My OperaCAPTCHA
Thursday, July 26, 2007 3:09:55 AM
Virtually every time you sign up for a new account with some online service, they use CAPTCHA to do a simple "human check." If you're not sure what CAPTCHA is, they're these handy little images like this:
The text is slightly distorted to allow humans to read it, but it's supposed to stump computers. The only problem is that with increasing artificial intelligence, computers can read these images just as well as humans.
Personally, I think that the problem is that CAPTCHA, by it's very nature, is too simple. To really trip up a computer program, you've got to pose challenging questions.
For instance, here's a CAPTCHA I created:
Please enter the solution to this equation:
<input>
Ok, ok... maybe this one's a tad too difficult (maybe it'll stop more than just bots
). All joking aside, a new type of CAPTCHA has been developed which is ingeniously simple for humans to understand, yet hard for bots -- it uses images already proven to stump the best OCR (Optical Character Recognition) software.
"reCAPTCHA" works in conjunction with the Web Archive's "Project Gutenberg" -- the Web Archive scans thousands of pages from old books, then tries to convert them into digital media. However, there's always several words on each page than the OCR software gets hung up on -- before, it had to take a take a wild guess that would later be corrected by volunteers. Now, the Archive sends images of these words to the reCAPTCHA project, which presents them as CAPTCHAs.
Here's how it works: people are presented with an image with two words in it -- reCAPTCHA knows what one word says, but not the other (the user doesn't know which word is known or unknown). If the user correctly enters the already-known word, access is granted. Furthermore once several different users enter the same thing for the unknown word, it becomes one of the "known words", and is sent to the Archive.
This is such a brilliant idea -- it's simple for humans to read, proven to stump computers, and helps digitize books. So, maybe my CAPTHA idea isn't necessary after all
.
When I'm feeling like a 30-second break, I go to reCAPTCHA's "Learn More" page and fill in 10 CAPTCHAs. I'm not going to make a huge difference, but I'll help digitize a couple pages without sacrificing hardly any time.
In other news, I just finished my final exam for my summer class. Yay! I'm finally on summer break for about a month before school starts again.
The text is slightly distorted to allow humans to read it, but it's supposed to stump computers. The only problem is that with increasing artificial intelligence, computers can read these images just as well as humans.
Personally, I think that the problem is that CAPTCHA, by it's very nature, is too simple. To really trip up a computer program, you've got to pose challenging questions.
For instance, here's a CAPTCHA I created:
Please enter the solution to this equation:
<input>
Ok, ok... maybe this one's a tad too difficult (maybe it'll stop more than just bots
). All joking aside, a new type of CAPTCHA has been developed which is ingeniously simple for humans to understand, yet hard for bots -- it uses images already proven to stump the best OCR (Optical Character Recognition) software."reCAPTCHA" works in conjunction with the Web Archive's "Project Gutenberg" -- the Web Archive scans thousands of pages from old books, then tries to convert them into digital media. However, there's always several words on each page than the OCR software gets hung up on -- before, it had to take a take a wild guess that would later be corrected by volunteers. Now, the Archive sends images of these words to the reCAPTCHA project, which presents them as CAPTCHAs.
Here's how it works: people are presented with an image with two words in it -- reCAPTCHA knows what one word says, but not the other (the user doesn't know which word is known or unknown). If the user correctly enters the already-known word, access is granted. Furthermore once several different users enter the same thing for the unknown word, it becomes one of the "known words", and is sent to the Archive.
This is such a brilliant idea -- it's simple for humans to read, proven to stump computers, and helps digitize books. So, maybe my CAPTHA idea isn't necessary after all
.When I'm feeling like a 30-second break, I go to reCAPTCHA's "Learn More" page and fill in 10 CAPTCHAs. I'm not going to make a huge difference, but I'll help digitize a couple pages without sacrificing hardly any time.
In other news, I just finished my final exam for my summer class. Yay! I'm finally on summer break for about a month before school starts again.
How to use Quote function: