With fairly simple means I can block almost all of those mails on my own mailserver with practically zero false positives. In fact, the false positive rate was so low that I now directly reject those mails during the SMTP dialog. This frees me from checking my spam folder, and at the same time gives senders of legitimate mails (false positives) a notification that their mail has not been delivered.
- Wait 15 seconds until responding on the SMTP port. For real mailservers that's not a problem, but spammers typically use a very low timeout as lots of their mailadresses are invalid (and mailservers often don't even exist anymore).
- Enforce some SMTP protocol characteristics. E.g., spammers often try to "pipeline" all their data directly after connection establishment. "Real" mailservers don't do this (at least not until the remote server explicitly states that it supports pipelining). So we can just reject the connection if this occurs.
- Use Greylisting (http://en.wikipedia.org/wiki/Greylisting). Disadvantage: Seems to cause problems with some mailservers.
- Use zen.spamhaus.org blacklist. Disadvantage: There may be some false positives, but have not encountered them so far. You can mitigate disadvantages of Greylisting and the Spamhaus blacklist by not using the blacklist for a accept/reject decision, but for a greylisting/no-greylisting decision. So "good" IPs aren't greylisted, and dubious IPs must pass the Greylisting check.
- Strictly enforce SPF if the sender's domain configured it with "-all". This blocks all the rolex.com spam.
- Not yet, but I will use this soon: Add support for Spamhaus' SWL whitelist and do no further Spam checks for messages from trusted IPs.
- If a message passes all those checks use Spamassassin for a content check. We can allow a pretty high Spamassassin here as most spam mails have been blocked in the previous steps. By using Spampd with Spamassassin, Spamassassin can reject messages during the SMTP dialog. In addition, Spamassassin is configured to use Razor, to detect "known" spam messages.
Also don't blame the spammer for spam. Blame people. Fact is, there wouldn't be spam if it didn't work. The ugly truth is that one person in a million who can't find porn or Viagra on the Internet by themselves pays for all the spam.
Facebook is obviously big now for communication but don't write off SMS yet because it has one thing all the messaging wannabes still don't: ubiquity.
Of course the fact that AT&T charges me 25 cents per SMS sent AND RECEIVED (no other country does this to my knowledge; certainly Australia and Europe don't) or I have to pay $90 for a pretty basic level of service (ludicrously high by OECD standards) is another story entirely...