Week 8: Reinvention of the Wetware

If there was anything which has bent Time's Arrow into Time's Circle, it is the newest thing: Crowdsourcing. (Or as I like to call it, "Crowd Computing".) Poor Carnot! He must be rolling in his grave, although the free energy nuts will be loving it.

Years ago there existed roomfuls of individuals, many of them women, armed with pencil, paper and a love of arithmetic. These people were called "computers", and were employed to perform the many calculations required by science of the day. You see, before the advent of the electronic gadgets we know today as "computers", human brainpower was all we had. And science needed it -- lots of it.

When ENIAC came along, everyone expected these wondrous electronic gizmos would soon be powerful enough to solve all of Humanity's problems. Feed information in, press some buttons, and get the answer out with a "ding".

Now in the early 21st century a simple household PC can perform, in one second, more calculations than all the human brains ever have in the history of pencils. And yet, in an irony of astronomical proportions, even our most humongous petaflop machines are outclassed by a child or an insect on something as pedestrian as seeing things. This is why, 120 years after Karl Benz, cars still can't drive themselves.

It's this "fuzziness" that neural networks do so well, which is required for many of the tasks that are the goals of AI and which gum up even the biggest logic machines. So what's the solution? Link up a whole bunch of ready-made, ready-to-go neural networks! It's the return of human computers in the form of MTurk and the like, which provide quick, neat answers to the simplest questions. Problems which would leave the entire Google server farm smoking out.

And all it costs is pennies. Literally.

In a New Media Utopia I can picture models like MTurk being used for things like the eradication of spam, perfect internet dating, and the accurate tagging of every photo in existence. Or more seriously, the solution of such science and medicine problems which require a wetware touch. And as Zittrain suggests, all this could be presented to kids as game and done at zero cost.

Hallelujah! But here comes my cynical voice. If New Media is sticking it to the traditional corporate bastards, as described by Deuze, then it could, nay will, result in the opposite effect. I can see traditional corporate entities using the crowdsourcing model to reassert and restrengthen themselves, at very little cost.

There is the obvious use of crowdsourcing's ultracheap labour to produce work -- break the project down into many tiny subtasks, farm them out, and only employ a team to join it all back together. Suddenly whole development teams are no longer required. Project budgets are slashed. We already see a very nice form of this model in Open Source, although that relies on the benevolent effort of many dedicated individuals.

But I can imagine the CEOs will find a far more insidious use for crowdsourcing.

Example: One day, someone published a crack for Blu-ray encryption on Digg. Owing to Digg's popularity-voting model, this information quickly rose to the top rankings and stayed there. Then all hell broke loose. Digg found itself being threatened by the biggest corporate heavies you could imagine: the AACS consortium, which in turn represents the movie studios and film distributors. Facing legal annihilation, Digg had no choice but to capitulate and pull the offending material.

But then there was a backlash by Digg's user base. The crack codes were inserted into things like forum messages, where they grew virally and couldn't be eliminated. In the end, here is what Digg's management said to their users:

"...after seeing hundreds of stories and reading thousands of comments, you’ve made it clear. You’d rather see Digg go down fighting than bow down to a bigger company. We hear you, and effective immediately we won’t delete stories or comments containing the code and will deal with whatever the consequences might be."

So why the cynicism? Isn't this Power to the People? Yes and no. Sites like Digg exist on a model of popularity and rankings. But what if Digg's user base had thought to themselves, "Hey AACS is right! Using Blu-ray crack is piracy!" and voted the encryption information down? Then the cracks would have stayed buried. And this, I foresee, is exactly where crowdsourcing comes in. How hard would it be for AACS to simply employ an army of MTurk wet-bots to log into Digg and vote the crack down? At pennies on the dollar? By extension any corporation, facing something it wants buried or something it wants done, could utilise crowdsourcing to make it happen at little cost. Suddenly the our New Media model is not so democratically flattened as we thought. And as Zittrain says, the crowd would not even know what they are doing, even if they are from Sudan.

(Oops sorry for the long blurb! Just I had a bit to say on this issue.)