We shall see if it does a better job of blocking these relentless professional spam attacks.
Let us know if you have any unexpected problems with posting, we may have to work out a few bugs at first.
Because spam attacks have been getting more sophisticated I have activated a new anti-spam service called Mollum
- voicesweb's blog
- Login or register to post comments
Comments
One spammer thru it already -
One spammer thru it already - I may have to set multiple barriers.
Multiple barriers is inconvienient but it works - heres a tip
Here's a tip - if you are a known poster here, (or a loacla user of our calendar) email us at voices@voicesweb.org and we will give you account special powers that will let you bypass the spam checks.
If you are a real person who wants to post here, email us and we will give you an account with those special powers.
Right now I have the spam check set to use one spam blocking tool after another - you have to fill out TWO captchas to post.
But, if you email us and request it, we will give you the power to post without a spam test.
Trying an experiment - back to just one spam barrier for a week
Two spam barriers was 100% effective in blocking this new type of spam from professional spammers, but, it required legitimate posters to fill out two CAPTCHA forms. (CAPTCHA are those annoying things that make you type in a series of numbers and letters to prove you are a human being, not a bot. Professional spammers hire people from asia to sit at a computer all day filling out those CAPTCHA forms so the spamming script can get thru the barrier.)
You may find my analysis of this new spam technology interesting. It looks like the spammers run a script to collect some keywords from the page or website they are trying to spam. Then the script uses google to search the web for pages with similar keywords, and grabs random sentences from those pages. the script then creates a fake post that looks to a computer like it MIGHT be real, because it has related keywords and a semi random structure. The script then slips in one link - just one, to fool the bayesian filters - to the spam website.
Tricky, eh?