Smart Mobs

by Howard Rheingold

  Folding@home (http://www.folding@home.org) is a Stanford University project aimed at understanding the structure of protein in order to develop better treatments for diseases. In March 2002, the popular search engine Google bundled a Folding@home client with the custom search tool it distributes to millions of users.17

  SaferMarkets (http://www.safermarkets.org) seeks to understand the causes of stock market volatility (“You and Your PC Can Help Stabilize the Global Economy”).

  Evolution@home (http://www.evolutionary-research.org) searches for genetic causes for extinction of species.

  Distributed computation is only one example of how peer-to-peer arrangements can assemble scattered resources to create collective goods. Disk space is another resource that can be shared over the Net. File sharing, however, is not about the quantity of computer disk space that p2p memory can aggregate but about the social arrangements that enable the members of a p2p community to copublish and share information. Napster is probably the most well known example of a p2p arrangement for sharing the contents of individual participants’ disks—challenging traditional notions of intellectual property and the existing commercial music industry in one stroke.

  Peer-to-Peer Power

  The story of the “killer app”—the software application that turns an underused technology into an industry—is a central and recurring myth of Silicon Valley culture. The PC was a toy for geeks and gamers until the electronic spreadsheet transformed it into a business tool.18 Email and the Web were the killer apps of the Internet. And Napster was the killer app that awoke the world to the disruptive potential of p2p power. When millions of college students started trading music files in the new MP3 digital recording format, they strained the carrying capacity of large-scale university Internet connections, alerted the vested interests in the existing intellectual property industry that a frontal assault had been launched on their livelihood, and demonstrated that teenagers can ignite world-changing p2p ad-hocracies.

  While finishing his freshman year at Northeastern University, Shawn Fanning spent a lot of time hanging out with other geeks on Internet chat channels. He noticed that his friends were going to some trouble to exchange music files encoded in the new MP3 digital format. Fanning decided to create software that would allow people to search the Internet for the MP3 files they had trouble finding and to exchange them. He incorporated some clever ideas that were circulating in the p2p world, such as building an enormous distributed database by enabling every user to make some disk space available to the file-sharing community. Because they liked music and didn’t like to pay for it and had PCs and high-speed Internet connections, college students drove the Napster epidemic.

  Fanning founded Napster, Inc. in May 1999, dropped out of school, and moved to northern California during the height of the dotcom bubble. Pictures of his stubbly head became an icon in the entrepreneurial pantheon. Napster quickly became the world’s largest community for sharing music files because it allowed easy searching, had a user-friendly interface, let the users communicate with each other through instant messaging and chat rooms, and enabled them to share each other’s bookmarks. The social network multiplied the impact of the network of computer storage. At its height, 70 million users were trading 2.7 billion files per month.19

  The social system for sharing resources was as revolutionary as the application of p2p technology to distributed file sharing. In an article titled “The Cornucopia of the Commons,” Dan Bricklin, the inventor of VisiCalc, the killer app of PCs in the early 1980s, pointed out: “The genius of Nap-ster is that increasing the value of the database by adding more information is a natural by-product of each person using the tool for his or her own benefit. No altruistic sharing motives need be present, especially since sharing is the default. . . . In other words, nobody has to think of being nice to the next guy or put in even a tiny bit of extra effort.”20

  Naturally, the Recording Industry Association of America (RIAA) wasn’t happy about millions of downloads of songs that had formerly fed revenues into their companies. Some recording artists, most notably the band Metallica, also became outraged at this sudden threat to their livelihood. The legal battle began.

  In July 2001, Judge Patel of the Ninth Circuit Court ordered Napster to remain offline until it had shown that it could effectively block the trading of any copyrighted work. Metallica and rapper Dr. Dre settled their legal disputes with Napster, ending all legal actions between the parties. It was agreed that the artists would have final say over which of their songs could be shared on the platform, with the provision that they would make certain “material available from time to time.” In September 2001, a proposed settlement was announced between Napster and the National Music Publishers’ Association (NMPA), in which Napster agreed to turn into a fee-based service with the music being licensed by the publishers to the users. However, by that time, downloads of music from alternative file-sharing services outnumbered Napster at its peak, and the fee-based service never took hold.21 Napster filed for bankruptcy in June 2002.

  Although judicial counterattacks killed Napster, other ad hoc networks that enrich their participants—arguably at the expense of intellectual property that belongs to others—thrived. Like other driving technologies of the age of smart mobs, p2p technology is undergoing rapid evolution. The post-Napster generations of file-sharing schemes directly addressed Napster’s perceived weaknesses.

