On the pro side, the book is short and full of instructive examples. The insane unforced errors surrounding tetra-ethyl lead, and the pretty competent responses to unforeseen side effects of DDT and CFCs. Smil then details the mostly-failed dreams of zeppelins, supersonic flight, and cheap fission, which seems to have mostly-failed due to competition from planes, insufficient demand, and technical challenges + regulation, respectively. Finally, Smil talks about hyperloops, self-fertilizing cereals, and fusion, which seem to be silly, unexpectedly hard, and even harder than expected, respectively.
On the con side, I think I found the book somewhat unsatisfying. What I wanted was a dissection of whether these failures were plausibly foreseeable in advance or not, and what kind of mistakes the techno-optimists of the past made. Alternatively, some heuristics for what kinds of technical challenges are tractable and what kinds are unexpected swamps like nitrogen fixation. Smil doesn't really seem to make that sort of thing his mission, but some lessons are extractable nevertheless.
Notes:
• Discusses inventions that were promising but had unexpected problems, inventions that were hyped and didn't meet the hype, and inventions that were anticipated but never arrived
• Ethanol in gasoline was originally meant to be an anti-knocking agent! Not just blatant corn lobbying
• Tetraethyl lead was GM's patent, other anti-knocking additives weren't
• Smil suggests a pretty explicit campaign to cover up toxicity of TEL, tobacco style
• TEL was ultimately banned NOT because of the LEAD POISONING, but because the new and powerful EPA wanted to get rid of photochemical Nox smog, which required catalytic converters, which were poisoned by TEL emissions
• DDT was a unprecedentedly powerful insecticide but bad for the environment, most famously birds, and hazardous to humans in large exposures
• DDT played a pretty vital role in the eradication of malaria outside of Africa
• Smil suggests it might have worked out better if not for the overzealous crop spraying that led to massive environmental contamination and rapid development of resistance
• Refrigerants need to be volatile, nonflammable, nontoxic, nonreactive, lighter than air to not be an asphyxiant in the case of a leak, and cheap
• CFC ozone interaction not noticed empirically, but via theoretical concerns about chlorine-ozone reaction
• Replacements for CFCs were ~5-10x more expensive. Modern HFCs don't contain chlorine so ozone not a problem, but are greenhouse gases with 1000x-10000x the warming power of CO2
• Smil says people thought zeppelins were the future, but it's unclear if this was because of an overestimate of zeppelins or an underestimate of airplanes?
• Some zeppelins were used for recon and patrol in WWII!
• Hindenburg disaster was big because it was filmed, "first media event of the 20th century". Put an end to passenger zeppelins, but they were pretty doomed anyway, mainly because of speed and capacity.
• Fission power was originally thought to be uneconomical by Manhattan Project theorists, but initial development was pushed for standard Cold War reasons
• Discusses regulatory obstacles but doesn't discuss why those obstacles suddenly started being put in place (they were pre Three Mile Island and Chernobyl)
• Smil seems to also attribute a large share of the responsibility to the sudden collapse in electricity demand growth during the 70's, from decadal doublings to 1.5x, 1.2x
• Breeder reactors were supposed to be the future but didn't work out, apparently because the technical problems involved in separation were just difficult and expensive, and because the initial case for the importance of breeders thought U-235 would be rarer
• Supersonic flight was considered "the obvious next step in air travel" since the main salient fact about airplanes was their drastic increase in speed over previous modes.
• But it seems like the economics just basically didn't line up - supersonic flight means more drag, more fuel, narrower hull, fewer passengers, more expensive mechanical components. At the margins of standard subsonic jet liners, customers just preferred price and comfort over speed.
○ Classic "actual time saved goes as 1/x with increasing speed" sort of situation
• Potential lesson to be learned from fusion: the heuristic that makes people think it'll be good is "steady-state costs low", bc fuel is cheap and environmental externalities are low. But if the mechanisms required are super complicated, those probably require lots of expertise and maintenance, which is also a steady-state cost!
○ Intuitively, maintenance of sophisticated machinery seems like it has more "exposure to innovation" than fuel or emissions, which are more just "properties of the power source". And this is kinda true, kinda not? Innovation could definitely make fuel cheaper or mitigate emissions. But there are more obvious arguments that place bounds on those via energy conservation, whereas there's no obvious argument for why maintenance cost/complexity will be fundamentally constrained by physics.
• Man Smil just kinda comes across as jaded and bitter about innovation hype, which is fair, tech journalism sucks
• The last chaper makes the case that people think of innovation as like Moore's law due to availability, but that the material foundations of society are actually the limiting factor, and those technologies progress much slower / are much harder to innovate on
○ Smil says you can have electricity without microprocessors but not microprocessors without farming. Which is true but not relevant? The valid version of this argument is "you can dump lots of cheap electricity into expensive microprocessors to get more output but can't really dump lots of cheap microprocessors into electricity generation to get cheaper electricity"
○ Which seems semi-true, idk maybe you could eke out some improvements in power plants with arbitrarily powerful control systems
○ But mostly, oh BOY does this argument get fucked by AGI. Unless Smil wants to argue that AGI is impossible or that the current systems are near the physical limit and couldn't even be improved on by massive amounts of cognitive labor, then the possibility of AGI seems like it undermines the whole basis of Smil's "resource foundationalism" position. Like yes, those resources are important and hard to innovate on. But if we find a way to turn electricity into labor, doesn't really matter, line go up until physical limits.