This book is a set of sober, science-based calculations and assessments. It’s technical but not inaccessible. The author is disappointed with current energy debates, proposed policies, and news headlines, because they aren’t based on sound reasoning. He wrote this book to fill in the science and inform the conversation.
I’ve summarized my takeaways from the book in this post, but I also recommend that you check out the book. At 163 pages, it’s no dense tome. The chapters are laid out by topic, and the pages are peppered with clear and compelling graphs. If you’re only curious about one of the topics, by all means, read only that chapter.
My takeaways, below, are a study guide you can refer to when you’re reading an overly optimistic book on energy or climate change. Before you get too excited, flip to the appropriate chapter and let Smil take some of the air out of an overinflated idea.
I’ve also listed in each section a few opportunities and challenges which would make the technology more viable. This is a flip of Smil’s cynicism: he outlines the technical problems clearly. It’s up to us to solve the challenge.
Chapter 1: Electric Cars
Reality: Electric cars are only a niche piece of the market, and will be long into the next decades.
Myth: All cars will be electric cars in the near future (next 10-ish years).
Smil’s main points:
Slow adoption rate. Hybrid cars have taken more than 10 years to claim less than 3% of the market. Why would we think all-electric cars, which require much more infrastructure investment, would adopt to 100% in that same amount of time? Likely technology adoption rates in most researched, published scenarios put the likely share of pure-electric cars at no more than 25% by 2050. (p25)
We don’t produce enough energy to charge 100% of cars, and can’t scale up to it quickly. Assuming that the overall demand of a midsized electric car is around 6MW/year, if all American cars suddenly became electric, we would immediately need new power generation equal to 25% of all of the energy used in the United States in 2008. (p26)
We can’t produce that much more energy soon. It took 15 years (1993–2008) to spin up that quantity of power the last time we did. (p26)
Battery performance is sub-par and degrades quickly. Lithium ion batteries lose power even when idle and their performance degrades over time and with temperature. Tesla engineers expect the car battery pack to degrade by as much as 30% in 5 years (p29)
Opportunities and challenges:
In large cities (where electric cars for commute make the most sense), 30–60% of cars are parked curbside. Since most electric car scenarios envision overnight charging in garages, how would these curb-parked cars be charged? (p25)
If all new cars were electric, 98% of cars would still be burning fossil fuels. How do we get old cars off of the roads more quickly?
Electric cars pull a lot of load; how do we manage the charging of each electric car so that we don’t create a new energy peak?
Gas-fueled internal combustion engines might be a more efficient way to get energy into a car for some time to come. American energy is largely oil-produced; the tradeoff of miles per gallon of gas directly in the car vs. through an electricity generation process and transport through the grid is comparable– about 38mpg for an average electric car in 2008 (p27). With this in mind, one challenge is to improve the efficiency of the gas-powered combustion engine. Such projects are underway, e.g. DiesOtto by Mercedes-Benz. (p28)
What’s a better energy storage solution than Lithium-ion batteries?
Reduce the power consumption of an electric car, or increase the ramp-up speed of bringing additional (sustainable) power online.
What are the main causes behind slow hybrid adoption– do they apply to electrics, and are these things we can change?
Chapter 2: Nuclear Electricity
Reality: Nuclear electricity is one of few ways to bring on large amounts of power per plant on relatively short timescales, but it’s not going to be cheap, fast, or ethically straightforward.
Myth: Nuclear energy will singlehandedly solve the world’s energy needs by providing huge quantities of clean energy at minimal cost in the near future.
Smil’s main points:
Building a nuclear energy plant is very, very slow and expensive. Between 1972 and 1992, the cost of building a new 1 GW nuclear power plant in the United States increased more than 10x. This was due mostly to increased safety regulations. The plants are now much less likely to become meltdown sites, but the adoption rate is very slow. (p36)
Nuclear fission is not in our near future. Smil particularly addresses the hope of liquid metal fast breeder nuclear reactors. The subject has been funded and researched since the 1940’s with no promising commercial outcomes yet. (p38–39)
Opportunities and challenges:
A big reason why new nuclear plants have been expensive is that the laws changed while construction was already underway. Another (relatedly) is that they don’t follow a standard design. There may be an opportunity to greatly reduce construction cost and time by designing and implementing a standardized nuclear plant now that regulations are more settled.
