I've had some more general lessons running around my head than the ones I identified in the previous post, but I haven't found a way to make it a coherent post out of them. So below I have just listed them out, in order to capture them before some new crisis displaces them.
- "Is" is the wrong verb. A lot of pro-nuke cranks are fond of saying, "Nuclear power is safe." This reveals a significant confusion on their part, because it is absolutely clear that nuclear power isn't inherently safe. Most large industrial objects are not. Cars, planes, trains, tanks, ore smelters, oil rigs, coal mines, oil refineries, combines, bulldozers - all of them are in fact quite risky when not taken seriously on a continuous basis. What the supporters should say instead is, "Nuclear power plants can be run safely." I believe that statement is true. I also believe that it is hard to make that sentence true.
- Tail risk is incomprehensible. Our modern high-technology industrial civilization is a very recent occurrence in the history of our little blue planet. As a result, accurate records about various geophysical phenomenon are very short, and probably non-existent for some types of events. Scientists have become pretty good at teasing clues out of the geological record, but the risk to nuclear plants in many areas is still not quantified. But even if it could be quantified, humans aren't good at assessing the numbers. For instance, a flight on a first-world commercial airline is safer than driving, but many people still refuse to get on a plane. Similar problems arise when talking about nuclear power plants. But in that discussion the risk is often either wildly over-estimated or completely dismissed, and neither is correct.
- Profit is corrosive to safety. As I've said before, here and elsewhere, I do not trust private, profit-driven corporations to run nuclear plants safely. The reason is that safety costs money. Resources spent on safety frequently won't translate into profits, in either the short term or the long term. But executives get rewarded for performance on a time scale that is relatively short. The massive mismatch between when cost-cutting might backfire and when an executive gets to drive new Porsche means that it will almost always be in a decision-maker's interest to go for short-term rewards.
- Technology matters only somewhat. When a safety discussion becomes heated, often people will join in to say that one reactor type or another is better, and if only that reactor had been used, this or that problem wouldn't ever happen. But nothing can be made fool-proof, and the specifics of implementation and operation can render inoperable a safety feature in any design . The organizational culture that surrounds the low probability, high impact nature of nuclear power plants is just as important as the technology that is in use.