So much talk about new energy supplies ignores the wisdom we supposedly learned in the ’70s about "negawatts" being the most efficient, effective, and environmentally friendly source of power around.

It’s good to see that we might finally make some progress in this direction, learning to shave demand peaks and save a bundle (and open the way for integrating more renewables into the grid):

In a yearlong trial run that ended in the spring, 200 or so homes on Washington’s Olympic Peninsula engaged in a daily bidding war for electricity. It was a sort of robotic Ebay auction in which the thermostat in one house, say, bid against the clothes dryer in another for scarce electrons. The loser would turn off and wait for prices to drop before jumping back onto the grid. Engineers at the federally funded Pacific Northwest National Laboratory showed that by equipping appliances and thermostats with a few cheap microchips and Internet connections, they could cut peak demand by as much as 50%. That’s a big number, because 8% to 12% of peak demand for power capacity comes during the busiest 1% of hours. Most of the extra supply comes from inefficient gas-turbine generators.

CenterPoint Energy (nyse: CNP – news – people ) of Houston, for example, plans to install 2 million Internet-capable electric meters over the next five years. The utility likes the $120 devices: They eliminate the need for meter readers and contain wireless chips that communicate with Internet-enabled appliances in the home, letting consumers use a simple Web-based program, say, to raise the air-conditioning thermostat when electricity prices rise or turn on the dishwasher in the middle of the night when prices are low. "We’re on the threshold of being able to digitize the system," says Thomas Standish, the head of regulated operations at CenterPoint. The grid "is one of the last things that can be completely transformed by this technology."

That won’t happen overnight. One of the biggest and oldest networks around, the U.S. electric grid seems hopelessly stuck between the 19th and 21st centuries. It’s broken up for historical and regulatory reasons into 8 regional transmission systems and some 130 smaller "control areas." While many systems are computerized, grid operators at the higher levels still communicate largely by phone and fax. Hundreds of thousands of switches and circuit breakers must be operated manually, and the main transmission lines have little instrumentation to monitor the second-by-second flow of electricity from unpredictable new sources like windmills. "If Thomas Edison came back to life, he’d recognize our electric utility system immediately — and that’s not a good thing," says Jesse Berst, publisher of, in Redmond, Wash.

Each year U.S. electric utilities waste tens of billions of cubic feet of natural gas on "spinning reserves," for example, generators that are running below top efficiency so they can supply electricity on a moment’s notice. From 5% to 20% of capacity is in reserve at any given time.

The problem of matching supply and demand will worsen as utilities increase the supply of green energy under mandates like California’s, which requires 20% renewable electricity by 2010. Windmills, for instance, are unpredictable and must be paired with gas turbines that can be throttled up when gusts die down. Utilities need as much as 1 megawatt of spinning reserves for every 2 megawatts of wind power, says Douglas Houseman with CapGemini, a Paris consultancy. To supply the extra juice, utilities turn on inefficient single-cycle gas turbines, which turn 25% of the fuel’s energy into electricity, as opposed to 45% for combined-cycle plants that use turbine exhaust to make steam to run a second generator.

The ideal solution is to counter fluctuations in wind power by changing demand, as in the Washington experiment, but it’s a big job. Houseman estimates the average home uses about 3 kilowatts of electricity at peak hours; General Electric (nyse: GE – news – people )’s most popular windmills generate 3 megawatts. "You’ve got to pick 1,000 homes to turn everything off in — or 10,000 homes to turn something off in — to counter one windmill," he says.

Going to open source was a big step for Itron (nasdaq: ITRI – news – people ) of Liberty Lake, Wash., which controls approximately 60% of the electric meter market and until four years ago was committed to protecting itself with proprietary technology. The change came as Itron’s research group in Paris realized it could exploit inexpensive new wireless chips and open-source software to create a so-called mesh network of electric meters that passed information among themselves, like firemen in a bucket brigade, instead of relying on expensive and less reliable individual connections to the Internet.

CenterPoint figures the project will cost $550 million or so, which it will recover from customers if regulators approve a charge of $2.50 a month over the next 12 years. Consumers could save several times that much if the system cuts peak energy demand, however, since prices at peak hours are set by the least efficient, most expensive generators on the grid. As communications networks become more widespread, utilities will be able to balance flow by harnessing everything from solar panels to back-yard generators in order to supply electricity when and where it’s needed. This country has an estimated 220 gigawatts of what you might call amateur power: emergency generators, industrial fuel cells and other user-owned power plants, compared with 1,000 or so gigawatts of central station capacity. Only 1% of the amateur power is connected to the grid now, but Portland General Electric (nyse: POR – news – people ) in Oregon has hooked up 21 large customers with 43 megawatts of generating power that can be turned on electronically and supplied to the grid.

Factoring in the software and systems to control all those units, utilities would spend $75 to $150 per kilowatt of generating capacity, says Steve W. Pullins, an analyst with consultants Horizon Energy Group in Maryville, Tenn., versus $1,000 or more for peaker plants that can be turned on rapidly to supply peak loads. "But instead of 25 generating assets in your portfolio, now you’re looking at 25,000," Pullins says. "Our traditional, older control systems aren’t capable of handling it."

There are other barriers to transforming the grid across America. GE, which supplies the National Grid operator in the U.K., would dearly love to outfit the U.S. network with control software and thousands of sensors to report voltage and other information on a second-by-second basis. That way grid operators could, say, automatically lower the cost of transmitting power from a region with excess to capacity to one with a shortage, instead of letting the problem fester until the peaker plants turn on and phone calls and faxes fly. But utilities and regulators alike protect their markets and are unlikely to support any such national electricity authority.