Devise a good/safe/reliable method for storing a small volume of highly radioactive waste to be sequestered for thousands of years. No country has one yet. (p43)
“Nuclear generation is the only low-carbon-footprint option that is readily available on a gigawatt-level scale. That is why nuclear power should be part of any serious attempt to reduce the rate of global warming; at the same time, it would be naive to think that it could be (as some suggest) the single most effective component of this challenge during the next ten to thirty years. The best hope is for it to offer a modest contribution. “ (p43) Assuming nuclear power should be a part of a sustainable solution– especially as a high-reliability, high-power complement to wind/solar/etc.’s fluctuating lower-wattage contribution, what further construction is necessary? (p40)
If you’re firmly anti-nuclear, figure out how to redistribute the research funds. Nuclear research received 96% of all funds appropriated by the US Congress for energy-related R&D between 1947 and 1998, a total of $145b in 1998 dollars (p43)
One challenge is to create accurate public perception surrounding the social and planetary costs of nuclear energy. Waste disposal and other issues are not all worked out, ethically – but where do these impacts stand in relation to coal and oil in terms of human and environmental degradation per watt?
Chapter 3: Soft Energy
Reality: “Soft energy” is the theoretical matching of natural renewable energy flows to local power consumption. Small and local sound appealing, but are not inherently better. While there may be a place for local energy generation, a full solution likely includes many energy sources, depending on their individual economics.
Myth: The idea posits that by decentralizing power production and localizing it by community, we can eliminate inefficiencies such as infrastructure investments, transmission line power losses, and power company office workers. The result is cheaper, renewable, locally conscious power for all.
Smil’s main points:
Soft energy assumes an imminent world shift to renewable energy. This would be nice, but it’s debunked in chapter 8 (p47).
Soft energy is only a small portion of current energy usage, nowhere near the touted adoption rate. No country, as of 2000, uses local power as a major (even non-negligible) energy source. The author of the theory proposed in 1976 that soft energy would account for 33% of United States energy by 2000. Instead, it was less than 0.5% (p47).
Forcing transition to local biogas generation failed in Maoist China. As part of the Great Leap Forward, communities were mandated to produce biogas as fuel by use of a digester that would use waste products, plants, human sewage, etc. as raw inputs. Typical output was not enough fuel to cook rice three times a day (p50).
Big projects leverage economies of scale. Producing power in large plants reduces the cost of construction, transmission, infrastructure, and all the other rolled-in costs of soft energy — to the extent that it can be cheaper per watt than small, local installations.
Opportunities and challenges:
The biogas generators in China were not maximally efficient. Proper biogas generation requires completely anaerobic digestion, precise input mixing, and temperatures above 20C. Is there an opportunity to create a more self-managing digester? What other local power solutions become plausible if made usable by non-experts?
Small and local power generation might still be a worthwhile component of an full energy solution– how can it be approached economically and with appropriate expectations? In what situations is it more effective or efficient than traditional power plant scenarios?
Chapter 4: Peak Oil
Reality: The world has a lot of untapped oil; we’re not about to run out, and resource harvesting rates are asymptotic, not bell-curved. (Not Smil’s point, but we actually have the opposite problem: 2 degrees celsius is the international standard of “let’s not warm the earth any more than that”, but oil company reserves show 5x more than the allotted CO2 we can afford to emit as part of their currently valued net worth– see the Do the Math campaign from Naomi Klein and Bill McKibben).
Myth: Based on the incorrect belief that resource extraction curves fit a bell curve, “peak oil’ is the idea that we will run out of oil to extract and then our industry will drop precipitously, returning us to a hunter-gatherer lifestyle within a few thousand years (p62–64)
Smil’s main points:
The model is incorrect. The best fit for a resource extraction model is logarithmic, so while we will hit an asymptote on oil (and thus be unable to keep up with increasing demand), we won’t suddenly be out of energy. (p66)
We have a lot of oil already recoverable. The United States Geographical Survey sets a 95% probability there are 400b barrels of oil that can be extracted from currently known fields. At a current global rate of ~100m barrels consumed per day, this would last us 4000 years (p68).
There’s probably a lot more oil we haven’t found yet. There are enormous major sedimentary fields both associated with existing land and deep underwater that most likely have a lot of oil. They have not been truly tapped until they have the same density of drilling as Texas. (p68)
Opportunities and challenges:
We aren’t going to run into a physical limit on oil in the near future, so the challenge is on human restraint. How do we make it plausible to give up our existing energy infrastructure when not forced to?
How do we align exponential growth in energy usage and logarithmic growth in oil harvest?
Chapter 5: Carbon Sequestration
Reality: Carbon sequestration is not something we can reliably accomplish in an energy-efficient manner with clear and permanent results. Many otherwise valuable carbon sequestration opportunities are decreasingly powerful due to the effects of global warming.
Myth: We can keep emitting as much carbon as we want, because we can just sequester it back out of the atmosphere.