  The main problem with Napster, from a p2p purist’s perspective, is that it was not designed to be truly a decentralized network. Although Napster users stored all the music files they exchanged on their own disks, the users had to go through a central server in order to find the music they wanted and to connect with the users who had it. Having control of that central server was what made Napster a business worth investing in. Like the telephone company, Napster aimed to profit by introducing its users to each other. Having control of the central server is what made Napster, Inc.’s owners vulnerable from a legal standpoint.22

  Gnutella was specifically designed by Tom Pepper and Justin Frankel of Nullsoft to be a totally decentralized system, with no central server.23 According to the company origin myth, “The name Nullsoft was chosen by Justin Frankel in 1995 to label software that he would develop for fun in his free time.”24 In 1997, Frankel began working on a software client that would enable people to play MP3 music. Later that year, Frankel connected with Tom Pepper, who hosted a Web site for Winamp, the MP3 player Frankel had developed. The Winamp software, released as shareware (free to download, with payment on the honor system), was wildly successful. Just as the Mosaic browser made the World Wide Web instantly popular, Winamp multiplied Napster’s success. America Online acquired Nullsoft in 1999.

  While technically an AOL employee, Frankel set out to create an un-bustable, untraceable, perhaps even indestructible file-sharing program. Frankel and his partner Pepper had definite social goals for Gnutella when they created it in March 2000.25 The owner of Nullsoft, AOL, was reportedly unhappy about this innovation, but by the time AOL tried to shut it down, the enabling software protocols were out of the bag.26 It’s no wonder AOL tried to shut down GnutellaNet before it could propagate. It enables the sharing of not only music files but video, text, pornography—anything that can be transformed into digital format. Unlike Napster, there is no deep-pockets single owner to hold responsible. Gnutella claimed to avoid the legal vulnerability of Napster by making the owners and controllers everyone who uses the Gnutella client software. The GnutellaNet Web site proclaims that the Gnutella service was designed to be “anonymous,” “designed to survive a nuclear war,” and “withstand a band of hungry lawyers.”27 It is possible that GnutellaNet’s creators underestimated the opposition. Hollywood joined the recording industry in an all-out as-sault on file sharing when broadband connections made it possible to exchange pirated videos as well as music.

  Because Gnutella users connect to other users and not to a central server, the users’ personal computers
act both as clients and as servers. The Gnutella “servent” software is composed of a minisearch engine combined with a file system. In describing how the network of Gnutella users serves as a search engine and file-sharing facility for its users, the founders compare it to the game of “telephone”:

  When you say to GnutellaNet, “Hey, find strawberry-rhubarb pie recipes,” you are actually saying, “Hey, my close friends, could you tell me if you’ve seen any recipes for strawberry-rhubarb pie? And while you’re at it, ask your close friends too. And ask them to ask their friends.” It’s obvious that after just a few rounds of this, you’ve got a lot of friends working on finding that recipe! And, it’s pretty much impossible for any one person to know who asked the question in the first place.

  So suppose some guy, 6 degrees from you (your friend’s friend’s friend’s friend’s friend’s friend), has the world’s best recipe for strawberry-rhubarb pie. He tells the guy who asked him. That guy tells the guy who asked him . . . And ultimately the answer gets back to you. But only one person in the whole world knows that you’re the original person who asked. And guess what? In GnutellaNet, we even fix that. The guy you asked originally doesn’t even know that you’re the person who’s really asking the question.28

  The central dogma of p2p-as-cult is “every client a server.” In the opinion of some observers, this is not so much an innovation as a return to one of the Web’s founding principles. In “Gnutella and the Transient Web,” Kelly Truelove describes Gnutella’s effect on the Web:

  The Gnutella protocol restores the Web’s original symmetry, enabling even transient computers to effectively participate as servers. It’s far from a complete solution, and alternative systems may eclipse it. Nonetheless, this simple and idiosyncratic protocol is currently in the vanguard of the emergence of the transient Web. The transient Web has the potential to be every bit as disruptive as the conventional “permanent” Web, and possibly more so.29

  The value of Gnutella depends on the voluntary cooperation of its users, who need to give information to the system as well as to use the information they find through the system. This is increasingly difficult with such a large and anonymous user population. Gnutella’s weakness is, you guessed it—free riding. In “Free Riding on Gnutella,” Eytan Adar and Bernardo A. Huber-man report that Gnutella has a significant amount of free riding in its system: Nearly 70 percent of Gnutella users share no files, and nearly 50 percent of the system’s resources are contributed by the top 1 percent of users. The architecture of the system allows for anonymity and decentralized control, but it does not structurally encourage cooperation, rendering it vulnerable to the “tragedy of the commons.”30 The question that remains: Does p2p technology enable people to build public goods that can withstand large amounts of free riding, or will free riding end up destroying the p2p cornucopia?

  Jim McCoy, founder and CEO of Mojo Nation, set out to create a filesharing system that added three important new features: First, cooperation is structurally encouraged by requiring users to contribute at least as much as they take away; second, not only are queries anonymous, but nobody knows where specific files are stored; third, the “swarm distribution” model breaks up files into large numbers of small segments, distributed throughout the network. Swarm distribution makes it easier to find the most popular material and ensures its availability, even though the servers are only transiently available.31

  Mojo Nation shares the advantages of other open source software such as Linux; because the source code is available to any programmer who wants to tinker with it, an ever-growing community of developers improves the software. The downloadable client is used to publish and retrieve information from Mojo Nation—the collection of users who run the software at any moment. Mojo Nation brings together trust management, security through encryption, and a distributed accounting system.

