I think I’ve already ranted about the inaccessibility of those CAPTCHA images that are used to distinguish humans from bots– assuming, of course, that those humans have perfect eyesight and no perceptual processing difficulties. But as Blogger founder Evan Williams discovered last month, sometimes they’re even illegible to the target audience as well.

Even when they’re not that distorted, there are some usability issues that people tend to forget about when writing CAPTCHA scripts. How, for instance, is one supposed to distinguish ‘I’ from ‘l’ from ‘1’, or ‘O’ from ‘0’, without any sort of cues as to the typeface being used? And sometimes the extra distortion only makes things worse; a ‘3’ might look like an ‘8’, or a ‘g’ like a ‘9’.

And that, of course, is assuming one actually has a reasonable enough knowledge of the Roman alphabet to identify the distorted characters at all. Otherwise, one will be in the same boat Jim O’Connell of wirefarm.com was in when a Korean site forced him to enter a CAPTCHA in the Korean alphabet…