The Economist ran a nice story about one of the projects under my jurisdiction at Whirlpool.

I cannot honestly claim a lot of involvement in this success story, but I do claim the basic idea of a decentralized mechanism for power grid monitoring. My basic idea was to avoid the clunky and not really economic central command structures that were introduced so far. Rather, I suggested an approach where each appliance would sense the local load on the net as it experienced it. This information is used to decide when to delay activity, and when to resume. Result: a massively distributed control architecture where local decisions tend to support the global/regional goal:

• There is no worry about a central system knowing appliance states (and thus snooping on users).
• No command&control system that would take control away from the user/owner.
• No central station to fail.
• No expensive infrastructure to build, local information is enough, and the cost for the sensor can be justified by the avoidance of brownouts (something residents will really support).
• Oh, and it scales rather well, with every new appliance with this capability adding to the power of it. Not quite a network effect, but better than nothing

http://www.economist.com/search/displaystory.cfm?story_id=11482511