On Wednesday, November 23, 2005, 4:33:51 PM, Pete wrote:
> |
On Tuesday, November 22, 2005, 5:54:21 PM, Tech wrote:
I don't recommend increasing your rulebase sensitivity at this time. We've just done a lot of work on the back-end system - so it is difficult to predict what the benefit would be (if any) to nearly doubling the size of your rulebase. Also, as a rule, hyper-sensitive rulbases are reserved for special applications on large systems with multiple MTAs. In the past we have made rulebase sensitivity adjustments available to smaller systems on a case by case basis but in light of recent changes to our tuning mechanisms I'd like to stay closer to the rule-book for a while and gather more data. I believe that recent changes in our back-end systems will shortly mitigate much of the leakage, and since the tuning engine has also recently been rebuilt it is unlikely that rules in the 0.1-1.0 band are going to provide enough benefit to offset the costs of doubling your rulebase size. Recently there have been a number of very aggressive days by the "blackhats" and that probably explains what you and others have experienced. This appears to be part of a fairly significant shift in the way spam is generated and the beginning in earnest of a number of new trends that will redefine "how spam is done". Consider: This E-Mail came from the Message Sniffer mailing list. For information and (un)subscription instructions go to http://www.sortmonster.com/MessageSniffer/Help/Help.html |
Sorry to respond to my own post...
There's a lot more to this note that didn't make it to the list. My trusty TheBat! ate everything after the first image. I'll make a paper out of it and post it on the SortMonster site later - there's some interesting stuff happening.
Best,
_M