Accountability

Although users benefit from anonymity, there is also a loss of accountability. The system can't attribute actions to anyone. If the system also lacks the capability to control the use of resources, users might take advantage of the system by consuming them unfairly. P2P systems can regain control of and make clients accountable for the use of system resources by using the micro payment and reputation models.

Micro Payment

Peer-to-peer systems can divide the resources among participants using the micro payment model. With this model, the consumers of the service make small payments of some kind to the producer for small units of work. By keeping the work units small, the producer minimizes the risk to itself. Payments continue with each transaction, until one of the parties is not satisfied, or until all the transactions have been completed.

Although the term micro payment seems to imply cash or something equivalent, it can be also be a Proof of Work (POW). A POW is designed to slow the consumer down, to prevent the consumer from overwhelming the system or using more of the resource than it should. This model is suitable for use by identifiable and private clients.

Reputation

P2P systems can work with pseudonymous or public consumers over a longer period using a model built upon reputation. Consumers consult a reputation server to obtain information about potential producers. Using this information, the consumer offers work to the producer of choice. The producer consults a reputation server to discover the reputation of the consumer. Based on the reputation, the producer may reject or proceed with the transaction. Fewer resources are given to peers that have poorer reputations.

Although the prior example used a centralized reputation server, it is possible to use a system such as PGP's web of trust, or the X.509 Public Key Infrastructure.

Note

Accountability systems like these can also significantly reduce freeloading, or the tendency of people to take resources without giving anything back for them. A September 2000 study conducted by Xerox PARC concluded that 70% of Gnutella clients did not contribute to the shared content, causing overall system degradation.

Free Haven, a peer-to-peer document storage service, is an example of a system that provides accountability based on reputation in an anonymous environment. Free Haven's servers work together to publish documents by accepting contracts between themselves to store each other's data. A server broadcasts a positive referral about a peer when the peer successfully fills a contract for the server. Servers obtain additional storage space on other servers for better reputations. Documents are fragmented into shares, and shares are paired as buddies. Buddies are not stored together. Each share knows what server its buddy is on. A share periodically queries its buddy to see if it's still alive. It broadcasts a negative referral about a server if it notices that its buddy doesn't respond.