"It shouldn't be surprising that we're using a fair amount of electricity to keep the Internet economy hot," says Mills. "This virtuous economic circle is very powerful, and everyone wants to act like it's free, like there's no energy cost to it, like living without food." Mills and Huber want to highlight the dependence of so-called new-economy companies, and all the wealth they inspire, on the oldest of the old economy -- power companies. "The digital economy," they wrote in the Wall Street Journal, "which most everyone loves, is completely dependent on the big central power plant, which most everyone hates."

So while Silicon Valley lobbyists busied themselves arguing for more H-1B visas, did they forget to worry about what keeps the machines running for all those workers to type away on? "They got very focused on bringing in enough labor and talent, and they forgot about power," says Mills. "Twenty years ago, it was inconvenient to have a brownout. It's a killer in the silicon economy. It's blue-screen death."

That specter of the blue screen of death has been more than enough to encourage Silicon Valley CEOs such as Intel's Craig Barrett to start throwing around their political weight to advocate for the construction of new power plants -- and fast. And who can blame them when they've been reduced to shutting off lights and bracing for rolling blackouts?

But increasing demand for electricity is only one factor in California's current energy woes -- and the overall increase has not been especially dramatic for a time of economic boom. According to the California Energy Commission, electricity demand in California grew at a rate of about 2 percent a year in the late '90s -- the height of the Net boom. It's not a staggering percentage. In the go-go late '80s, between 1987 and 1990, U.S. electricity consumption grew 3.3 percent a year, according to the Electric Power Annual, a publication of the Energy Information Administration.

California's power problems are rooted in a mix of factors, most importantly a flawed deregulation scheme that gave little incentive for power generators to build more capacity in the short term to meet demand. As part of the 1998 deregulation, utility companies sold off much of their power-generating capacity to a handful of wholesalers. But once they enjoyed the market power to raise prices, wholesalers haven't been in a rush to raise capacity. They've taken a wait-and-see attitude about building more power plants -- a position that's somewhat understandable in a state like California, where stringent environmental regulations mean the construction of a new plant can take an average of seven years, compared with three years in a more permissive state like Texas. And with utility companies unable to pass on wholesalers' rapidly rising prices to customers without government permission, their road to bankruptcy has become painfully obvious.

"The whole California electricity crisis is not a crisis related to rapid increase in demand. Demand growth has not been out of the ordinary," says Jonathan Koomey, a scientist and group leader of the End-Use Forecasting Group at Lawrence Berkeley National Laboratories. Although no one could have predicted how quickly California's recession would turn from bust to boom, moving from a power surplus to increased demand, a 2 percent annual increase is hardly a never-before-seen consumption surge.

So what's the explanation for Mills and Huber's 8 percent figure? "If the claims that they're making are true, you'd expect to see vast increases in electricity demand and you are not," says Koomey. Scientists at Lawrence Berkeley have refuted Mills and Huber's assertions point by point, based on their own research. They put the figure for all office, telecommunications and networking equipment at 3 percent of the total electricity used in the United States.

It comes down to a war of watts. For example, Mills and Huber argue that after factoring in all the networking and telecommunications equipment required on the back end, like routers and servers, a PC and its peripherals connected to the Net use 1,000 watts of power, which is as much the electricity used by 10 100-watt light bulbs. But Koomey and his group think that figure is wildly exaggerated. Koomey says that a PC consumes only 50 to 200 watts and that factoring in the back-end equipment adds only about 15 watts to the PC's electricity consumption.

Mills and Huber also assert that a Palm Pilot that is plugged into the Net consumes as much energy as a refrigerator -- a nice sound bite that has been widely quoted. But is it true? Koomey says he sent an e-mail to Mills requesting documentation: "I am trying to reproduce your estimate about the electricity use of a Palm Pilot equaling that of a refrigerator, because of the network electricity use. Is there a place where this calculation is documented?" Koomey says he repeated this collegial request eight times over two months, but got no response.

In a rebuttal to Mills' testimony ("Kyoto and the Internet: The Energy Implications of the Digital Economy") to the House Subcommittee on National Economic Growth, Natural Resources, and Regulatory Affairs, Koomey wrote, "In the past year and a half, I have been witness to an extraordinary event: An analysis based on demonstrably incorrect data and flawed logic has achieved the status of conventional wisdom."

Mills defends his 8 percent figure by saying that it's only an estimate: "The total may be somewhere between 5 percent and 6 percent. Our number is an estimate. No one knows for sure. We know it's not zero. That's why the whole debate is sort of silly in a way, if you're focusing on a few percentage points."

But Koomey contends that a difference of a few percentage points does matter. "This whole set of bogus numbers has gotten so much play because it is convenient for certain interests to make it sound like growth has gone up at a ferocious pace," he says. "The people who want to build power plants would like everyone to think that there is huge demand growth so we can reduce environmental regulations. I think that we have to build power plants, but I also think that people are trying to create a sense of urgency to make it easier for them to argue for their interests."

The Free Software Project
Read Andrew Leonard's book-in-progress on Linux and open source -- and post your comments.