I started the Freenet Project in 1998 with the goal of building a network for truly free communication, and of all the things we have learned since then, perhaps the most salient is that the biggest threats to P2P networks come not from without, but from within the network itself. This is something that the current file sharing networks are now learning the hard way, with those organizations who wish to stop them now infiltrating the networks to sue individual users for providing certain files. And while Freenet has always been designed to protect the identity and security of people who access and publish information from attackers and prying eyes, it's design has never been able to protect the identity of people who operate nodes in the network from one another.

Recently Oskar, who was one of the original contributors to the project and who is now working on his PhD in Mathematics, and I have been discussing the mathematical mechanics behind large scale networks. As a part of this discussion it dawned on us, that because science now believes that human relationships really do form a "small world" (between any two of us, there are only six degrees of separation), with the right algorithms it should be possible to find data fast even in a network where peers only ever talk to peers that they already know and trust. We believe our methods for doing this provide to key to making peer-to-peer networks that are both dark and searchable: secure and efficient. For those who wish to constrain the free flow of information, such networks could be the biggest nightmare of all...