From: tlhackque <tlh...@ya...> - 2010-09-14 10:02:59
|
>>There's a minimum amount of data required. This is by design - hash-based systems are risky when there is little data to hash, because the risk of collisions significantly increases. I understand - but nonetheless am surprised that there is no response - even -d doesn't say "too small, not processed". >>The best way to combat spam is with a combination of techniques. For the messages you are describing, a tool like SURBL would probably be ideal. Well, these folks seem to be clever (or lucky). Despite my best efforts so far, a lot of this is finding its way through. As you can see below, they're slipping by just under the radar; I had hoped that pyzor would help.. X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on localhost X-Spam-Level: **** X-Spam-Status: No, score=4.9 required=5.0 tests=BAYES_99, RCVD_IN_BL_SPAMCOP_NET,RCVD_IN_DNSWL_LOW,RCVD_IN_SORBS_WEB,T_RP_MATCHES_RCVD autolearn=no version=3.3.1 X-Spam-Date-Scanned: Tue, 14 Sep 2010 00:59:55 -0400 X-Spam-Bayes: score=1.0000 summary=Tokens: new, 31; hammy, 1; neutral, 91; spammy, 12. X-Spam-Pyzor: Subject: Hot Discounts on Boner Pills --------------------------------------------------------- This communication may not represent my employer's views, if any, on the matters discussed. Message: 4 Date: Mon, 13 Sep 2010 09:04:00 +1200 From: Tony Meyer <to...@sp...> Subject: Re: BUG with reproducer: pyzor silently fails to check certain e-mails. To: pyz...@li... Message-ID: <AAN...@ma...> Content-Type: text/plain; charset=ISO-8859-1 > What seems to be happening is that the message text consists of a single data > line - a url. There's a minimum amount of data required. This is by design - hash-based systems are risky when there is little data to hash, because the risk of collisions significantly increases. > I'm actually seeing quite a bit of spam like this, where the subject is some > illicit offer, and the text is just a url. The best way to combat spam is with a combination of techniques. For the messages you are describing, a tool like SURBL would probably be ideal. Cheers, Tony |