Stand aside, Moore's law. Here comes Chu's law.
Running the Energy Department, a sprawling, $27 billion agency, is no easy job. The nuclear arsenal falls under its oversight. And then there's the nation's migration to a clean-energy economy, which, by the way, Congress might want to chat about soon.
Yet despite these many demands on his time, Energy Secretary Steven Chu is nearing publication on a pet research project that he has led with a small band of physicists. And far from rehashing old work, the study will be an economics paper -- his first in the field -- that could upend modern ideas on how energy efficiency works.
The project, which has run since early 2011, began with refrigerators. For decades, the government has placed minimum energy standards on household appliances like fridges, once a notorious power hog. The expectation has been that, while purchase prices might temporarily bump up, electricity savings would balance that expense down the road.
It seems a reasonable assumption. And it's one that has, until recently, powered most of the Energy Department's models, said Chu, sitting on the couch in his expansive top-floor office, with Smithsonian views, in an interview with Greenwire.
The thing is, historical data don't show it to be true. There is no bump, he said.
"You really can have your cake and eat it, too," Chu said. "You get higher performance. You get lower cost. And you're saving tons of money. And by tons of money, I mean the cost of ownership going down threefold, fourfold. [It's] really dramatic."
The preliminary result, in revision for publication, amounts to a data-driven defense of the government's power to spur "directed evolution," as Chu calls it, in manufacturing. For energy efficiency in appliances, market forces aren't always enough to capture all the creativity of U.S. engineers -- but well-designed standards can, at almost no penalty.
"This is something that actually drives the engineers to make a better product," he said.
DOE has long flagged technological innovation as one of the best ways to mitigate climate change while also benefiting the economy. But while the agency has understood the benefits of innovation in general terms, it has had little grasp on how, or how fast, innovation can advance efficiency, said Robert Van Buskirk, the study's lead author and a physicist and energy analyst at the Lawrence Berkeley National Lab.
This work will advance that understanding, if from an unexpected source.
It's "a group of physicists trying their hand at empirical economic analysis," Van Buskirk said.
The stakes are high. Allowing more aggressive minimum standards in household appliances could make a true dent in electricity consumption; together, they account for about 21 percent of household demand, the largest fraction after heating and cooling, and consume 6 percent of the total energy used by most advanced countries.
Even before Chu's paper is released, DOE has begun to employ the team's work in its cost-benefit estimates for several appliances, including refrigerators. But the team has faced much skepticism in its work, both from inside the government -- it took some time to sell the White House on the idea -- and from industry.
Indeed, the Association of Home Appliance Manufacturers filed a scathing evaluation of the agency's move earlier this year. The department, it argued, "is simply making up numbers," and its forecasting approach is "pseudo-intellectual" and "simply wrong."
"Perhaps in recognition of the underlying lack of theoretical and explanatory foundation for this concept," it adds, "DOE has tried to cover over the situation ... as if more questionable data somehow creates reliable truth when, in reality, the old axiom of garbage in, garbage out still applies."
Despite this vehement response, a number of academics found it a step forward for the agency to include estimates of technological learning in its rules. Previously, it assumed businesses never got better at building identical products. It was a conservative bias and due for revision, said Richard Newell, former head of the Energy Information Administration and director of Duke University's Energy Initiative.
Still, Chu is entering a field full of potential mines with few clear answers, Newell said.
"[It's] one of the hardest areas of science and economics," he said.
For Chu, there are few more reliable stories than the evolution of refrigerators.
He still remembers the old, tiny unit owned by his parents in the 1950s. Back then, a fridge ran some 8 cubic feet; today, they average 22 cubic feet. There was no automatic ice dispenser. And certainly, it was not frost-free. Chu had to heat kettles of water, putting them inside the freezer as he chipped away at built-up ice with a butter knife.
"Now, I wasn't doing this because I was a good child and I wanted to help my parents," Chu said during a lecture last year while first describing his research project. "I did this to keep the ice cream hard. But it was a real pain."
As the decades rolled by, manufacturers added features, like frost-free, which blasts hot air into the freezer to melt forming ice. As might be expected, the energy use of fridges went zooming up, far faster than their expanding waistlines. Their size was limited only by the typical American kitchen. And those got bigger, too.
