The future of P2P file sharing

Social networks: The future of P2P file sharing

By Janko Roettgers

Since the recording industry sued Napster in December 1999, the P2P community has tried to evade persecution by making their systems more autonomous and less traceable. Unfortunately this process has also adversely affected the user experience. This may not be so obvious when we look at P2P just from a technological perspective. Today's networks serve many more users than the first generation of file sharing platforms. They also enable us to move bits around the network much faster than before.

But second and third generation networks and services like Fasttrack / Kazaa, Gnutella and Bittorrent lack one thing that Napster had: Community. Napster enabled users to engage in discussions and discover new content simply by browsing the shared folders of random people who happened to attend the same chat room. This sort of accidental collaborative filtering is very unlikely in modern networks.

With the music industry suing users worldwide two main concepts emerged as possible solutions: the concept of anonymity and the concept of darknets. Anonymous file sharing is not a new idea, but with increasing legal pressure more and more users demand technical solutions that allow tham to share content without concealing their identity. Unfortunately in most cases anonymity goes along with a loss of community. If I don't know who is downloading a file from my hard disk I won't have a chance to engage in a discussion about that file.

On the other hand, darknets are small networks that are based on trust rather than on anonymity. The idea of such webs of trust isn't really a new idea either, but it recently received lots of attention when a group of Microsoft researchers envisioned it as a possible future of P2P. The problem with trusted communities is that they face a dilemma of differenciation. If their users trust too many people they could face intrusion. If they are too paranoid they'll suffer from a lack of new content. However this is only true if we define such a group or network as a closed system. I don't think this is necessarily true for today's darknets - people exchanging content with their pals over Instant Messaging or in real life - and it doesn't need to be true for technical more innovative darknets that might eventually be able to replace today's P2P networks.

An interesting model for these future darknets emerged when the social networking site Friendster.com started in spring of 2003. Friendster connects people based on their immediate relations and allows to build quite impressive personal networks. Originally the website connected people up to the fourth degree. Average users could easly be connected to more than half a million fourth-degree friendsters. To sustain a feeling of intimacy Friendster recently reduced the personal network horizon to three degrees. Even with these limitations it is very well possible to build a personal network of tens of thousands of people.

Friendster made the idea of social networking popular, but it is by far not the only platform offering such services. Other websites have introduced significant additions. Tribe.net for example allows to form topic-related networking groups. Orkut offers it's users the chance to distinguish between a general, a professional and a personal profile, offering individual access for many bits of information. The decisions made when completing such a Orkut profile are based on calculated risks. How much can I gain from making such information accessible to other people, and how high are the risks? If I publish my e-mail-address publicly I might risk getting even more spam. Not publishing it at all might make it hard for people that are important to me to get in contact with me.

Such decisions based on calculated risks could also help us to create very effective social P2P networks. Instead of relying on the trust of a closed community, each participant could determine on a case by case basis how much he wants to risk. I might want to share my whole digital music collection with only a handful of close personal friends. However I might be willing to take the risk to share a few hundred files with everyone in my extended personal network. Introducing different layers of groups and relationships might even expand each participant's network horizon.

Unlike closed groups, a social P2P network can not be compromised that easily because each user has different trust settings and in fact, a different network. And finally social networks will almost automatically introduce a whole bunch of collaborative filtering mechanisms. Napster made it possible to accidentally discover new content by connecting with strangers. Social networks automatically connect you to people with similiar interests, making it much easier to find what you want without even knowing what to search for.