The Fundamentals of Turbulence
The physics, engineering, and economics of our current technologies contribute.
Energy Markets Are Like a Broken Thermostat
Most reading this are blessed with a functional HVAC system that maintains a set temperature. The system holds the temperature by measuring the process variable (indoor temperature) and comparing it to the setpoint (desired temperature). If the process variable strays, the system turns on within seconds and runs as long as is required to return the indoor temperature to the desired set point.
Now imagine your thermostat can only read the temperature from four hours ago, and the HVAC unit can only run in 12-hour blocks. Instead of holding the setpoint, the indoor temperature will constantly overshoot and undershoot the set point, like a roller coaster. 
Energy technologies work this way. Projects can take years to develop, demand is driven by impossible to forecast factors like weather and economic growth, and capital cost dominates total cost, meaning projects operate continuously once online. Instead of our ideal thermostat, we have the broken one. Fossil fuels require constant investment to hold production steady. When the thermostat turns the HVAC off, its ability to run during the next cycle degrades.
If new technologies shorten the market response time, markets are more stable. Shale gas drilling is much faster to market than alternatives like offshore drilling.
I probably don't need to tell you that shale gas took over providing the marginal cubic foot of gas from offshore rigs around 2010.
Markets that help with planning are also important. United States natural gas futures are thickly traded, with both producers and buyers regularly participating. Decisions are made based on the EIA weekly storage report. If storage is too low for the season, prices go up. If storage is too high, prices go down.
Price volatility is inherent to energy markets as long as supply and demand are inelastic in the short term. "High prices are the cure for high prices, and low prices are the cure for low prices."
Energy Productivity Deteriorates
Productivity can deteriorate because energy projects are often construction projects. Knowledge required for high productivity is in human minds, and humans come and go.
All drilling engineers dread picking up a stacked rig (one that has not been actively drilling). New workers show up and don't know how to put it together. Managers or mechanics from other rigs robbed parts. And the workers have no idea how to work together. It is like the first day of fall practice. The first few wells drilled might incur a 30%-50% cost premium suffering through the shake-out period. After continuously running, productivity increases through a myriad of improvements. If the rig gets laid down, the workers get laid off, and the learning disappears into the wind. Sustained success only comes from technologies like better drill bits that go faster no matter who is on the brake handle.
Building nuclear, coal, or hydroelectric power plants suffer the same problems. Want to build a giant dam cheaply? Henry Kaiser has been dead for some time. The cost of building coal power plants in the United States skyrocketed after construction slowed and complexity from environmental controls increased. In the 1960s and 1970s, a core construction industry and the Atomic Energy Commission had a building machine cranking out nuclear plants in short periods at low costs. Once construction paused and regulations increased construction complexity, prices skyrocketed. A 40-year-old tradesman or middle manager in 1970 is over 90 today. High productivity light water reactor construction is only coming back the hard way.
Source: Lovering, Yip, Nordhaus
The increase in cost variance is striking. Some teams handled the extra complexity better than others. Now lets look at historical coal power plant capital costs:
Source: McNerney, Farmer, and Trancik
Changes in design or slowdowns can reset cost curves.
The alternative is continuously building cookie-cutter designs. Construction crews traveling around the country building Starbucks stores are more productive than workers building a custom house. There is at least one country where nuclear construction prices have not skyrocketed - South Korea. They have continuously been building plants for decades. But when they tried to construct plants in the U.A.E., even they struggled.
An efficient construction-related energy industry can lose its ability through mandated design changes or pauses in construction work. Luckily not all energy productivity is tied to construction-like effects.
Manufactured Technologies Embed Learning
Chad Syverson is an economist at the University of Chicago studying industrial organization. In an interview he tells us why knowledge in factories is different:
Syverson: Regarding the data, as a car is being made, there are things constantly being recorded in the factory's information system, either in an automated fashion or by workers manually inputting information. So Steve Levitt, John List, and I were able to see every step of the way whether the step went right or wrong. And then we looked at subsequent defect rates for every car that was made – about 190,000 over the course of a year.
