[HCS-D]Re: new student group

Ivan Krstic krstic at fas.harvard.edu
Wed Dec 8 15:40:25 EST 2004


Gregory N. Price wrote:
> I don't know how fast CRM114 is, but SpamAssassin takes roughly a
> second, which can be made something like a quarter-second with spamd.

CRM114 uses sparse binary polynomial matching that's roughly 4 times 
faster than SpamAssassin.

> So all we have to avoid is a situation like HASCS's where some
> legitimate mail with no resemblance (for a human) to spam gets blocked
> for reasons like mentioning a domain whose IP address once belonged to
> a spammer.

This is the product of an RBL-style approach (I think FASCS uses XBL). 
As long as we're only employing tagging-based spam filtering ala 
SA/CRM114, this won't happen.

-IK


More information about the hcs-discuss mailing list