[ale] Spam prevention

David Corbin dcorbin at machturtle.com
Thu Jun 29 19:24:20 EDT 2006


On Thursday 29 June 2006 06:37 pm, Fulton Green wrote:
> On Thu, Jun 29, 2006 at 05:37:33PM -0400, David Corbin wrote:
> > I've been using Spam Assassin for years, but I'm really getting a bit
> > disappoitned with it.  I easily have 30 plus SPAM that get through a day,
> > and a fair number of "false positives".
> >
> > Is there a better Spam solution out there these days?
>
> Besides the currently-in-testing Google Mail for Domains?  OK, seriously:
>
> First off, you may want to explore tweaking the scores of the various
> flags and / or the spam threshold, to take care of the false positives.

I have done that quite  a bit, but over time, things have become less 
successful for reasons you mentioned below.

> For the spams that are passing through, read below:
>
> I have seen that SpamAssassin works best when used in conjunction with
> various blackhole lists.  They usually require a subscription fee.
> However, some hosting providers (such as the one I'm with) simply roll in
> the service fees into your account fee.  So what you'll get is a set of
> additional SpamAssassin flags that indicate which blackhole lists raised
> their flag.

hosting provider? what that :)  I thought there were some 'free' black lists, 
and I *thought* I was using them, but I could be wrong.


>
> Most blackholes examine either the sender IP address to see if it's in a
> dynamic pool and / or was recently reported to be spamming.  Some
> blackholes will also examine content.  True, SpamAssassin has a
> traditional strength with text-based content examination, but newer
> obfuscation techniques (HTML trickery, embedded images) are meeting with
> greater success by the spammers.
>
> The only real issues I've seen with false positives after using those
> techniques seem to be with lists I've opted into (e.g., BusinessWeek).
> However, when I asked one of the blackhole maintainers about it, they
> claimed that the lists in question weren't dropping "test emails" that
> requested opt-opt.
> _______________________________________________
> Ale mailing list
> Ale at ale.org
> http://www.ale.org/mailman/listinfo/ale



More information about the Ale mailing list