Most of the empirical learning-by-doing literature has looked at unit costs, such as how many worker hours it took to make a unit, and then examined that over time and traced out the learning curve that way — how fast people adapted, for instance. Our more detailed data let us learn something about where the knowledge resided inside the organization and how it moved around.
There are a few facts that are important to understanding that in this setting. One is that a lot of learning happened early, as is pretty common. So, for example, defect rates fell 70 percent in the first two months of production. Now, as it happens, the factory only ran on one shift for the first two months of data we observed, and then starting in the eighth week, the second shift started. The second shift's training was to watch the first shift for one week. That was it. They weren't on the line itself. Once the second shift comes online, they are right at this new, lower defect level that the first shift achieved. So you immediately know that it's not just being on the line for a while that leads to improvements.
Two, there is a high correlation between defect rates for a particular operation across shifts. Operations don't go wrong with equal frequency. There is a right tail of processes that go wrong a lot of the time, and then there's a left tail where things never go wrong. That's true across shifts. So if some operation is problematic on the first shift, it's problematic on the second shift, even though the workers are different.
Three, we were able to see absenteeism every day at the factory and in which part of the production process the absent workers were placed. There is a positive relationship between absenteeism rates and defect rates along a set of operations on the line, but it's very weak.
So those three things suggest it's not the workers who are carrying the knowledge, which, again, is substantial. Defect rates over the course of the year came down 90 percent total.
What happened is the factory had a set of practices to take knowledge from the workers and as quickly as possible put it into the capital of the factory — either the physical capital, such as changing a faulty part on the line, or the organizational capital, such as workers conveying information to each other.
Factories physically embed the technology in the factory instead of in workers' heads. In our drilling rig example, the penalty suffered would be higher, except the rig itself acts partially like a factory with embedded knowledge. Technologies with more construction, like thermal power plants, fare worse.
Factory technologies also handle complexity and change better. Governments have imposed numerous safety and environmental standards on cars, yet productivity and costs have continued to improve.
Factory products often require few skills for installation or use. We want our energy technology to have knowledge embedded in factories instead of humans and be easy to use. Solar panels, batteries, and nuclear microreactors are underrated technologies for this reason.
There is a vast opportunity for simplification in small-scale generation hookups, whether solar or another source. In ten years, it should be normal to go to Home Depot, buy $1500 worth of solar panels and batteries, install them as a Saturday project, and be off-grid capable by the end of the day. An older house might require an hour from an electrician to make the final hookup. The same simplicity will be available for utility-scale installations of any manufactured energy technology.
Preserving the Tradition
The knowledge stuck in the head of petroleum engineers and specialized service company employees is even more construction-oriented than the knowledge of running a drilling rig. But the increment of drilling an onshore well is small, and even in the worst slumps, drilling must continue given the natural decline rate of wells. Continuous drilling preserves knowledge through the hard times. Once demand comes roaring back, replicating the knowledge is expensive, but the dominance of fossil fuels means the price adjusts to accommodate training.
Mega-scale technologies like nuclear plants completely stop new construction in bad times. Knowledge goes extinct. The power plant construction sector is notoriously cyclical. Mega facilities like coal plants suffer as well. Smaller, factory-built technologies like gas turbines have prevailed as America's electricity demand flatlined in the 2000s.
Shale wells, gas turbines, solar panels, wind turbines, and nuclear micro-reactors have advantages in sustaining their craft through pricing cycles because of their smaller size.
The End of Energy Turbulence?
Manufactured energy technologies with rapid install times still won't end energy turbulence, though they can reduce it. Each technology sees a period of rapid expansion until it saturates the market. Once growth normalizes, factories close, and the supply chain becomes more fragile. It becomes highly optimized around factors like cost. The US natural gas industry has recently started this transition. If economic growth or other factors change the demand for additional energy capacity, the response will be delayed, slowing economic growth. The faster a technology's production capacity ramps and the smaller the increment it delivers, the less disruption there will be.
