[Spambayes] SpamBayes now filers less than 50% of my spam.

[Bill Yerazunis]

   From: David McNab <david at rebirthing.co.nz>

   Rob Anderson wrote:
   | I am having the same problem. I have 311 ham and 1093 spam in my training
   | database yet SB only catches about 50% of the incoming spam.

   Same with me.

I'd check to see if you're having a configuration problem.

CRM114 is running essentially the same code (ok, maybe a 2x improvement
according to the SB torture-test corpus) and I'm still solidly in
"ivory snow" accuracy.

   Maybe time to switch to a challenge/response system - one that when
   receiving messages, sends a 'challenge' email to the sender, asking them
   to reply. Once replies are received, senders end up on the whitelist,
   and their emails are delivered without further challenge.

See "CAMRAM", which does this only when a Bayesian says "I can't tell
for sure".  

   I'd suggest switching to challenge/response based filtering. That'll
   keep the spammers at bay for a few months.

Except that there's no path back to the spammer for the challenge;
they do their best to cover their tracks as their IP address is the one
thing that can cost them their upstream connectivity.

thus, C/R systems _will_ stop them.

The problem is that C/R will annoy the rest of us.

    -Bill Yerazunis