Bram Cohen and BitTorrent
Abstract
Bram Cohen solved a problem that had made the internet’s economics backwards: the more people wanted a file, the harder it was to get, because every download hit the same overloaded server. His answer, BitTorrent, released in 2001, turned that around. It broke a file into pieces and made every downloader also an uploader, so that popularity became a supply of bandwidth rather than a drain on it. The design’s core rule, tit-for-tat, gave you faster downloads the more you uploaded to others, which quietly solved the free-rider problem that had crippled earlier peer-to-peer systems. Within a few years, measurement firms estimated that BitTorrent accounted for something like a third of all internet traffic. Cohen, who describes himself as having Asperger’s and built the protocol partly because he found the problem interesting, never turned that scale into a fortune. The company he founded to commercialize it struggled for fifteen years to make money from a protocol that was free, open, and mostly used for things its author could not charge for, and in 2018 it was sold to a cryptocurrency entrepreneur who, Cohen said, did not pay the full agreed price.
The Flash-Crowd Problem
In 2001, distributing a large file to many people was an exercise in punishment. A person who released, say, a Linux disc image or a video paid for a server, and every downloader pulled the whole file from that one machine. If the file became popular, the server’s bandwidth bill exploded and its connection choked, so that the moment a release succeeded was the moment it became unavailable. This was the “flash crowd” problem: demand destroyed supply.
Peer-to-peer networks had tried to fix this by having users share files directly with each other. Napster and its successors let a downloader fetch a file from another user rather than a central server. But these systems had a weakness Cohen understood precisely: nothing made anyone share. Most users downloaded and then closed the program, or configured it to upload as little as possible. The generous few carried everyone else, and the networks were chronically short of upload capacity. Economists call these people free riders, and a P2P network full of them collapses toward the same bottleneck as a single server.
Cohen had seen the shape of the answer at MojoNation, a startup building distributed file storage that split data into encrypted chunks spread across many machines. He left MojoNation in April 2001 and spent months designing a protocol whose central concern was not sharing files but forcing cooperation.
Tit-for-Tat
BitTorrent’s design starts by chopping a file into fixed-size pieces, each with a checksum listed in a small .torrent file that describes the whole download. A coordinating server called a tracker tells each participant which other participants have the file, and from then on they trade pieces directly, forming a swarm. Because the pieces can be verified individually, a downloader can collect them from dozens of strangers at once and reassemble them in any order.
Two rules made the swarm work. The first was rarest-first: a client preferentially downloads the piece that is scarcest in the swarm, which spreads rare pieces quickly and keeps any one piece from becoming a bottleneck that could strand everyone if its only holder left. The second, and the one Cohen is known for, was tit-for-tat choking. Each client uploads preferentially to the peers who are uploading fastest to it, and temporarily refuses (“chokes”) the rest. The effect is that the more you give, the more you get. Freeloading no longer works, because a client that will not upload gets throttled by everyone around it. The cooperation that earlier networks had begged for was now the fastest path to your own download.
Cohen demonstrated BitTorrent in February 2002 at CodeCon, a small San Francisco conference he had co-founded with his roommate, the cryptographer Len Sassaman. To stress-test the software with real load, he later admitted, he gathered free pornography and offered it as bait to draw in beta users, because the protocol only proves itself when a real crowd shows up. It worked. The swarms held.
A Third of the Internet
BitTorrent spread because it made large-file distribution cheap for anyone. Linux distributions shipped their disc images as torrents. Software projects, game patches, and later high-definition video moved over it. It also became the dominant tool for online piracy, because the same property that made it good at distributing a free operating system made it good at distributing a copyrighted film: the more people wanted it, the faster it went, and there was no central server for a rights holder to shut down.
By the mid-2000s, network-measurement firms were reporting that BitTorrent made up a very large share of all internet traffic, with estimates commonly in the range of a quarter to a third. The exact figures were disputed and varied by region and method, but the direction was not: for several years BitTorrent was one of the largest single sources of traffic on the internet, and internet service providers began quietly slowing it, a practice that became a central exhibit in the net-neutrality fights.
The protocol itself was neutral, and Cohen was careful to say so. He had built a way to move bytes efficiently. What people moved was not his to decide, a position that kept him at an awkward distance from both the pirates who made BitTorrent famous and the industries that wanted him to stop them.
The Company That Could Not Charge
In 2004 Cohen founded BitTorrent, Inc. with his brother Ross Cohen and the businessman Ashwin Navin, to build a company around the protocol. The problem was structural and never went away: the protocol was free and open, anyone could write a BitTorrent client (many did, and the popular ones were made by other people), and the largest use of the technology was one the company could not legally touch. In November 2005 BitTorrent, Inc. signed an agreement with the Motion Picture Association of America to remove links to infringing content from its own website, aligning itself with the law and away from the traffic that had built its name.
The company tried to sell legal content, then to license the technology, then to build new products on top of it, and none of it produced a business proportional to the protocol’s reach. It was renamed Rainberry, Inc. In June 2018, after fourteen years, the company was sold for a reported $140 million to Justin Sun, founder of the Tron cryptocurrency, who wanted BitTorrent’s large user base for his blockchain ambitions. Cohen left within weeks. He later said publicly that Sun had not paid the full agreed amount, and that he “eventually got most of it” with no explanation for the shortfall. The most-used file-transfer protocol on the internet had made its inventor comfortable, not rich, and had ended up as an asset in a crypto acquisition.
Chia
Cohen’s next project circled back to an argument he had been having with the cryptocurrency world. Bitcoin and its imitators secure themselves through proof of work, which pays miners to burn enormous amounts of electricity on pointless calculation (see The Cryptocurrency Revolution). In 2017 Cohen founded Chia Network to build a cryptocurrency secured instead by proof of space and time, which uses unused disk storage rather than raw computation, on the theory that idle hard-drive space is abundant and cheap where electricity is scarce and dirty. Chia launched its network in 2021. Critics noted that it created its own problem, a run on hard drives and accelerated wear on the drives used for “farming,” and that greener than proof of work is a low bar. But it was recognizably the same instinct that had produced BitTorrent twenty years earlier: take a resource that is sitting idle across millions of ordinary computers, and design the incentives so that using it collectively solves the problem.