ReCaptcha: Stop Spam, Help Digitize Books

TangoWhiskey

Touchdown! Greaser!
Joined
Feb 23, 2005
Messages
14,210
Location
Midlothian, TX
Display Name

Display name:
3Green
You know those funny distorted images (captcha's) that you have to figure out to log into so many sites? I noticed two major sites using a new kind of Captcha... and it's free for web site owners... and it solves a VERY unique business problem in the process. They serve up two words (one known, one unknown) that you must type in. The unknown word comes from scanned texts where the optical character recognition couldn't (with certainty) determine what the word is. So, by solving the captcha, you help digitize old books... after all, humans can figure out the word images better than computers!

It looks like this:

recaptcha-1.gif


Notice that their implementation also solves a classic problem with Captchas: this version has a work around for vision impaired people--an auditory method of compliance to meet ADA Web Accessibility standards.

Check out this site and see what the students at the School of Computer Science at Carnegie Mellon University have come up with:

http://recaptcha.net/learnmore.html

I'd love to see POA use this one!!
 
Last edited:
Wow, that's cool - I never saw it before today, when I finally signed up for Twitter. They're using it. I saw something about books but I didn't know the significance. A very creative solution to the problem!
 
I personally hate captchas, so I've taken bits and pieces from my scripts on www.spamfreetext.com and used them on other contact forms. So far, I've had considerable success.

The ones that seem to work best are actually the simplest ones:

1. IP matching. The IP address that connects to the mailer script must agree with the one that accessed the form page during the session;

2. The timer script. Submissions that are entered too quickly for a human to have completed the form are assumed to be robots and are silently discarded;

3. A simple preg_match that checks fields that should not contain hyperlinks for "http" and "www," and silently discards those messages.

-Rich
 
Back
Top