Hello Stefan,

Tuesday, October 2, 2007, 3:14:03 AM, you wrote:

> Pete McNeil wrote:

>> I will respond off list.

> Did you try to contact me?

Yes.

> I didn't see anything from you.
> If yes which e-mail address did you use?

I used the address you used to post to the list. <[EMAIL PROTECTED]>

<snip/>

> PS: In the last two weeks I see more spam catched by IMail's spam filter
> (IMail Premium 2006.21). Everything that is caught by that filter has
> passed Sniffer. Anything special going on?

Different filters will always catch some things that are missed by
others. That is why diversity is important in spam fighting - it makes
it difficult for blackhats to craft messages that will get through.

As for anything special going on - the blackhats have been
continuously ramping up their volume, rate of change, and diversity.
That's not news because they have been doing this for a while now. The
rate of increase continues to climb and so leakage will continues to
climb and there will be good days and bad days for all filters of all
types.

With 2000-3000 connections per day you are going to see highly
variable results. In contrast the message rate on one of our spamtrap
processing servers is currently more than 3400 per minute. Many of the
production servers we monitor average 400K per day or more.

If your system were to get the focus of just one of the new stock-push
campaigns in any real way for only one or two minutes you could easily
exceed your traffic for an entire year in spam leakage on that one
campaign alone - and each message would be unique. (of course, your
server would probably fail before that actually happened)

My point is only that 35 messages out of 3000 is "in the noise" given
the environment today on the Internet. You will likely see highly
variable numbers in that range and it is significantly more likely
that you will see a dramatic increase from time to time than any kind
of decrease.

All that said, we are evolving with the problem and we are about to
release a new version of SNF in wide beta that will help - especially
as more nodes are deployed. The new system employs collaborative
real-time learning system. We hope to have a wide beta package
available this week - if you would like to get a jump on that then let
me know off list and I will send you the current test package.

_M

-- 
Pete McNeil
Chief Scientist,
Arm Research Labs, LLC.


#############################################################
This message is sent to you because you are subscribed to
  the mailing list <[email protected]>.
To unsubscribe, E-mail to: <[EMAIL PROTECTED]>
To switch to the DIGEST mode, E-mail to <[EMAIL PROTECTED]>
To switch to the INDEX mode, E-mail to <[EMAIL PROTECTED]>
Send administrative queries to  <[EMAIL PROTECTED]>

Reply via email to