Human Intelligence Validation
I just finished reading my good friend Brian’s post about “captchas”, or the annoying things where you type out the garbled letters in an image to prove that you are not a computer program when you sign up for things. Recently, I also read about a new service from Amazon called Mechanical Turk. Basically, MT allows developers to submit “human intelligence tasks” to the system for people browsing to solve. These are generally easy tasks like “identify the best photography of this building” that would be incredibly difficult for a computer to do. On Mechanical Turk, the people who complete the human intelligence tasks get a couple cents per task to spend on Amazon. Interesting idea.
So, after reading Brian’s post, I thought to myself, “Why don’t we just combine the two ideas?” For people signing up for a new service, make them solve a human intelligence task that somebody else needs solved. If they answer in a semi-intelligent manner, then they were probably a human and you can go ahead and let them in. This way, instead of giving people impossible and pointless tasks like deciphering garbled letters, we can give them tasks more suited to human analysis AND create value for people who need tasks solved at the same time. Plus, if anybody cracks this method of human verification, they will have effectively created an efficient algorithm for solving human intelligence tasks quickly, which would mean that they can quit trying to hack web logins and go become rich beyond their wildest dreams.
It’s a win-win situation! Amazon … I’m looking at you.
Comments (3 So Far)
1
Grant Hutchins says:
Well, with a captcha, the computer already knows which is the correct answer. With Mechanical Turk, the problem is that the computer doesn’t know the correct answer, which means as an authentication system it doesn’t hold any water. Plus, you’d get a ton of bots submitting false data.
2
Sean says:
Yeah, I suppose you’d have to make it wait until the system got confirmation of a “good answer” from the submitter of the human intelligence task. So it would be slower, but still kinda cool.
3
Cousin Kate says:
Sean,
I can just see the wheels in your head spinning on that one, and the rest of the family is all scratching their heads saying, “duh I don’t get it!” and so I will tell you the standard family line, “Wow Sean, I don’t really know what you are talking about, but it sounds like a great idea!” Ha Ha!!!
Make a Comment