The problem of unreadable CAPTCHAs

A CAPTCHA is usually a severely contorted image of random characters that you are supposed to key in correctly to be able to get into a website. The idea is to thwart automated scripts/programs - the sort of stuff that spammers (among other bad guys) use to create hundreds of fake accounts for spamming. Now as optical character recognition (OCR) becomes sophisticated a simple image or one that is not significantly distorted, does not present a sufficient challenge to an OCR program. As a result we are seeing increasingly more distorted/noisy CAPTCHAs that are beginning to push the limits of even human readability - like the one below from Yahoo India’s e-mail sign-up page:


And then there are CAPTCHAs that just want to mock you, like the one blogger so affectionately threw at me this morning:


Since CAPTCHAs cannot be read by screen readers, a lot of sites also include an audio version of the CAPTCHA. I hadn’t heard one till recently (I happened to click on the accessibility icon accidentally), but what I heard was stuff of nightmares! Turn down your speakers a bit this stuff is not pleasant:

Obviously, the intent here is to stop sound recognition programs in their tracks by introducing a lot of background noise. The overall effect however brings images of prisons, misery and torture to one’s mind. (Or may be I watched one war movie too many this year!)

And then there is this minor matter of spammers with an army of CAPTCHA cracking humans at their disposal. Or may be that explains why CAPTCHAs are becoming increasingly painful even for legitimate users.

November 14, 2008