Comments on: quiet for a little while longer… in the mean time, the spamming continues http://www.waking-dream.com/2006/05/18/quite-for-a-little-while-longer-in-the-mean-time-the-spamming-continues/ Bridging print, new media publishing, and cultural anthropology. Fri, 04 Jul 2008 20:05:53 +0000 http://wordpress.org/?v=2.5.1 By: mrz http://www.waking-dream.com/2006/05/18/quite-for-a-little-while-longer-in-the-mean-time-the-spamming-continues/#comment-70 mrz Tue, 30 May 2006 18:26:07 +0000 http://www.waking-dream.com/?p=73#comment-70 P.S. I hope the grading went well. And that all is well, in general! P.S. I hope the grading went well. And that all is well, in general!

]]> By: mrz http://www.waking-dream.com/2006/05/18/quite-for-a-little-while-longer-in-the-mean-time-the-spamming-continues/#comment-69 mrz Tue, 30 May 2006 18:25:15 +0000 http://www.waking-dream.com/?p=73#comment-69 Those spam bots could be using some kind of hidden Markov sentence generator. They basically scan the thread for all previous text, then pass it through a hidden Markov generator which creates random sentences that seem likely for the thread. Pepper the new text with links to Spam, et. voila: spam that (hopefully, for them) flies under the radar of a Baysian filter. Why they are bombing old threads isn't clear. Either they are probing your blog to see how to crack any spam filtering, or they are just using dumb algorithms that say "Oh! High post count = hot thread! Good spam target!" Which is probably a good heuristic for a message board running something like InfoPop or whatever, but in a personal blog, it's probably not that smart. A friend of mine has a simple "Turing Test" that you are required to pass to help thwart spammers. It is just a hidden text box. Since most automated spam programs try to fill in every text box they can, if any text appears in that text box, he just silently rejects the message. Of course, this won't work if everybody does it, but it is good if you are getting inundated and don't know what to do. Those spam bots could be using some kind of hidden Markov sentence generator. They basically scan the thread for all previous text, then pass it through a hidden Markov generator which creates random sentences that seem likely for the thread. Pepper the new text with links to Spam, et. voila: spam that (hopefully, for them) flies under the radar of a Baysian filter.

Why they are bombing old threads isn’t clear. Either they are probing your blog to see how to crack any spam filtering, or they are just using dumb algorithms that say “Oh! High post count = hot thread! Good spam target!” Which is probably a good heuristic for a message board running something like InfoPop or whatever, but in a personal blog, it’s probably not that smart.

A friend of mine has a simple “Turing Test” that you are required to pass to help thwart spammers. It is just a hidden text box. Since most automated spam programs try to fill in every text box they can, if any text appears in that text box, he just silently rejects the message. Of course, this won’t work if everybody does it, but it is good if you are getting inundated and don’t know what to do.

]]>