Smil’s main points:
Global warming reduces the likelihood that we can count on forests and trees as permanent carbon sinks. Sequestration of carbon in forests fluctuates to the extent that some years forests can produce more carbon than they sequester (p80). In the near future, Tropical forests’ carbon impacts will change in the near future mostly due to deforestation, but many other forests will be limited by water and soil nutrient availability, especially a lack of nitrogen. This is one of the effects of global warming. We will also have more carbon-releasing wildfires across these forests due to longer droughts from global warming. (p82)
Sequestering the amount of carbon we emit in trees would require truly enormous new forests. Planting mixed forests sequesters carbon at the rate over which the trees mature — so in 10–80 years after planting (depending on the type of tree), the tree must continue to live to hold carbon, but it does not offset new emissions (p82). Offsetting just 10% of 2005 carbon would require a planting as big as the combined forests of North America and Russia, or a ~15% increase in tropical forests (p82).
There is opportunity to sequester carbon in soil, but global warming makes this type of sequestration uncertain in the long term. Soil stores about 4 times the carbon that is stored in land plants (p83). Tropospheric ozone levels are increasing and can reduce plant productivity thus slowing soil sequestration. Uncertainty caused by global warming means we can’t know whether soil will net store carbon from plants or net emit it from decomposition (p83)
Biochar could improve carbon sequestration in soil, but there are logistical and environmental challenges. Soil with biochar stores 2.5x carbon as soil of the same type without it (p83). However, there is currently no supply chain set up to source waste biomass (p84).
Biochar can provide only a small piece of the solution; 900 million tons of straw (the total amount produced by affluent counties) turned into biochar (ignoring the logistical and application challenges) would sequester only 2.5% of the CO2 emitted globally in 2005 (p84).
Pumping CO2 into basalt might have a small effect, but it’s unproven. The idea is to trap CO2 in basalt layers beneath the Indian Ocean and/or the Juan de Fuca tectonic plate. This method, if functional, could only trap 4% of American CO2 emissions — so, this would be less effective than raising car emissions standards (p87)
We don’t have an infrastructure to capture, move, and sequester carbon. Most sequestration solutions depend on pipelines of CO2 and other infrastructure that we don’t have and which will take time to build. (p87)
Sucking carbon out of the atmosphere is highly experimental. One specific project posits the deployment of artificial trees that circulate a carbon-sucking liquid (likely an aqueous solution of calcium hydroxide). Each of these trees could theoretically suck up to 90k tons per year, so it would take only 160k of these to remove half of global carbon emissions from 2005 — assuming the trees have access to lots of wind (for high throughput) or high elevation (where carbon concentrations are higher). Circulation of the fluid and extraction of the carbon from it may be energy-intensive. Then, the problem of sequestering the carbon is yet unsolved (p89).
Capturing carbon at its source is a good idea, but companies are not incentivized to capture and sequester their emissions. “CCS” (carbon capture and sequestration) involves sorting CO2 out of exhaust at its source, transporting the CO2 (typically, in compressed form through pipelines) and injecting it into underground structures (p89).
Carbon sequestration through direct intervention by humans has unknown long-term effects. We don’t know what effect there is in injecting CO2 into underground structures. Sudden, catastrophic events might include earthquakes which rupture reservoirs and emit the CO2 gas directly back into the environment. Slow, long-term effects could include chemical reactions between stored CO2 and surrounding groundwater; some evidence suggests that this could result in heavy metals in drinking water reservoirs (p94).
Opportunities and challenges:
Soil carbon is currently at half of preagricultural levels (because of intensive farming practices), so there is opportunity to store much more carbon in soil while also improving soil productivity (p83). Biochar could be a piece of this practice, though its integration currently requires tillage of the land (which can be environmentally destructive) (p84).
Potential sources of biomass to pyrolize into biochar include crop residues and forestry waste. However, both of those include yet-unsolved logistical challenges, and might be environmentally destructive to collect (p84). Is there a better biomass source available? Can forestry and crop waste be collected effectively and nondestructively?
Oil and gas companies already use CO2 pipelines and injections to harvest oil, so there is strong technical feasibility for transport and underground injection of CO2. We could even use the existing infrastructure, as a profit incentive for oil & gas companies. Is there a way to use this infrastructure and build additional for carbon sequestration? (p90)
Improve the efficiency and deployment of carbon-sucking “trees” and other artificial techniques. Key areas of work with the “trees” mentioned are the energy required to keep the sorbent fluid circulating (especially in high wind), and the heat-intensive CO2 gas extraction process from the aqueous solution of calcium hydroxide (p89).
Figure out a long-term place to put sequestered CO2 such that it cannot rupture and leak into the environment.
Chapter 6: Biofuels
Reality: Biofuels could someday be a promising supplement to oil-based fuels, but right now massive deployment relies on processes that have not been proven commercially viable. Additionally, harvesting the biomass for biofuels can be environmentally destructive.
Myth: We can replace all of our gas and oil use with biofuels like corn-based ethanol.
Smil’s main points:
Ethanol is not an efficient energy source. The energy content of ethanol is 65% that of gasoline (p98).
Corn-based ethanol requires more land than we can use. If all of America’s gasoline were from corn derived ethanol, the growing of corn to cover American fuel use would require 220mil hectares, 20% more than American arable land (p101).