Technologies like wind, solar, nuclear, and geothermal have the advantage of low decay rates. If a shock, like a pandemic, lowers energy demand, their installed production potential is still available as demand rebounds, unlike fossil fuels. The difficulty of satisfying growing demand causes remaining turbulence. Imagine if we "Made America Great Again" and "Built Back Better" by going exclusively with traditional nuclear power plants that take ten years to build.
Prospective Factory Builder: "Hi, I'd like to see about electricity service to open a new factory."
Utility Representative: "I'll put you down on the waiting list! It will be about ten years until the construction of our new plant is complete. We could not foresee this latest burst of economic growth driving demand on our electrical system."
And if it is a state like Alabama or Arizona:
Utility Representative: "I'd also like to remind you that it is breaking the law in our service territory to consume electricity without a grid connection. Don't let one of those distributed electricity installers trick you into signing a contract. There is a reason why they hand out branded soap-on-a-rope as a promotional item. Have a nice day!"
Technology Drives Politics (Usually)
Does politics choose energy technology, or does politics react to energy technology?
Technologies Emerge on Own Merits
Nearly any potential energy technology receives government funding and research at some point. It looks like we choose technologies. In reality, the technologies that make more progress rise to the top. Despite government and societal pushes for ethanol or hydrogen-powered cars, battery-electric cars are the fast-growing technology. Was there another technology available to China and India besides low-tech, dirty coal power plants? Gas, renewables, nuclear, or even cleaner supercritical coal plants were out of reach during the early parts of the booms.
Politics reacts on the margin to what technologies emerge. If I ran on a platform of returning to only burning wood for fuel, I'd lose spectacularly. Politicians make mistakes on the margins trying to promote one technology over another, like shutting down nuclear plants early. Few try policies that would cut our total energy supply by 90%.
Let's say we did bring back a program of constructing gigawatt-scale nuclear plants. We'd have to pay a lot of tuition. At some point, we'd saturate the market. Once we stopped building the plants, our knowledge would rapidly decay. If any semblance of competition exists, more flexible technologies will fill the gap readily instead of restarting giant nuclear facility construction.
Politicians play all sorts of games around the natural rhythms of energy prices, but they don't control physics and engineering.
Politics Mostly Halts
Politics can kill existing or emerging technologies as they grow. Nuclear was an unstoppable freight train until it wasn't. We didn't dam up Yellowstone Falls, after all. The larger and more centralized the technology, the more vulnerable it is.
Political Equilibrium on a Centralized Grid
Electric grid capacity is a resource that has to be rationed, just like any other. Given the government's involvement, price is not the only rationing mechanism used. Few customers pay variable prices. Moving transportation, heating, and industry onto the grid only raises the stakes. Bad decisions driven by politics and attempts to shield voters from volatility in electricity prices will harm economic growth and likely discriminate against non-politically connected electricity customers.
Demand-side reforms on a unified grid are risky for politicians. Because electricity prices can go parabolic during critical events, customers could get huge bills. Politics demands the grid capacity to handle the worst-case scenarios for users who suffer from power outages. Rationality can prevail when users make capacity decisions in smaller increments like a neighborhood-size microgrid or a household. How much you would spend to go from 8755 hours to 8760 hours of power availability a year becomes a personal question rather than a societal one.
The grid will continue to play a major role in electricity distribution for decades. That doesn't mean we should reinforce its centrality.
We Can't Return to the Past
Dead people can't build power plants.
The Renaissance Doesn't Look like The Past
What are the market and our innovative energy startups trying to give us?
Smaller oil and gas producers gave us shale gas instead of offshore mega projects
Companies like Mitchell Energy, Range Resources, Petrohawk, and Enron Oil & Gas brought the new technology to the forefront instead of Exxon.
