[Spambayes] Maybe it's too late...

[Tim Peters]

[Thomas Juntunen]
> Recently, there was some discussion of licensing and the legal
> issues surrounding software. Someone said matters were arranged so
> some entity, such as Microsoft, could not swoop in and buy everything
> up.

The assets of the Python Software Foundation (such as the copyrights
covering the spambayes code base) can't be transferred to an entity other
than a public charity (under US tax law -- "public charity" has specific
legal meaning) or a government agency.  It's not just that a company can't
take the copyright from the PSF, the PSF couldn't sell it to a company even
if it wanted to.

> I thought of that when I got a message from the IETF's Anti-Spam
> Research Group mailing list about a spam-related patent, so I went
> to the USPO web site and did some searching. I was most interested
> in patent 6,161,130. (Sorry about the URL from hell).

<http://patft.uspto.gov/netacgi/nph-Parser?Sect1=PTO1&Sect2=HITOFF&d=PALL&p=
1&u=/netahtml/srchnum.htm&r=1&f=G&l=50&s1=6,161,130.WKU.&OS=PN/6,161,130&RS=
PN/6,161,130>

> I am not all that qualified to say, but the description of this
> patent -- filed in 1998 -- sounds a lot like the Bayesian, or naive
> Bayesian, technique to me. After you look at the abstract or
> description, take a look at the Assignee -- a certain enterprise from
> Redmond, WA.
>
> I am definitely not a lawyer, but this can't be a good thing, so I'm
> hoping someone more savvy about how SpamBayes and related engines
> work can tell me there's no real relation here.

Sorry, nobody can tell you that:  whether a patent is enforced is up to
courts, and software patent law is in such a sorry state that, even though
you're not a lawyer (and neither am I), I bet your guess is as good as an
expert's -- and no matter how bad a guesser you are <wink>.

That patent, and others, were discussed early in the life of this project.
I don't worry about it.  That specific patent makes a large number of
claims.  Some are obviously (to me) unenforcable, due to widespread
publication of the relevant art long before this patent appeared (e.g.,
Bayesian classification is about 50 years old!).  Other claims are specific
to server-side spam filters, which spambayes isn't (for most people).  Some
are simply ludicrous claims, such as the idea of retraining based on a user
moving a msg from one folder to another -- yes, that's a fine idea, but no,
IMO it's not novel enough to patent (it's an obvious approach to anyone with
ordinary skill in the art).  Then again, patents on obvious ideas have been
upheld in the past; and then again, others haven't.

If MS wants to pick on spambayes, I'm sure they could.  I doubt that they
do; but, if I'm wrong about that, I can't think of anything that would boost
contributions to the PSF faster <0.9 wink>.