Summary: spam thresholds?

Martin Sapsed m.sapsed at BANGOR.AC.UK
Tue Mar 25 09:35:19 GMT 2003


Hello again,

Martin Sapsed wrote:
> Since we've finally gone site wide with MailScanner we've had a few
> grumbles that we're not picking up enough spam! (Some people are never
> satisfied!)
>
> Would people mind mailing me with the spam and spamhigh numbers you use
> and I'll summarise the responses next week if anyone's interested?

OK - I had 12 responses (although one wasn't a direct response but
someone put the numbers in another message!) plus ourselves.

Average ordinary threshold was 5.6 and average High score was 13.3.

6 of the 13 are using 5 as the low score, we're now using 4.5 and the
lowest in standard use was 4.4 although someone said they used 4
personally but 5 for everyone else. The highest lower threshold was 9.

On the High SpamAssassin Score, a number of people either didn't use it
or left it at the default of 20. Just using the 6 who had changed the
value, the average was 10.

To some extent experience of the values will vary depending on the
version of SpamAssassin in use.

We've provided a web page for people to visit to request addresses for
whitelisting in order to address the false positives. It will be
interesting to see whether the Bayes stuff in 2.5x will address some of
the more cunning spammers, especially the ones who put HTML comments
with random content in the middle of the words that SpamAssassin looks for!

Thanks to those who responded...

Cheers,

Martin

--
Martin Sapsed
Information Services               "Who do you say I am?"
University of Wales, Bangor             Jesus of Nazareth



More information about the MailScanner mailing list