[Spambayes] Please do a plug-in version for Outlook Express as well!

Tony Meyer tameyer at ihug.co.nz
Fri Nov 26 06:42:34 CET 2004

> Thanks for the through explanation. I have lots of programs 
> running on my computer: ZoneAlarm, AVG, SpywareGuard, Spybot,
> MrPostman, etc. to name a few.  I just was trying to reduce
> the number of running programs by one!

A plug-in will take up more or less the same amount of memory and processing
time, so all you'd really gain is reducing the list of running programs in
Task Manager (which, given how many processes are running with an average XP
system, is nearly pointless).

> About the proxy, it seems to me that it doesn't come with a 
> default list of filter words.  It would be better to have a list
> of words in addition to the learning feature, so it will be
> ready to do something as soon as it is installed :)

This has been discussed a number of times.  The arguments against are:

  (a) You really need to supply both sample ham and spam messages to start
off with, but how do you choose these?  Are there examples of ham and spam
that will really be universally considered such?

  (b) SpamBayes learns really fast - you only need to train on a dozen
messages and results are fairly satisfactory*.  So you're not really gaining
that much.

  (c) The success of filters like this lies in the fact that it learns what
you personally consider good & bad mail, and this works against that.

All that said, it would be simple enough to do if someone was going to do
the work in (a) of picking the messages (and could be optional).  IIRC, one
of the developers (Skip?) even started work along these lines, but there
really isn't much demand for it, and so it's never got very far.

If you strongly feel that it would be a good addition, the best move would
be to add a tracker to the feature requests (RFE) on sourceforge:
<http://sf.net/projects/spambayes>.  That way it won't get lost in list
traffic, and it's easy for other people to add their pleas.


* I use the 'bigrams' option, still experimental with 1.0.1, which probably
makes learning quicker.  But it should 
still be quick even without that.

Please always include the list (spambayes at python.org) in your replies
(reply-all), and please don't send me personal mail about SpamBayes.
http://www.massey.ac.nz/~tameyer/writing/reply_all.html explains this.

More information about the Spambayes mailing list