NOVA Science Now! aired a program a couple of weeks ago that included an interview with Carnegie Mellon University computer scientist professor and creator of
reCaptcha, Luis von Ahn.
I'm sure you've seen a captcha request before. It's the little box at the bottom a form that asks you to type in a string of letters in order to prove that you're a human and not an evil machine attempting to do bad things.. We use them on Intershame. They look like this:
The NOVA piece explains that Professor von Ahn was troubled by the considerable amount of time human beings were wasting filling out captcha forms on the internet, so he devised a way to make use of the time we spend filling out these mundane captcha requests and sold his solution to Google. His solution is what we know as reCaptcha. You've probably seen it before. Below is an example of what von Ahn's reCaptcha looks like:
The reason reCaptcha asks for two words instead of one is the secret behind how the reCaptcha system gets humans to perform meaningful work. You see, machines are attempting to
digitize hundreds of thousands of books. When the machine is scanning the pages of these books, there are times where the machine is unable to figure out what a word is. reCaptcha takes these unknown words and uses them as the puzzles humans must solve. In order to verify the human solved the word puzzle correctly, reCaptcha requires you enter two words - a known word and one of these unknown words. If you correctly enter the known word, reCaptcha will assume you correctly entered the unknown word. So, let's give it a try...
That worked! So my solution will go into the Google Books repository, right? Does this mean I can look forward to reading the digitized version of
Chicago and Its Fuck-pools of Infamy? Not quite.
The reCaptcha system takes the results from many different users and gives each word solved a confidence rating. When a word reaches a high enough confidence level, that word is accepted into the system. So in order to pull off a reCaptcha hack, you'd have to organize a large group of people, decide on what word you want to attempt to change and somehow reload the reCaptcha over and over until the word you're attempting to change is displayed.
I don't know of any
group that could pull that off. ;)
