OK, here’s a cool CAPTCHA tool I can actually get behind
Every now and then a smart anti-SPAM solution comes along that’s actually worth taking a close look at. ReCAPTCHA™ is one of those tools.
Everyone who reads my blog already knows that standard CAPTCHA utilities have been hacked. And you already know that the best form-based anti-SPAM tools require a modicum of human intelligence to unlock the comments form and allow a comment to be posted.
And that’s why I’m so excited about ReCAPTCHA. Not only is it human intelligence-based, but it’s free as well. And to make it even better, if you use ReCAPTCHA, you’ll actually be helping to spread literacy around the world.
Here’s how it works:
The people behind the ReCAPTCHA project have taken hundreds of old books that are no longer protected by copyrights and scanned them into image files. Their goal is to take the scanned images and convert them into actual text so they can be stored, read and searched by anyone who wants to read those books.
Since Optical Character Recognition (OCR) is an inexact science, and many of the words in these books do not translate properly when OCR technology is used, the project owners are counting on people like you and me to do the translation for them.
ReCAPTCHA software displays random word images taken straight from the scanned books. Every time someone types a displayed word into the CAPTCHA form, another word from a particular book has been translated!
Now here’s my idea:
ReCAPTCHA is certainly cool, but it doesn’t have all the features and power of Advanced Textual Confirmation (ATC). So I’m willing to create a special version of ATC that uses the ReCAPTCHA image database. That way you can help with this worthwhile literacy goal without losing the security that ATC provides.
Interested? Let me know.
September 7th, 2007 at 2:19 pm
So wait, the CAPTCHA is an image of a word that you DON’T know a textual value to?
Haha, sounds a little broke to me.
September 8th, 2007 at 10:40 pm
No, no. The captcha consist of two words: for one the textual value is known, for another — no. You have to correctly decrypt the first value. The answer to the second is remembered. When statistics is collected, the second word will also have the known value.
October 18th, 2007 at 7:59 pm
I thought of this about a year ago but did not have the ability or the resources to put it into action.
I’m excited that it has been done.