RRbanner.jpg
logo

need more stuff?

December 19, 2009

A Modest Proposal: CAPTCHAture

Jim Hanas
While I wasn't actually at MediaBistro's eBook Summit last week, I gleaned from the tweets that I was not alone in being surprised to find out that CAPTCHAs -- those annoying little words you have to type to prove you're human -- are being used to transcribe old newspapers and books for Google, which recently acquired the company that came up with this ingenious crowd-sourcing concept.

There is poetry, to be sure, in the fact that you now have to be willing to work for free on the Internet (even if unwittingly) to verify your humanity -- since even robots are too smart to join blog farms -- but it occurred to me that there might be even greater literary possibilities here. Why stop at a few distorted words? Perhaps we should require customers to write sonnets or short stories or monographs, just so we can be really, absolutely sure that they are human and not machine.

And instead of these grainy words, they can be given trending topics out of which they'll be be required to fashion their works. "To help fight spam, please write a novelette about Tiger Woods and mesothelioma in the space provided," a typical prompt might read. Then the work-product would stack up pretty quickly, not like this meager transcription plan. It would fill volumes and shelves and whole libraries that could be merchandised at great profit -- even at e-book rates -- since it would all be produced free of charge. A special, iTunes-like service could be set up to allow readers to access this vast store of CAPTCHAture at a reasonable flat rate. And if they forget their password? No problem. "Please paste a roman a clef about Barack Obama and vampires in the space below." And the wheel will turn and turn, smoothly and delightfully spam-free.

Just a thought.

Comments

YAY! I'll put that on my list of advantages to being vision impaired. Now, if I could just get reimbursed for all the time I've wasted trying to register, only to come across one of those things.

Your ideas are intriguing to me, and I wish to subscribe to your newsletter. As they say. I've always wondered why some of the phrases are so peculiar; I never suspected it was because someone was making money.

I'm partial to the real-sounding nonsense words YouTube uses. They order me to kill.

The obvious work-around for having one's labor exploited without compensation in this way is to correctly transcribe one of the two words shown, and incorrectly transcribe the other. There is a 50-50 chance that the word you correctly type is the "test" word with an already known answer, so you will pass the bot test. In that case your wrong answer will be recorded as a correct answer for the "research" word whose answer is not yet known. I assume they repeat research words several times before deciding upon a correct answer, but at least that way you can decline to add value if you prefer.

Publishers lack sufficient data to backup most of their claims. They can———t afford quality marketing research. They———re just making guesses based on their experiences and expertise, for better or for worse. This sort of guesswork does a good job in stable business climates but is terrible in a dynamic situation.

Post a comment

Powered by
Movable Type 3.2