On Friday, Matt Yglesias made the point that only socialist state control seems capable of creating a robust nuclear power industry. After all, the only countries building nuke plants these days are the ones where governments are making the decisions. David Frum replied with a series of wildly overbroad assertions ranging from false to highly misleading, with no evidence or links to support them. (Nuclear power has an impressive effect on conservative error-to-word ratios.) Matt replied in turn, and in doing so echoed a familiar misunderstanding:

That said, obviously you need a certain amount electricity that can be relied upon irrespective of how windy it is or whether the sun is shining. So I’d happily see the nuclear share of the pie grow at the expense of coal and oil as the provider of that baseload electricity.

This notion has really grabbed the public imagination. It’s become conventional wisdom that the grid can only incorporate a limited amount of renewable energy; ergo, we need coal and nuclear power plants for “baseload” electricity. Clean energy skeptics wave the word “baseload” around like a talisman.

There’s far less to the claim than meets the eye, though. As Amory Lovins points out, it’s a category error: baseload is a characteristic of aggregated demand, not of any particular kind of supply. He distills the counter-argument:

Baseload: The electricity system doesn’t rely on any plant’s ability to run continuously; rather, all plants together supply the grid, and the grid serves all loads. That’s necessary because no kind of power plant can run all the time, as Stewart says they must do to meet steady loads. I repeat: there is not and has never been a need for any particular plant or kind of plant to run all the time, and none can. All power plants fail, varying only in their failures’ size, duration, frequency, predictability, and cause. Solar cells’ and windpower’s variation with night and weather is no different from the intermittence of coal and nuclear plants, except that it affects less capacity at once, more briefly, far more predictably, and is no harder and probably easier and cheaper to manage. In short, the ability to serve steady loads is a statistical attribute of all plants on the grid, not an operational requirement for one plant. Variability (predictable failure) and intermittence (unpredictable failure) must be managed by diversifying type and location, forecasting, and integrating with other resources. Utilities do this every day, balancing diverse resources to meet fluctuating demand and offset outages. Even with a largely (or probably a wholly) renewable grid, this is not a significant problem or cost, either in theory or in practice—as illustrated by areas that are already 30-40% wind-powered.

Right now our power system might be characterized as Security Through Oversupply. We’ve built enough power plants to create the maximum level of power we might ever need at a given point in time; but since “peak load” times are relatively brief, most of the time dozens and dozens of large power plants are cycled down, sitting idle. As population and per-capita power use rise, the size of peak load is rising as well. The STO response is to build more plants.

The alternative will be Resilience Through Diversity: just-in-time, just-enough power from multiple, redundant, diverse sources spread over large geographical areas, managed by a reliable, intelligent power grid incorporating distributed storage. Peak load will be shaved by load spreading and efficiency; failures will be localized and self-healing rather than cascading and catastrophic; intelligence will replace brute power.

Utilities face, imminently, some very large investment decisions. Should they invest in nuclear and “clean coal” power because they will “have to” have some baseload power on the grid in 10-15 years when the plants are completed? No. For the next decade it will be a huge challenge just to get to the level of renewables integrated in Spanish and Italian grids today (30-40 percent). In the ensuing time, an enormous amount of money and engineering will go into grid resilience and intelligence. It is far too early to predict what level of renewables will be “impossible,” but whatever that level turns out to be, it is certainly far distant.

This is the green pitch to utilities: Rather than spending the next decade or two building nuke and CCS plants, with all the attendant management hassles, public opposition, lawsuits, and cost overruns, why not spend it reducing demand, creating a more resilient grid, and diversifying the generation portfolio? The former is just a more expensive version of what exists now. The latter is a revolution, a platform for innovation that will make the internet look like, um, the electricity industry.

A pitch isn’t enough, though. For a fusty industry like utilities, revolution is to be resisted, not celebrated. The key is not just asking utilities to use full cost accounting, but to start building such accounting into markets via regulation, legislation, and large-scale investment. Once the financial and legal incentives are correctly aligned, even utilities—slow and regulator-dependent as they are—will respond. Until then, until they really start trying, we shouldn’t trust them about what parts of the old system are “necessary” in the new.

(For a longer and more detailed response to the “baseload” shibboleth, see Lovins’ “Four Nuclear Myths” [PDF].)