An inspiring new view of reCAPTCHAs

Recently I got an entirely new, inspiring, view of what reCAPTCHAs really are, after watching the TED Talk below:

Luis von Ahn: Massive-scale online collaboration
http://www.ted.com/talks/luis_von_ahn_massive_scale_online_collaboration.html

Until then, I used to find CAPTCHAs a real PITA, an annoyance. Now I find them inspiring, and the tecnology behind it to be awesome. I am so amazed with it that I felt like sharing that with you.

I won't spoil the surprise... please watch the video above and we'll talk after the break.

One thing is sure: I'll never fell annoyed by reCAPTCHAs again!

Thanks for the post. I noticed the transition from one word to two words, and apparently one of the two was easy for some reason. Now I know why. At this rate we will be translating alien languages and communicating with ETs pretty soon :slight_smile:

I knew that in reCapcha, that one word didn't matter and I could recognize that word. So, if that word had a significant amount of letters, I would use shorter words to get past the capcha.

Now with new found knowledge of why this is so, I still don't think I will change my habit..

Thanks for the link.

I did a reCAPCHA yesterday that involved the word Bragg. That is an important figure in my field. They must be digitizing old science books. I won't do the shortcut. I know it too.

Unfortunately the spammers are now paying people to fill out the captchas for them.

That video made my day. I thought the transition to two-word captchas was another annoying idea by some dickhead. Apparently not.

i watched the video... now i know who's to blame! lol

captchas can be annoying if you used yahoo a lot, it made little impact on spam bots as people pointed out, paying people to fill them out seems to be enough...

i still believe there's a better solution out there to authenticate yourself as a human, i once came across a captcha as a picture puzzle, re arranging the squares to make the image complete....

when i think of the better solution i'll let people know lol... the 2 word system just means you use ocr on both words hope it gets the correct word of the 2 and away you go (seems less secure than 1 word to me)

I wonder if they deliberately put a line through the "real" one (the one from a book) to make it look more "captcha-like".

I think captchas fall down in this way:

  • Some spammers need a lot of captchas solved cheaply.
  • They set up a porn website.
  • Before you can download an image/video you have to enter a captcha (allegedly to stop people data-mining their site)
  • The captcha is actually copied from a web site that they want to authenticate themselves to (for spamming purposes)
  • The person wanting the porn solves the captcha for them. They get their porn as a reward.
  • The site use the solved captcha to spam someone

In fact the "double captcha" works well here. They would use one generated by themselves to make sure that people don't get porn for nothing (which would mean they don't have to work too hard at the captcha). The other one is the one they really want solved.

After watching the CAPCHA video, I was left with one major question.. How does the computer know that the capcha that your translating from a book is translated correctly? I guess next time I read a book online and see a word that clearly doesn't fit I can blame some dyslexic kid trying to sign up for yahoo! XD

robotlover17:
After watching the CAPCHA video, I was left with one major question.. How does the computer know that the capcha that your translating from a book is translated correctly? I guess next time I read a book online and see a word that clearly doesn't fit I can blame some dyslexic kid trying to sign up for yahoo! XD

The "HCR" (human character recognition!) is not done based on just one identification/recognition. The same word is sent to several reCAPCHAs, and the word more recognized is the one accepted as the correct one. This way, even if one kid is retarded, his recognition of that word won't count.

Ohhh that makes much more sense thanks for explaining! :slight_smile: He also talked about the problem of a lack of motivation.. Why don't they just create a website where people can go and solve as many CAPCHAs as they want and recieve a certain incentive for it.. For example, they could say if you solve x number of CAPCHAs they would give a few options of books to download free offline. Makes sense, right? XD

As I indicated earlier, porn websites give you a reward for solving captchas.

Indeed any website which provides something you want (eg. a pirated movie) after you type in a captcha does the same.

But if you have free time you're willing to spend on noble tasks, there are plenty of online scientific projects in which you can participate, like classifying galaxies (first one of the kind), searching for exoplanets or gravitational lenses. Human are much better at recognizing pattern that computer programs (for the time being), no prior knowledge is required to take part.

Real scientific papers have been written thanks to these datas (just check "galaxy zoo" on arXiv, for instance).