  Mojo Nation incorporates an economy of incentives, using micropay-ments called “Mojo” to reward users for distributing and uploading files to the network. Every user is expected to contribute something, whether it is system resources or digital cash, for the transactions they make within the community. Users earn Mojo by acting as a server, allowing their bandwidth or disk space to be used, or providing other services. A market mechanism enables buyers and sellers to determine prices, and prices can be advertised. As a result, users create a reputation system of sorts since the quality of service and the reliability of service providers is constantly under review and tracked by agents, which check for good connectivity and proximity to resource providers.

  All files that are distributed on Mojo Nation are broken into hundreds or thousands of pieces, and no files are stored in their entirety in a single location. When a user requests a file, it is automatically downloaded in parallel streams from the nearest peers that have chunks of it available. If a peer is offline, it is likely that another peer will have the needed chunk. Documents that are requested more frequently are distributed more widely, minimizing bottlenecks that would slow down the system when a majority of users try to download the most popular files simultaneously. Encryption is used to cover the tracks of people making requests and to make it impossible for participants to know exactly what content they are storing on their PCs as part of their contribution to the system. Mojo Nation ceased operations as a commercial enterprise in February 2002, replaced by the noncommercial Mnet project.32

  The first peer-to-peer networks linked social networks into cooperative ventures that shared computing cycles, files, and bandwidth. The next generations of p2p sociotechnology include p2p systems that share decisions and judgments.

  Sheep That Shit Grass

  Cory Doctorow, thirty-year-old online auction addict, Internet jack-of-all-symbol-manipulating trades, and award-winning science fiction writer, is also the most enthusiastic p2p proponent I’ve met. When I learned that he was working on a p2p scheme that would enable people to share their decisions about what they find interesting, I called him. He was living in Toronto at the time. A year later, I found his name while looking through a list of publicly accessible wireless Internet links in San Francisco (see Chapter 6, “Wireless Quilts”). The combination of p2p and wireless technology and his passionate defense of online collective action drew me to Doctorow’s Potrero District apartment. I suspected he had some ideas about where p2p might be heading, and I wasn’t disappointed.

  The hallway leading to Doctorow’s apartment is decorated with his landlord’s collection of faux-Tiffany lamps, Star Trek commemorative plates, and what Doctorow calls “framed assemblage sculptures made from symmetrically arranged junk jewelry.” Doctorow’s own apartment is filled with paper ephemera from Disney parks. Indeed, it turns out that his interest in Disney memorabilia was one of the motivations for creating OpenCOLA.

  Here is Doctorow’s “elevator pitch” for OpenCOLA:

  The idea is that you have a folder on your desktop, you put some things in it you like, and it will fill up with things that you’ll probably like. It figures out what you’ll probably like by finding peers in the network who have taste similar to you and telling you what they think is good. The software fetches documents from peers and from various Internet servers, puts them under the noses of people that it thinks will like them, then watches what the peers do when they get them: Do they attend to them, or do they throw them away? These implicit, observed decisions are aggregated and the result is a “relevance- switched” network where documents automatically migrate to the attention of people who’ll probably like them, based on human decisions.33

  Doctorow, in addition to being cofounder and chief evangelist for Open-COLA, is a winner of the John W. Campbell award for new science fiction writers. He has a beard, chain-smokes, and seems to pull epigrams from the air. My favorite Doctorow epigram is his description of Napster as a solution to the tragedy of the commons dilemma. Doctorow refers to Napster users, who provide the same resource that they consume, as “sheep who shit grass.”

  I sat on Doct
orow’s couch while he smoked filter cigarettes and spun rants in response to my inquiries. Above his head, perched on a shelf high on one wall, is the small box that quietly provides wireless Internet access to any nearby geek who knows the node is there. His notions about finding something interesting in the seething chaos of the Net are sufficiently novel that it took a while to wrap my mind about the idea. “Relevance switching” is a way of creating your own self-updating map of the network by querying the social networks of people who share your interests.

  Doctorow and two friends in Toronto ran a technical services company during the dotcom era. Like many contract service companies, they yearned to create something to call their own. Doctorow and his partners, John Henson and Grad Conn, wanted to create a tool that they personally lusted after—some kind of software agent that “would keep us abreast of things we were interested in, and would inform us about things we should be interested in.”

  They looked at how search engines help people find things by indexing every document on the Internet and at how some collaborative filtering tools (like the one Amazon uses) are able to recommend books or music people might like by looking at the tastes of people who make similar choices. Would it be possible to build a collaborative filter that would include every document, music, graphic, video, and software program on the entire Internet, and at the same time take into account the tastes of every person on the Net, somehow keeping track of how all other people’s tastes compare with one’s own? Why not think big?