of the torrent.
Newer files consume more upload bandwidth

DIME occasionally hosts a “free leech” period. During a free leech, downloading doesn’t affect your ratio but uploading still counts. Users downloading behavior changes during free leech periods, demonstrating that users understand how files lose “resale value” with time:

during free leech, demand for old files increased 60% to 70% while the demand for new files did not change significantly … these findings imply that users are typically downloading significantly more copies of newer files than older files, but that during free leech users react to the change in prices by consuming many more older files.
Downloads increase across the board during free leech
Interestingly, there was essentially no increase in the number of seeders during free leech, either overall or among those with low share ratios. Given the increase in the number of active downloads during free leech, more downloads are supported by the same number of seeders during this period.

During free leech, demand spikes for older files that users wouldn’t otherwise have downloaded because of low resale value. The number of seeders in the network doesn’t change during free leech, so the supply of bandwidth for older files is there even when users avoid them due to the resale value penalty. It seems like everyone in the system would benefit from lowering the download cost and increasing the resale value proportional to the torrent’s age.

Kash, Ian A., et al recommend that DIME change their central pricing mechanism so that it decreases the download price over time:

In conventional markets, the price of services that have too much supply and too little demand naturally drops. But on DIME, all transfers are credited equally, so prices remain fixed. One can imagine adopting a credit system in which uploads and downloads convert to credit based on the prices of files. In such a system, one can attempt to adjust the price of torrents by slowly lowering the price over time, by making all files beyond a certain age cheaper, or by making the price depend on the seeder to leecher ratio in the torrent. This would attract more reluctant downloaders, and give additional hints to seeders about how to best direct their efforts.

Users with slower internet will have a harder time maintaining a ratio since they upload alongside other users with more bandwidth in order to share a fixed amount of data. This difference in earning potential shows up in the data. Lower earning users with slow internet connections seed files for longer to compensate.

Left: CDF of time spent seeding by users’ upload bandwidth / Right: User download amounts bucketed by bandwidth

At the end of the day, upload/download ratio requirements introduce an economy where users have to maintain a positive balance of an in-network currency. DICE effectively loans in-network currency to users until they hit 5GB and become subject to ratio requirements. Older files are harder to make a profit off of, so users avoid downloading them outside of free leech. Users with slow internet connections have less earning potential and they work more as a result.

Altering the ratio system so that older files cost less to download would be an improvement, but it is still a bandaid. Ratios are centrally planned economies in disguise: the digital equivalent of setting the price of every book in the world at $0.01 per page.

What do we actually want in this system? Well, newly released movies cost me more to buy or rent than they do when that movie has been out for a few years. Great art costs more than mediocre art. Every member of a torrent’s swarm has their own optimization to solve:

  • How much bandwidth could I allocate?
  • How valuable do I think this file is?
  • How many others are seeding this file?

Seeders and leechers should negotiate spot prices with each other depending on the content and timing. This wouldn’t be a mandatory part of the experience; sensible defaults and automated torrent clients could make it seamless. Still, I’m sure that people uploading new popular content to private torrent trackers would want to manually put a premium on the file for the first downloaders. Likewise, someone seeding a very old file might be willing to transfer a file at half price if the alternative is not earning any upload credit at all.

If it looks like a duck, swims like a duck, and quacks like a duck, then it probably is a duck. Private BitTorrent communities are closed economies with bad centrally planned pricing. They are a huge improvement on top of public trackers because we’re comparing a poorly coordinated economy against pure tragedy of the commons. When we try to tune the market mechanisms of a BitTorrent community, the result sure looks like a protocol token. After reading this paper, I’m pretty convinced it quacks like one too.

If you liked this post, follow me on twitter. I’ll be writing more on p2p and blockchain.