Then, in the 1970s, under the guidance of Arthur Rosenfeld, the legendary energy efficiency expert who spent most of his career at the Lawrence Berkeley Lab -- which Chu once directed -- California instituted the first conservation standards for refrigerators. It was the start of a campaign that saw the state's energy use decouple from its economic growth, a development now known as the Rosenfeld effect. But with those first standards, industry was quick to say this constituted end times for consumers.
"The manufacturers said, 'You can't do this,'" Chu said during his lecture, which he gave under the auspices of the National Academies. "'The American consumer can't afford it. ... You're forcing people to do something. You're taking away people's choice.'"
Yet despite these requirements, and the national standards that followed, prices kept falling. This continuous drop has many possible causes -- outsourcing to China is certainly in the picture -- but Chu and his colleagues are convinced that their study of historical price data shows that much of this savings is driven by the simple fact that, as manufacturers make more of a product, they get better at doing it.
In economics, this is a concept called learning-by-doing, and it's not new. It was pioneered and applied to plane construction in the 1930s and gained wider circulation during World War II's Liberty ship campaign. It's become a common assumption in manufacturing that as deployment of a widget doubles, costs go down, a result called a "learning curve"; Boston Consulting Group built a whole business on it. Yet, with few exceptions, government regulators have not embraced the tool.
As DOE went to revise its basic fridge standards last year, Chu -- who gets visibly excited by energy efficiency and keeps the Rosenfeld effect close in his mind -- was gobsmacked to see that the agency did not include any sort of technological learning in its models. There was a flat line. No price change. And so he started reading.
"I said, 'This is kind of weird,'" he said. "Because every time there was an estimate of how much that would cost to implement these standards, it seemed to be consistently overestimating. Those were not accurate."
Talking with Van Buskirk and others, he decided to take a new approach. Since fridges were the first appliance regulated in the country, decades of price data exist, along with data for appliances like washing machines and air conditioners. Instead of making simple, conservative assumptions, did this data tell a different story?
Above all, Chu is an experimental physicist. His take on economics mirrored that.
"We decided to take a very empirical physicist approach," Van Buskirk said.
At first, the analysis seemed simple. After a few months, Chu was discussing it publicly.
Using American and European data, two main points popped out: The overall decline in lifetime ownership costs -- over 40 years, appliance energy use dropped more than fourfold -- and the surprising lack of price hikes after appliance rules tightened. But, on further thought, perhaps those phantom bumps weren't so surprising, Chu said. While washing machines aren't as innovative as computers, they're not coal boilers, either.
As always, Chu takes it back to refrigerators. When new standards came into place, manufacturers still faced a competitive market. And they found with better insulation, they could use smaller compressors, and compressors are the most expensive part of a fridge -- at least, after that stainless steel cladding that ties the kitchen together so well.
Another way to look at it: Compare fridges to cellphones, which are limited by their batteries, spurring rapid efficiency gains, since consumers value processing power and mobile talk time. Even today, consumers don't necessarily buy appliances based on electricity use, defying notions of a perfect, rational market, Van Buskirk said.
"The effect of standards on a fridge is like ... battery lifetime on your cellphone," he said.
It seemed so logical that Chu submitted a paper to Science only a few months after their work began. The study had already been sent for review when the team decided to dig deeper into the statistical analysis, giving more rigorous thought to other reasons why the price did not increase while also adding more data and more appliances. After this additional work, the preliminary results seem to hold.
Chu had not previously worked with many people on his team and has proved to be a demanding mentor for Van Buskirk -- whose colorful career has included stops in Harvard, Eritrea's Energy Department and Hong Kong -- and Colleen Kantner, a postdoc at Lawrence Berkeley pulled into the project after only a few days on the job, when a personal email from Chu appeared in her inbox.
Chu never lets his preconceived notions color his evaluations, Van Buskirk said.
"[Chu] asks a scientific question and puts forth the hypothesis," he said. "And initially you say, 'Oh, no, that's wrong.' And then you check it. And you recheck it. And you find out he's right."
