In early 2002 i suggested that we could solve some computer problems and south africas street-kid problem by setting up a network of street=kids with basic education to handle tasks computers still struggled with. At the time we were concerned with low-false positive, agentless remote detection of defaced web-sites, but also ran into the idea when we first built e-or, our early web application scanner. I suspect i didnt broach the subject with enough sensitivity (and in retrospect suggesting that remote controls for automatic gates could be replaced by 2 low cost street-kids (one as a spare)) might not have helped my cause..

I recall the meeting ending badly with Roelof screaming that i never take anything seriously..

Today i learned about the .  service. MTURK simply attempts to solve problems difficult for computers to solve by enlisting the help of humans (and rewarding them with money in amazon accounts). The service, as they point out is named after the chess playing automaton that shocked the world in the 18th century (it simply hid a chess master inside).

The idea is a cool one.. (if nothing else, ponder captcha solving) with an API being handed to prospective developers:

-snip-

All software developers need to do is write normal code. The pseudo-code below illustrates how simple this can be.

read (photo);
photoContainsHuman = callMechanicalTurk(photo);
if (photoContainsHuman == TRUE) {
acceptPhoto;
}
else {
rejectPhoto;
}

-snip-

If this were closer to April i woulda considered it an April fools joke, but like the streetkid idea.. i also suspect it fill a much needed gap till our algorithms catch up..