I ran into an interesting project this morning called reCAPTCHA. In the spirit of distributed computing solutions, such as folding@home, it tackles a difficult problem by splitting it up and farming the pieces out. What makes this interesting is that instead of having computers solve the problem, people do.
ReCAPTCHA actually tries to solve two problems. The reCAPTCHA project pipelines the unrecognizable words from a book scanning OCR effort into a freely available web service for verifying your humanity, a CAPTCHA. Instead of each CAPTCHA puzzle being a necessary but regrettable waste of human effort, reCAPTCHA harnesses this otherwise lost resource. How brilliant is that!
The web service looks very interesting to me. I’m due to revisit a submission form soon that contains a CAPTCHA that I wrote several years ago that I know has been broken. The system contains a very extensive blacklisting system, so the weakness of the CAPTCHA has never been enough of a problem to warrant its replacement, but I’m curious to see what difference this service will make.
Feed
Subscribe to this blog…
I think the new wave of useful captchas is really fascinating. This one, in particular, is a great sounding idea. The problem I see, is that there is no context to determine what the word actually is. So many of the words that come up are impossible to type correctly. And there is little in the information about the captcha itself that explains to someone what the heck they are looking for. It says “type the two words” which would tell me to type exactly as I see them. Which is actually wrong for what we are trying to achieve. I would pass the captcha but send back a false positive on the OCR word.
reCAPTCHA is little bit over-styled…
It will be interesting to see reCaptcha because I steered away from using captcha in web development a while ago preferring instead to write user editable humaniser questions.