Ethanol-providing crops can contribute to environmental degradation. In corn crops, nitrogen fertilizer runoff is a key negative effect (p103). In other crops, such as sugar cane, expanding need for arable land can contribute to deforestation (p101).
Opportunities and challenges:
Sugar cane is better than corn for ethanol production because it requires minimal fertilizer and has a higher power yield per hectare of planting (p104). The United States has high tariffs on Brazilian sugar cane ethanol, so it is not commonly imported into the United States (p105). Is there a way to grow sugar cane (or a higher power yield crop) in a place with more favorable trade conditions where sugar cane can be sustainably grown?
Cellulosic ethanol is a promising technology to turn waste into biofuel, but it is yet unproven as commercially viable. Though it will take decades to scale up this industry, it is worthwhile to research potential processes for creating cellulosic ethanol (p108).
Since energy density is lower in biofuel than in gas (and we have inefficient vehicles) miles per gallon would be low and fuel weight could be significant in a biofueled car (p114). Can this inefficiency be decreased?
Chapter 7: Wind Power
Reality: Wind power has several challenges, particularly in infrastructure and height of wind harvest, to overcome before it can come close to promised power production quantities. If you’d like to delve more deeply into wind power challenges, I found Ramez Naam’s post on the subject more thorough than Smil’s chapter.
Myth: Wind power will provide all or nearly all of the energy we need in an infinitely renewable manner in the near future.
Smil’s main points:
Wind can theoretically be a major source of renewable energy, but never 100% of power needs. In 2007, global electricity production was 1,800 TWh (p125). Globally, about 1,200 TW is dissipated within 1km of the earth’s surface– and therefore harvestable (p121).
Wind power requires a lot of land (in windy places). A reasonable assumption of wind power capacity factor (the actual power output divided by the maximum theoretical output) is no higher than 25% based on measurements in Europe, so 4.1 TW of installed capacity would cover half of 2007 global power needs — which would cover a space equal in area to four Frances, assuming 2 W per square meter (p125).
Opportunities and challenges:
North America is particularly well suited for wind power generation because there is a high land area of strong winds areas distributed across the land. But the continent also has prolonged calms and excessively strong winds; both conditions halt wind power generation (p128). How can wind power be stabilized or complemented to provide steady renewable power at peak times?
There are not currently many high voltage power transmission lines from America’s windiest sites to its most populous cities. How can we get wind power to where it’s needed efficiently?
Winds are steadier at higher altitudes, but transportation logistics of very tall wind turbines is already a challenge for the technology. What are creative ways to harvest high-up wind energy without requiring the transport of massive structures? (Companies such as Altaeros have creative approaches to this problem.)
Chapter 8: Pace of Energy Transitions
Reality: It takes many decades to transition between energy types. Humans took hundreds of years to move from wood, to coal, to oil, and we should expect a similar timescale to move away from oil and coal to any next energy staples.
Myth: We just have to solve a few key problems, and then we can expect mass adoption of renewable energy sources in our lifetimes.
Smil’s main points:
Energy transitions are slow by nature. Oil took 50 years to climb from first commercial production to a 10% market share, and we continue to depend on prior dominant energy forms: coal, wood (p138).
From data up to 2008, we are not currently transitioning off of oil. In 2008, energy from new renewable energy sources was less than 2.4% in the United States. The American dependence on foreign oil has climbed steadily since at least 1973 (p135).
Quick energy transitions destabilize economies. Any new technology adoption requires a heavy up-front investment (estimated by Smil at at least $3 trillion. This is needed both in the energy sourcing and transport infrastructures, and in the “primary movers”, the major users of the new energy (such as cars). Primary movers take years to become efficient (p138). Quick changes in primary energy sources leaves less time to build infrastructure and primary movers. Transition also requires people who had invested in the old system to write off of that infrastructure investment (p142).
Renewable energy doesn’t work well with our existing energy grid. This is an energy transmission problem; the population centers of the United States are at the coasts, but the best spots for wind and solar are far from there. We don’t have high-voltage transmission lines between them, and thus no good way to move that power (if generated) to where it is needed.
Opportunities and challenges:
To what extent is it possible to adapt existing infrastructure to clean energy sources?
How can we anticipate the market by creating primary movers that will work renewably with the new energy system we seek but also functionally within the existing system?

I liked this book because it was straightforward, in good faith. Smil is uninterested in convincing you that climate change is real, or that we need to change our energy usage, production, et cetera. He just wants to explain, in detail, why widely touted solutions and expectations will not work.
I encourage you to read this not as discouragement, but as a starting point and an opportunity for further ideas in the space. I enjoyed Smil’s work, but found him quick to write off genuine improvements in our carbon economy just because the effect they can have is small.
Challenging problems, well defined, make for a good place to start.