But my question still remains: how are we going to "teach" the filter? :) I do not trust to AI you know... somebody must check periodically the spam candidates, otherwise the Org-Team members could lost their importante messages...
Posted by Andrey Golub on Wednesday, January 24th, 2007 at 13:10
Andrey,
If you would follow the link, you coud see that it has nothing to do with any AI.
Anyway, since we are all busy people I will explain.
Greylisting is playing with a nice RFC feature, it exploits the "try later" response. Whenever an email comes to our server the server says "temporary problem, try later", if it is a spammer - the chances it will retry are really small. If it is a normal server - ti should retry in a few seconds. Our server then accepts the message.
Posted by Michael Tabolsky on Wednesday, January 24th, 2007 at 15:18
Thanks for the great explanation, Michael. Now it's clear and I confirm the approach is working fine, I am getting much less spam in the last days!!!
P.S. just for fun: Also Google says you're the best man explaining how the greylisting works,- check this out
"ask Google about greylisting explain" :)
http://www.google.com/search?q=greylisting%20explain
(at least for today your Blog article appears as the first Google entry!)
Posted by Andrey Golub on Monday, February 5th, 2007 at 10:04
RSS