CAPTCHAs have become very popular among the blogger and even among the wiki crowd to prevent spamming by automated bot scripts. Wikipedia's article on CAPTCHAs is one of the few places where accessibility problems of the CAPTCHA system are discussed.

However, the only recommendation Wikipedia gives is to resort to audio as opposed to visual CAPTCHAs.

But there is a far better solution. You don't need images, you don't need audio files, all you need is a little imagination, plain text and a tiny little script in the background. Contrary to popular believe, it doesn't take multimedia to demonstrate how dumb computers really are. They may be able to process plain text, but they aren't able to understand it. So why not simply ask them trivial little questions?

• What is the first letter of this question?
• How many letters are in the word 'four'?
• Tell me any number between 1 and 10!

Questions like these have any kind of bot script completely stumped (unless it already knows them, of course). Yet, screen readers can read them to their users and if you resize them for the benefit of the visually-challenged, they don't get all pixely.

So I suggest that instead of installing a library that draws random letters and digits on a GIF, you simply sit down and come up with 10 or 20 questions that anybody can answer. Should be easy enough.

This idea and the sample questions were taken from the Question Asker Extension for the Oddmuse wiki engine, courtesy of the apparently brilliant Brock Wilcox.

¶ 15:31