Few have engaged in a similar research for appliances, due to a lack of interest or a lack of data. (Typically, learning curves are used for long-term forecasts on solar panels or carbon capture prices, for example.) But one scientist who has delved into the question -- Kornelis Blok, the science director of the Dutch consulting firm Ecofys -- found that the preliminary results from Chu's team matched well with his findings.
"We found learning all the time, within a certain band," he said.
That band of uncertainty fell between price reductions of 10 percent to 30 percent for doubled production, and Blok found it difficult to untangle why it varied. It's a challenge that points toward the fundamental debate about learning curves: Digging into historical data, researchers might identify links between prices and efficiency standards, but finding the definitive cause remains a slog in managing complexity.
"The concern with using [learning curves] for policy is that's purely a correlation," said David Popp, an energy economics researcher at Syracuse University. "It doesn't say anything about what causes costs to fall."
Even deft analysis will leave lingering uncertainty about how long the trend will go.
"Learning often doesn't go on indefinitely," added Ed Rubin, an expert in technological modeling at Carnegie Mellon University. "If you're going out 50 years, is there an empirical basis that costs will come down for whole period? Usually, the answer is no."
The dynamic found by Chu does not have an obvious limiting physical factor like Moore's law, in which, once chip density gets down to a couple of atoms, things get hairy. But his team believes it has made its model predictive. Indeed, the sheer length of some of their identified trends, lasting up to 60 years, gives team members confidence that their predictions will bear out over the next few decades.
There is one obvious limit, though, Van Buskirk added. Say every house has solar shingles and local electricity storage. Once the household's energy use is less than the power hitting the building, everything will be up for grabs.
"The dynamic will completely change," he said.
'But is it perfect?'
It's not easy to change how regulations work, even if you're the Energy secretary.
Within DOE and at the White House's Office of Information and Regulatory Affairs, resistance was steep toward adopting learning curves. How could Chu say the new approach was any good? Well, over the last 30 years, the agency's estimate had been way off, and the new approach would be better, Chu said.
"They said, 'Yes, but is it perfect?'" Chu said, eliding what was likely a testy discussion. "We said, 'Well, it doesn't have to be perfect.' And convinced them of that."
In the end, the agency adopted the approach in price forecasts for a host of standards: for fridges and freezers, of course, but also dryers, microwaves, window and central air conditioners, and furnaces. It is proceeding gingerly -- few bureaucracies are immediately responsive to new economic evidence -- but it is heading in a set direction.
"I was happy to see they're doing this," Carnegie Mellon's Rubin said. "If it's used properly, it'll give much more realistic answers to these kinds of questions."
It's debate that's only likely to spread, added Blok, the Dutch researcher. Right now, Europe is asking whether its base standards are ambitious enough. For example, recently it took about two years to put new minimum standards -- "C labels," in the E.U. system -- in place for televisions. When it came into effect, no C-level TVs were left.
"The standard was already way too conservative," Blok said.
It is an open question whether DOE would have moved toward learning curves had Chu not been the secretary. But certainly his presence, and his personal attention, has sped the effort along. It's a dynamic that the career employees have learned to, if not appreciate, at least accept, Chu said.
"They've gotten used to it," he said. "I've been here nearly three-and-a-half years. Because it's not only in standards. It's in every nook and cranny. They've gotten used to it."
Meanwhile, Chu has seen his ever-roving eye move to some newer features of the home, like the small black-box transformers that dangle at the end of every laptop cable, churning alternating into direct current. Manufacturers could make the boxes 55 percent or 95 percent efficient, with the difference costing no more than a dollar or two.
"The price will never motivate a buyer, because you're looking at what kind of computer you want," Chu said. "You're not looking at how efficient is the charger. ... [And] left to their own devices, guess what the manufacturers chose?"
Another black box gets Chu especially exercised: the always-on video recorders and tuners proliferating in living rooms. With some of these boxes, hitting the off switch only turns off the power light -- everything else keeps whirring. Some don't even spin down their hard drive, he said with the parsimonious incredulity of someone who has stuffed each of his homes full of high-performing insulation.
And now, his interview time expiring, Chu delivered the coup de grace on his indictment:
"Two of those set-top boxes," he said, "now consume more energy than your refrigerator."