Freenet
Initially, each node has no information about the performance of the other nodes it knows about. This means that routing of requests is essentially random. But since different nodes have different randomness, they will disagree about where to send a request, given a key. So the data in a newly-started Freenet will be distributed somewhat randomly.
As more documents are inserted by the same node, they will begin to cluster with data items whose keys are similar, because the same routing rules are used for all of them. More importantly, as data items and requests from different nodes "cross paths", they will begin to share clustering information as well.
The result is that the network will self-organize into a distributed, clustered structure where nodes tend to hold data items that are close together in key space. There will probably be multiple such clusters throughout the network, any given document being replicated numerous times, depending on how much it is used. This is a kind of "spontaneous symmetry breaking", in which an initially symmetric state (all nodes being the same, with random initial keys for each other) leads to a highly asymmetric situation, with nodes coming to specialize in data that has closely related keys.
There are forces which tend to cause clustering (shared closeness data spreads throughout the network), and forces that tend to break up clusters (local caching of commonly used data). These forces will be different depending on how often data is used, so that seldom-used data will tend to be on just a few nodes which specialize in providing that data, and frequently used items will be spread widely throughout the network. This automatic mirroring counteracts the times when web traffic becomes overloaded, and due to a mature network's intelligent routing, a network of size n should only require log(n) time to retrieve a document on average.
Keys are hashes: there is no notion of semantic closeness when speaking of key closeness. Therefore there will be no correlation between key closeness and similar popularity of data as there might be if keys did exhibit some semantic meaning, thus avoiding bottlenecks caused by popular subjects.
There are two main varieties of keys in use on Freenet, the Content Hash Key (CHK) and the Signed Subspace Key (SSK).
A CHK is a SHA-1 hash of a document and thus a node can check that the document returned is correct by hashing it and checking the digest against the key. This key contains the meat of the data on freenet. It carries all the binary data building blocks for the content to be delivered to the client for reassembly and decryption. The CHK is unique by nature and provides tamperproof content. A hostile node altering the data under a CHK will immediately be detected by the next node or the client. CHKs also reduce the redundancy of data since the same data will have the same CHK.
SSKs are based on public-key cryptography. Currently Freenet uses the DSA system as its public key infrastructure. Documents inserted under SSKs are signed by the inserter, and this signature can be verified by every node to ensure that the data is not tampered with. SSKs can be used to establish a verifiable pseudonymous identity on Freenet, and allow for documents to be updated securely by the person who inserted them. A subtype of the SSK is the Keyword Signed Key, or KSK, in which the key pair is generated in a standard way from a simple human-readable string. Inserting a document using a KSK allows the document to be retrieved and decrypted if and only if the requester knows the human-readable string; this allows for more convenient (but less secure) URIs for users to refer to.
A network is said to be scalable if its performance does not deteriorate even if the network is very large. The scalability of Freenet is being evaluated, but similar architectures have been shown to scale logarithmically.
Freenet is currently undergoing a major re-write incorporating a number of fundamental changes. Version 0.7 [3], the latest as of 31st May 2006, aims to create a scalable darknet, where users only connect directly to other users they know and trust. The purpose of this change is to protect users who may be placed at risk simply by using the software, irrespective of what they are using it for. In the new model, users will choose to whom they connect, and only those users will know that they are running the software. Previous darknets, such as WASTE, have been limited to relatively small disconnected networks. The core innovation in Freenet 0.7 will be to allow a globally scalable darknet, capable of supporting millions of users. This is made possible by the fact that human relationships tend to form small-world networks, a property that can be exploited to find short paths between any two people. The work is based on a speech given at DEF CON 13 by Ian Clarke and Swedish mathematician Oskar Sandberg.
Other modifications include switching from TCP to UDP, which allows UDP hole punching along with faster transmission of messages between peers in the network.
See also:
Cypherpunk