Linus is dubious... but we get some praise

[Chris Fortune]

Bogofilter rocks!  The big selling points are ease of use / installation, accuracy, and the   s p e e d !    A larger programming
team may have bloated the code unneccesarily.

BTW, Kevin O'Connor's comments on "random word spam" are not factual, go to http://spamconference.com and view the 1st speaker on
the 1st video, Terry Sullivan  (Volatility and stability in spam features), who applies some very interesting statistical analysis
to the spam corpus and concludes that spam is highly stable in it's features and is evolving in discrete steps.  The so-called
"random" spam we've been seeing in the last few months is an example of discrete evolution, but is also surprisingly stable within
it's subclass.

Bayesian classification is highly appropriate and bogofilter leads the way IMHO.


----- Original Message -----
From: "David Relson" <relson at osagesoftware.com>
Cc: "bogofilter" <bogofilter at aotto.com>
Sent: Sunday, January 25, 2004 7:53 AM
Subject: Re: Linus is dubious... but we get some praise


> On Sun, 25 Jan 2004 07:25:41 -0500
> Greg Louis wrote:
>
> > There's an interesting thread on the linux kernel mailing list --
> > off-topic for them but interesting for us.  A good starting point is
> > here:
> > http://www.ussg.iu.edu/hypermail/linux/kernel/0401.3/0106.html


> Indeed, a lot of ideas and testing have gone into spambayes and there
> have been many interesting discussions.  Sometimes it surprises me that
> bogofilter, with its small core team, does so well!  Evidently we have
> enough of what's needed.
>
> Good work, fellows!
>
>