stating the obvious archives | about

Content as Currency
Oct 03, 2000 :: Michael Sippey

As Napster heads back into court this week, hordes of "music lovers" will likely flood the Napster network, hunting for that perfect rip of whatever it is that gets their blood pumping. And regardless of the outcome in court, attention of the popular press will likely circle back to the viability of Napster alternatives -- Gnutella, Freenet and other distributed, person-to-person file-sharing systems.

With perfect timing, First Monday published Free Riding on Gnutella, by Xerox PARC researchers Eytan Adar and Bernardo Huberman yesterday. The study, which has been widely discussed on mailing lists and weblogs since its initial publication on the PARC site, contends that the vast majority of Gnutellans use the system only to search for files, not to share their own. With a convincing academic rigor (read -- charts and graphs), they argue that Gnutella is suffering from a tragedy of the commons: "70% of Gnutella users share no files, and 50% of all [search] responses are returned by the top 1% of sharing hosts," they report.

Adar and Huberman argue that the anonymous nature of Gnutella is a key factor in its apparent demise. "In order for distributed systems with no central monitoring to succeed," they conclude, "a large amount of voluntary cooperation is required, a requirement that is very hard to fulfill in systems with large user populations that remain anonymous."

While Gnutella may indeed be suffering from a tragedy of the commons, I don't believe that removing anonymity would make much of a difference. Leaving aside the obvious legal implications of adding user identities to the Gnutella network (wouldn't the RIAA just love that), it's not the ideal architectural solution to the Gnutella problem. Even if there weren't legal repercussions to logging on as Michael Sippey and sharing that bootleg copy of Kid A that I happened to get my hands on, I'd still just log on, point my Gnutella client to an empty directory on my hard drive, and search away. After all, the incentive of people knowing that they swiped Kid A from me isn't enough to encourage me to share. And conversely, the disincentive of people knowing that I'm searching without sharing isn't enough to encourage me to point my Gnutella client to a richer directory.

A more appropriate solution to the Gnutella problem would be a market-based approach, where the content itself is used as currency. Imagine a P2P system that operates like the informal "community" of folks who run warez FTP servers. You want to download something? Well, you better upload something in return...

  • You log on anonymously. After all, who you are isn't important, only what you've got to share.
  • As you share files, you "price" them, in a currency denominated in other files. You could require users who want to download a particular track of yours to have on their drives a certain number of songs by the same artist, or in the same genre, or of a similar bit rate, or be sitting on a certain line speed. Or any combination of those, or any other musical metadata. (Example: "You want this rare b-side from Sleater Kinney? You have to be sharing an entire Sleater Kinney album, or at least fifteen songs released on Sub Pop since 1990.")
  • When you search for a particular track, the app would return not only the available files, but also their associated "prices." Your app would automatically recognize if you have the appropriate number or type of files shared, and flag those tracks in the UI as available to you.

There's one key infrastructure component required to make an application like this work...and several interesting implications.

First, the infrastructure. The component required to make this system anything other than a pipe dream would be a CDDB-like service (or network of services) that would authenticate MP3 collections. Imagine this -- every time you add a file to your shared directory, the app would read the file and authenticate its artist, album, song title, genre and any other appropriate metadata, and tag the file appropriately. It would use public-key cryptography to ensure that an authorized authentication service would be the one doing the authenticating. And until a file is authenticated and tagged, it couldn't be shared on the network. Without such a service, people would flood the network with counterfeit files, tagged to look like the real thing.

Second, the implications. "Pricing" for new files on the system would likely vary wildly at first, as people relied on their own value judgments to set prices ("I really love this new Britney Spears, so it must be worth five Backstreet Boys tracks."). But as files were copied and distributed around the network, prices would likely reach some sort of equilibrium -- more suppliers, more "competition." If one user under priced others offering the same goods, their system would likely get swamped with requests, reducing the availability of that track and driving customers to other higher priced "vendors" that offer faster downloads.

Over time, pricing for widely available files (aka "commodity goods") would likely drive to zero, or at least the "marginal cost" -- i.e. bandwidth -- of providing such files. (Users on reliable high-speed lines could potentially charge slightly more for their files.) But new and "rare" files would likely remain expensive. The existence of any pricing scheme could give record labels and content producers access to a radically new form of market research. In the right hands, the average cost of any particular file on the network (as expressed in terms of other tracks on the network), combined with an index that represents of the availability of that file, could prove more informative than the Billboard charts.

Finally, what I've been describing is a barter-based system. Such a system could evolve its own currency (to facilitate faster and easier transactions), spawn its own market makers, or both.

Removing anonymity won't solve Gnutella's problems. An individual's identity (and associated reputation) doesn't carry much influence on the digital commons. On the digital commons, content is king -- and authenticated content can be turned into something more powerful than the rule of kings: currency.

Editor's note. Many thanks to Steve Champeon, Rogers Cadenhead, Lance Arthur, Greg Knauss and especially Jason Kottke for their input to this piece. Discussion lists sure are useful.



Other pieces about ecommerce: