Tell me more ×
Super User is a question and answer site for computer enthusiasts and power users. It's 100% free, no registration required.
up vote 22 down vote favorite
3

From what I understand, Captchas are text that have been distorted by the application of filters, noise and other miscelaneous algorithms. Therefore, to find out whether the person's reading ability is that of a person, you compare what they answered to what the known answer is.

Now, reading up on ReCaptcha, it says that the words that are displayed are those that cannot be translated by OCR. In addition, recaptcha is being used to translate those images. How can it tell whether you are indeed right in your reading or are just making stuff up?

If it knew what it said, it wouldn't be used in recaptcha as translation material. If it doesn't know what the text says, then how does it validate your answer?

I'm guessing this is probably some probability based analysis with huge sample sizes before it flags anything as translated.

Does anyone know where the answer to this is?

share|improve this question
2  
Of interest is the 4chan / anonymous prank on the Time poll. "Marble cake, also the game", which exploited flaws in the crowdsourcing verification of the second word. – DanBeale Oct 1 '11 at 11:59
1  
The hack @Dan mentioend: musicmachinery.com/2009/04/27/moot-wins-time-inc-loses – BlueRaja Oct 2 '11 at 11:17

closed as off topic by slhck♦, random♦ Oct 7 '11 at 2:31

Questions on Super User are expected to relate to computer software or computer hardware within the scope defined in the FAQ. Consider editing the question or leaving comments for improvement if you believe the question can be reworded to fit within the scope. Read more about closed questions here.

2 Answers

up vote 33 down vote accepted

Book pages are basically photographically scanned, and then transformed into text using "Optical Character Recognition" (OCR) and fed to the web in the form of an image with one word that is known to the computer program behind reCAPTCHA and one word that is not yet known.

The user then types both words out and if they solve the one for which the answer is known, the system assumes their answer is correct for the new one. The system then gives the new image to a number of other people to determine, with higher confidence, whether the original answer was correct. Therefore, the system is a self-improving service that gets better with time.

http://www.google.com/recaptcha/learnmore

share|improve this answer
up vote 22 down vote

This is why reCaptcha has you enter two words. One of the words is already known, and one of the words is not known. Whether you pass or fail the captcha only depends on how you answer for the word that is known. Your answer for the other (unknown) word will be used, along with other responses to the same word, to turn it into a known word as well.

share|improve this answer
4  
...which is also why over time it gets more and more frustrating to use and convinces you that you're an idiot / robot when you fail it for the 5th time in a row. :-( – Sirex Oct 1 '11 at 3:08
Strange...I have never failed one that I can remember, perhaps just luck on my part. – Paul Oct 3 '11 at 1:15
@Sirex I used to think that, but then I realized that this is only true if the size of the corpus text is constant or shrinking relative to the number captcha entries. The truth is that the corpus text is growing... the question is whether that growth keeps pace with the growth in overal captcha use. – Joel Coehoorn Jan 17 at 15:30
yeah i guess. I've seen plenty of reCaptchas which are just insanely hard. Where even the known word is ambiguous. – Sirex Jan 17 at 18:44

Not the answer you're looking for? Browse other questions tagged or ask your own question.