Yenta is alpha-test software. As is true of all alpha-test software, it's bound to have some bugs. But later releases of Yenta will do much more than fix bigs. What's ahead?
Much better determination of interests. The way Yenta currently determines your interests can be greatly improved. Right now, it assumes that HTML tags, signature lines, various header fields, and so forth, are actually useful content. In reality, they are generally useless, and cause Yenta to present a large number of irrelevant interests. This requires users to spend time turning them off. The next major improvement to go into Yenta is the incorporation of a brand-new version of the document scanner which understands these issues.
Support for running more than one Yenta on a given computer. Yenta talks peer-to-peer to other Yentas, and runs as an unprivileged user. Because of the way the Internet works, this makes running more than one Yenta on a single machine somewhat complicated. Currently, only the first Yenta that starts on a timesharing host can receive inbound connections from other Yentas. (A Yenta which starts later may initiate connections, but may not receive them. This makes reception of incoming messages somewhat slow and unreliable, since they won't arrive until that Yenta happens makes an outbound connection to one which is holding messages for it.) This limitation will be removed in the next release.
Internal bookmarks. If Yenta is used in a timesharing situation (see the paragraph above), each Yenta that starts must pick a unique port when it brings up its user interface -- after all, in this situation, one machine is hosting two or more "web servers," and there must be a way to tell them apart. (The port is the number you see in http://some.machine:12345/some/url.) But that means you can't use your browser's bookmark system to remember locations, because if your Yenta is forced to choose a new port the next time you start it, your browser will think that's an entirely different location. Yenta will soon support an internal bookmark system, which will work no matter which port Yenta must use on any given run.
Better security against traffic analysis. Currently, when one user is talking to another using Yenta, someone with a packet sniffer and access to the underlying network can determine which two Yentas are having the conversation. This does not tell the eavesdropper what is being said, nor does it necessarily indicate which user are speaking -- but it is often possible to make some very good guesses, if the attacker can finger the machines or otherwise knows which user is logged into which machine. And if the attacker knows what one of those users is interested in, it may be possible to make some good guesses about the other one. In a later release of Yenta, one-to-one conversations will be better protected against this form of traffic analysis, by using broadcast communications more extensively.
Better message filtering and reputation-handling. Yenta currently has hooks to enable filtering of messages, but they are not very sophisticated. Also, it has no easy way to write rules which can reject messages, sight unseen, based on the reputations of their senders. These extensions are planned for a future release. If spam and other forms of social misbehavior become common, they are likely to be added quickly.
A large number of small user-interface features. There are a number of rough edges, small missing abilities, and so forth, which will make Yenta more pleasant to use. For example, making it possible to have Yenta efficiently push updated pages, rather than waiting for the user to fetch them from the browser, would be a worthwhile addition. There are many others.
Note that this is a partial list. It gives you an idea of where we're headed, but there will certainly be a lot of other improvements, in addition to this list, by the time Yenta makes it to beta.
[ Are you a programmer? Want to improve Yenta's security? See the next page. ]