Amazon Mechanical Turk
Jul. 15th, 2006 12:34 pm![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png){kind=small}
yendiAmazon has a new service (in beta testing) called Mechanical Turk.
The idea behind it is nifty -- basically, it's "artificial artificial intelligence," or a project that seeks to gather data that simply cannot be automated (and is thus named after the famous chess playing "automaton" written about in Tom Standage's excellent book). Basically, it allows developers to create programs that require human input at one point, but which can then work with the data. From an average joe point of view, what it amounts to is a service that will pay end-users to verify data (such as checking if a restaurant is open, or offers certain items), or provide feedback (short surveys, essentially). It's a nifty idea, and worth taking a look at.
In practice, it's not there yet -- the pay rate is so tiny (1 cent per answer for easy questions, $.08 or so for ones that require, frankly, too much effort to be worth the time), that I just don't see anyone leaping on board yet. As a user, you're not going to make anything resembling decent money (and by decent, I mean, "enough to buy a used paperback," not, "enough to feed yourself for a month") participating.
Now, it's in beta, of course, and the money that a user can make are determined by the folks who submit requests. If the idea's actually viable, the marketplace will eventually bear it out, and there will be decent offers. So I'll keep checking on it. But in the meantime, the $.12 I made this morning from filling out "top three" lists about college memories isn't exactly going to change my life.
The idea behind it is nifty -- basically, it's "artificial artificial intelligence," or a project that seeks to gather data that simply cannot be automated (and is thus named after the famous chess playing "automaton" written about in Tom Standage's excellent book). Basically, it allows developers to create programs that require human input at one point, but which can then work with the data. From an average joe point of view, what it amounts to is a service that will pay end-users to verify data (such as checking if a restaurant is open, or offers certain items), or provide feedback (short surveys, essentially). It's a nifty idea, and worth taking a look at.
In practice, it's not there yet -- the pay rate is so tiny (1 cent per answer for easy questions, $.08 or so for ones that require, frankly, too much effort to be worth the time), that I just don't see anyone leaping on board yet. As a user, you're not going to make anything resembling decent money (and by decent, I mean, "enough to buy a used paperback," not, "enough to feed yourself for a month") participating.
Now, it's in beta, of course, and the money that a user can make are determined by the folks who submit requests. If the idea's actually viable, the marketplace will eventually bear it out, and there will be decent offers. So I'll keep checking on it. But in the meantime, the $.12 I made this morning from filling out "top three" lists about college memories isn't exactly going to change my life.
![[identity profile]](https://www.dreamwidth.org/img/silk/identity/openid.png){kind=small}
(no subject)
Date: 2006-07-15 04:40 pm (UTC)(no subject)
Date: 2006-07-15 04:45 pm (UTC)(no subject)
Date: 2006-07-16 12:46 am (UTC)(no subject)
Date: 2006-07-16 04:52 am (UTC)Neat concept though. And it WAS much better in the beginning, until people from luelinks figured out how to program a bot to do some of the work and got like HUNDREDS of HITS -- then they re-engineered it and there's less easy work to do.
I hope that changes.