Slash Boxes

SoylentNews is people

posted by LaminatorX on Tuesday January 27 2015, @02:34AM   Printer-friendly
from the croud-fleecing dept.

It turns out that while you're proving to the web server you're a human, you might also be pitching in to provide one of Google's services to its corporate customers. A woman filed a class action lawsuit against Google last Thursday in US District Court in Massachusetts, alleging that Google's reCAPTCHA service has harvested unpaid image-to-text transcription work from millions of web site visitors. Google markets reCAPTCHA as a service to web site owners; its customers include Facebook, Twitter, and Ticketmaster. Like other CAPTCHA implementations, reCAPTCHA challenges site visitors to type in the text corresponding to a visually distorted word. But reCAPTCHA differs from the others in that its images often contain two distorted words, as noted by the civil complaint:

One of those words is a “known” word, which the website user must enter correctly to access the website as a security measure. That is, because Google already knows what word is being displayed in the first distorted image, if the user enters the word correctly, Google knows the user is likely to be a human, and thus permits the users to continue using the website...

The other of the two words, however, serves no security purpose. The second word is an image with text that Google is attempting to transcribe. The sole purpose of the second word is to require the user to read and transcribe the word for Google’s commercial use and benefit, with no corresponding benefit to the user.

The lawsuit notes that Google makes use of optical character transcription for its own products such as Google Books and Street View, and also provides an archive digitization service to newspapers, including the New York Times.

This was apparently never a dark secret; the use of reCAPTCHA to "crowdsource" digitization of old printed materials was publicized as a feature by both Luis von Ahn (who invented reCAPTCHA as a graduate student at Carnegie Mellon University) and Google (who acquired the reCAPTCHA technology in 2009):

reCAPTCHA technology was developed not merely with an eye toward improving cyber security, but also as a way to harness and reuse the collective human time and mental energy spent solving and typing CAPTCHAs—a concept von Ahn has dubbed “human computation.” By constructing CAPTCHAs using words tagged as unreadable in the digitizing of books and other printed material, millions and millions of cyber users play a part every day in the digitization and preservation of human knowledge by transcribing words. Tests have shown that reCAPTCHA textual images are deciphered and transcribed with 99.1% accuracy, a rate comparable to the best human professional transcription services.

This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
  • (Score: 2) by c0lo on Tuesday January 27 2015, @02:48AM

    by c0lo (156) on Tuesday January 27 2015, @02:48AM (#138401) Journal
    'nough said.
    Starting Score:    1  point
    Karma-Bonus Modifier   +1  

    Total Score:   2  
  • (Score: 2) by Zyx Abacab on Tuesday January 27 2015, @03:21AM

    by Zyx Abacab (3701) on Tuesday January 27 2015, @03:21AM (#138409)

    4chan already got on that with the Renigger Project []. Unfortunately, Google caught on and adjusted the algorithm - especially where that particular word is concerned.

    Doesn't stop me from putting some random word in every time, though.

    • (Score: 0) by Anonymous Coward on Tuesday January 27 2015, @02:53PM

      by Anonymous Coward on Tuesday January 27 2015, @02:53PM (#138555)

      A while back google stepped up the difficulty of reCAPTCHA to be annoying. Since then I have always written "fuck" for the unknown word.

  • (Score: 0) by Anonymous Coward on Tuesday January 27 2015, @05:31AM

    by Anonymous Coward on Tuesday January 27 2015, @05:31AM (#138444)

    I've been experimenting a bit with this since at some point I got lumped with decoding house numbers from fuzzy street view photos. So far my experience is that google has no penalty for constantly getting it wrong. I've not been dropped back to having to decode two known things, despite being deliberately bad at interpreting bad photos. I would have thought that they'd present each photo to N people (N >= 3), and when ceil(N/2) provided the same value accept that as the truth, and also note any who got it wrong as not-so-reliable-at-this-ocr-stuff...

  • (Score: 2) by threedigits on Tuesday January 27 2015, @08:02AM

    by threedigits (607) on Tuesday January 27 2015, @08:02AM (#138468)

    Been doing it for years.

  • (Score: 2) by ikanreed on Tuesday January 27 2015, @02:47PM

    by ikanreed (3164) on Tuesday January 27 2015, @02:47PM (#138548) Journal

    What you douchebags don't realize is that if recaptcha fails, they're going to start doing something like Solve Media, where you have to put in a word from an advertisement to continue. I'll take old books over that any day.

  • (Score: 0) by Anonymous Coward on Wednesday January 28 2015, @09:11AM

    by Anonymous Coward on Wednesday January 28 2015, @09:11AM (#138810)

    On 4chan we used to call that "operation reNigger", systematically replacing the unknown word by "nigger".