Overview of Yenta

• Determining how agents are supposed to find each other, such that each agent does not have to know about every other agent out on the network, and
• Protecting their users' privacy while interacting with each other.

To make this more concrete, consider as a sample application a matchmaker, which tries to bring people together. Such a matchmaker could

• find people with compatible professional or personal interests, or
• build coalitions of people interested in the same topics.

One way that an agent might know about the user's interests is to have the agent scan the user's mail and forward some or all of it to other agents. Unfortunately, most users would not stand for having their mail forwarded to third parties, even if we assume that all the other agents are benevolent, which we cannot.

Yet matchmaking itself is a very powerful application, if we can do it safely. For instance, Ringo gets a great deal of its power from such matchmaking, even though it does not identify its participants to each other. Yenta is therefore an attempt to make distributed computation both possible and palatable, even in the face of untrusted other agents.

Other pages in this web describe:

• how Yenta functions,
• how it might be used,
• some of the more political aspects of its implementation (especially in its use of strong cryptography),
• an initial, simplified implementation to enable some simple, early exploration of these ideas, and
• its current status.

Note that this web of pages is quite bushy; there is no preferred, nearly-linear narrative stream. Instead, there are a multiplicity of branch points, depending on which issues interest you the most. In part, this structuring reflects the nature of hypertext; in part, it reflects the interrelatedness of many of these issues in my own mind. This is the sort of structure that, e.g., Minsky's Society of Mind should have been implemented in, and perhaps one day will be.