My own code used against me?

Despite my battles with Akismet on other sites last year, I’ve always appreciated the good job it does of fending off comment spam here. Not only does it keep almost all the spam out of the comments, it seems to let all the real comments through. Occasionally I take a look in the spam filter to see if any real comments have been caught, but I haven’t found any yet. I do, however, see trends in spam strategy.

The “names” of many spam commenters are obviously algorithmically generated in the hope of avoiding detection. Examples from the last couple of days are:

Spam comments themselves seldom have links anymore. Instead, they tend to be fulsome praise like this:

I must say, this blog is something else. The data here is very consistent with real life subjects. Thanks to you for communicating your thoughts here.

The expectation, I guess, is that readers will click the “Website” or “Home page” link associated with these oh-so-realistic comments and find themselves unable to resist a pitch for insurance, SEO services, fake Cialis, or (this is a new one to me) detoxifying foot pads.

These flattery comments are usually written with the sort of exotic word choice that indicates a non-native English speaker. Recently I’ve noticed comments that go a little further. Here’s one:

We now have go through a few using the subject material articles in your internet now, and i genuinely like your trend of jogging a blog. I further it to my favorites web site site file and probably examining spine again quickly. Make certain to analyze out my world wide web in addition and let me know what assume.|Fantastic summary, actually genuinely similar to a internet site that We’ve got. Keep in mind test it out sometime and sense totally free to go away me a comenet on it and tell me what assume that. Im typically uncover feedback.2

And here’s another:

love the life in London.It contours a beautiful outline to embed a tiny loop into a layered ,I do not normally do that, but I hope that we will have a mutual hyperlink exchange;3$4

Like the fake names, these came out a random text generator like my Dissociated TextExpander snippet, but with a corpus of blog comments as the starting text rather than Origin of Species. I love how misspellings come through to give that extra level of realism.

I remember this word salad approach being used many years ago by email spammers trying to poison the Baysian filters that had cut into their business. It didn’t work then, and I’m surprised that anyone would expect it to be more successful now.