More NetAss business: they seem to have filed a very broad patent for doing security analysis that may or may not have implications for my workplace. Here is a reference to an article about it. However, I did at the same time have a coworker tell me how well CRM114 works for him as a spam filter. This is a pure Bayesian analysis of the text of messages, which has a lot less of the overhead of SpamAssassin. It simply trains a network on the text of messages and then uses that.

CRM114 is actually a programming language, and the stuff that makes up the filters are very simple scripts in this weird language. The scripts allow for control via e-mail, such that you can forward a mail to an address with a command and password in it, and it will act on that, by whitelisting, blacklisting, or adding the message to the spam/non-spam neural networks. It's all quite interesting. Rather than buying in to all the heavy lifting that SA does, the principle of CRM114 is that we can analyze the message. Spammers can disguise the text in ways, but they can't disguise the message without rendering the spam ineffective. If they can communicate it to you, the CRM114 filters can learn it.