Linus is dubious... but we get some praise

[Boris 'pi' Piwinger]

Greg Louis <glouis at dynamicro.on.ca> wrote:

> "Especially if the "random words" in the spam end up being weighted by
> real frequency, you just _cannot_ use single-word bayes filters on it."

Why this discussion all the time? Also on the spamfilt
mailing list. I don't get. I never have a problem with
those. It just works out.

Who is seeing a problem?

Should that really be a difference betwenn full training and
train-on-error?

pi