[Spambayes] progress on POP+VM+ZODB deployment

[Sean True]

> >	In other words, Bayesian filtering (as popularized by the article
> >"A Plan for Spam") is only good for individuals, or small groups of
> >individuals who all like the same kinds of ham.
>

In the "real world", I suspect there are many companies who are probably
perfectly happy to block (or label) all "pony and farmgirl" messages,
independent of whether some lonely farm guy thinks they are ham. And for
corporate email, I think such practices are not unreasonable (granted, I've
tried hard
to not work places like that, or be in charge of MIS when I was).

There are legal standards that may require an employer to make a best effort
to keep the "pony and farmgirl" message away from those who might be
offended by even having to _label_ it as spam.

If users can nominate mail that violates community standards, and some MIS
person agrees,
a single filter might well be kept that would be a substantial help to a
large body of people.
As usual, for those with special needs (customer service), or special
privileges (the vice president for  reading naughty mail), more personalized
filters could be made available.

-- Sean