I was just reading an article on Coding Horror on captchas. It talked about sites using bad captchas (like ticketmaster), so that even humans cannot read them. Oh, if you don’t know, captchas are those little images that contain distorted text in which users must fill out before submit a form. You’ll see them on a lot of blogging sites so prevent bots from spamming the comments. Google also uses it for Gmail to prevent bots from signing up spam accounts.
Anyway, back to the topic, I was thinking whether I can develop some sort of algorithm to beat even the most advanced captchas currently deployed on the web. I mean, if humans can read them, then there has to be a way to do image analysis to recognize text, right? Maybe I should try making such a program, and then selling it on the underground markets in Russia. I’m just joking — about the selling part.
