New Scientist (October 22-28, 2005) reports that Australia and the EU are funding a vast sensor network (four to five square kilometers--which, in real units, is roughly 1.5 square miles) to be deployed on the Great Barrier Reef, which is either (depending on whom you ask) perfectly healthy or teetering on the edge of a fatal eco-coronary. The sensor network will help resolve this dispute by tracking the pollution that reaches the Reef as well as (through separate coral-mounted sensors) salinity, humidity, and light levels.

So What?
This is the biggest sensor installation I've heard of to date (excepting the world's cell phone network, which senses the locations of hundreds of millions of unsuspecting customers), and it's yet another indication that distributed sensors are coming into their own.

Amazon.com recently launched Mechanical Turk, a service named after the 18th-Century chess-playing "automaton." Turk solves a perennial problem: Computers are inadequate for many tasks--tasks that human beings are often quite good at. Billed as "artificial, artificial intelligence," the Turk is a system that allows programs to "invoke" Turk participants for any data-based task that can be done by an untrained human being. For example, given a picture of a landscape, a Turk participant might click "Yes" if there's a person in it and "No" otherwise. This sort of photograph interpretation is trivial for a person--but still impossible for computers. Amazon.com publishes an API with which to invoke Turk participants' services. (Come to think of it, geeks have long wished that human beings had an API.)

So What?
Mechanical Turk reminds me of the companies that paid you for watching advertisements (none of those companies, to my knowledge, survived the dot-com pop). The problem was that people willing to watch advertisements for money were not exactly a high-value demographic (duh). Mechanical Turk doesn't suffer from this problem: It only wants your brain, not your wallet, so it doesn't matter how poor you are. (Indeed, given the value of a dollar in poorer countries, Turk's most ardent participants will probably be in developing countries. Three cheers for a system that pushes wealth down the income ladder.) One potential problem (at least in the United States) is that piecework (which is effectively what Turk offers) is legal only if the effective wage equals (or exceeds) the minimum wage. But net, I think it's a good idea. In fact, from now on, I may get a Turk to blog for me.

In April of 2005 (I'm still catching up on my reading), Sony patented a technique for using ultrasonic waves to stimulate various centers of the human brain in order to produce extremely high-fidelity sensory impressions. The idea is apparently to build really, really immersive environments for games.

So What?
The problem here is one someone once raised regarding Star Trek TNG's "holodeck." To wit: If I had a fantasy machine that powerful, I wouldn't spend my time matching wits with a simulated Sherlock Holmes--I'd spend it playing chess with Aishwarya Rai.

Accenture Technology Labs has just put a new demo on the Web. Pickberry, a project we did to equip a California vineyard with a variety of sensors, wireless networking, and data analysis software, is the oddest use of these technologies I know of to date. We did it as a way to demonstrate some of the more surprising uses of sensor-based systems.

So what?
The point here, of course, is that as the cost of sensors and wireless networking drops to zero, and as data analysis software becomes more sophisticated, they'll be used for an increasing variety of applications. Perhaps the most important implication of Pickberry is that it should result in more consistent wine from season to season, earlier detection of soil problems, and more rapid response to pest infestations. Less importantly, it will provide new opportunities for practitioners of oenophilia to exhibit their expertise ("Please bring a bottle of Chateau de Pretentious made with grapes from the north side of the vineyard and monitored by Type 6 humidity sensors--the Type 7s they use today are vastly inferior, you know.").