Tesla, Sunrun, and others sell small scale solar and modular battery installations
Can't build new transmission lines for reasons x,y, or z? No Problem.
Any Size Utility-Scale Solar
Solar and batteries are modular, and the sun shines in most places. 10 MW solar plus storage plants are not dramatically more expensive than 500 MW ones. Higher DC to AC ratios, storage, and modularity allow solar plants to utilize the available grid connections. I detailed these strategies in a solar post.
Nuclear startups like Oklo and Radiant Nuclear are attempting to build 1-2 megawatt reactors.
These startups use fuels that reduce meltdown risk. Both companies are initially targeting the off-grid market where power prices are sometimes 10x that of wholesale markets. Building and operating non-grid plants allows faster time to market by skipping licensing steps with the Nuclear Regulatory Commission (NRC).
Oklo is the furthest along of any nuclear startup, starting the combined operating licensing process in 2020. They expect the construction of their first reactor to begin around 2023 and take a year. They already have customers like cryptocurrency miners lined up.
Radiant's design is more technically ambitious, and they hope to have a reactor running by 2025, meaning their licensing must start in 2022. Creating a reactor that can fit inside a SpaceX Starship and power a Mars colony drives their motivation.
It is hard to emphasize how much Oklo is breaking the NRC into licensing a technology that isn't a light water reactor (LWR). Their public documents show a constant back and forth of legalese arguing over what regulations should or shouldn't apply to a non-LW reactor 1/1000 the size of a typical LWR. It is possible follow-on designs could go through faster if Oklo successfully blazes a path.
Highly distributed technologies generally have higher generating costs at the plant site, but they avoid the cost of transmission and maintaining the grid, making them competitive at higher prices.
There are potential technologies that would have less impact on energy market turbulence.
Offshore wind needs scale. Someday we might see 50 MW turbines (Onshore has crept up to 3 MW turbines). Low cost and high capacity factors could drive adoption, but long construction and permitting lead times will not make our thermostat work better.
For geothermal to be competitive, it will likely be in increments of 25-50 megawatts. It requires both drilling and power plant construction. Over time the power plants could become more modular, and drilling technology could improve to reduce the importance of worker knowledge. It is possible to develop short cycle times and more flexible increments.
Not So Small Modular Nuclear Reactors
A 70-megawatt reactor might be small compared to a one-gigawatt monstrosity. It is still relatively large. NuScale started in 2012 and hopes to be selling power onto the grid by 2030. Its timelines are similar to traditional nuclear plants. Hopefully, the cycle time decreases once the technology is commercialized.
Other "Small Modular Reactor" designers have even larger plants. These companies' regulatory process is different than Oklo's is. NuScale had to go through an extra design step since it is selling the plant to utility customers. Other regulations like NEPA reviews await.
The Future We Want
Consensus in large groups is more challenging with 21st-century communications technology. Does it make sense to lionize 5-10 year lead time centralized plants on a quasi-governmental electricity grid? If the market is trying to deliver more flexible technologies, we should clear the regulatory path and see what happens.
In a previous post, I covered reforms to accommodate faster renewables growth.
For new nuclear technologies, one answer might be to create a different regulatory body than the NRC. The government designed the NRC to put existing gigawatt-scale plants in a straight jacket to prevent meltdowns. If new generators use dramatically safer fuel and technologies, they need a different regulatory paradigm. Startups led by crack engineers can build a nuclear microgenerator factory faster than traditional contractors can build a copycat LWR - if we let them. The NRC, "Big Nuclear" lobbies, and the fear of nuclear make this challenging.
The more distributed, responsive, and incremental new capacity is, the more stable energy prices will become. Fewer politicians will care about energy. The market is sending us these signals for a reason.
 If you are curious about the detailed math behind this, google "Engineering Controls Theory."