Early reactors required active safety mechanisms: You had to actively do stuff (such as pumping coolant) to keep them from melting.
Modern reactors are designed to be far more "passively safe" such that failures cause the reactor to shut down safely even if the operators do nothing.
Fukushima failed because power failed, failing the coolant systems. That was an active system and not a passive. Look at MSR test that were done at the Idaho National Research Laboratory. The heat transference of using molten salt shutdown the react even in the absence of power. This is a passive system.
The EBR-II tests you're referring to were liquid metal sodium, not molten salt. Molten salt is much different (though it also can be used in reactors and may be able to demonstrate passive safety as well)
Three meltdowns, up to 1 death total from radiation (acutely and estimated long-term). That's a huge improvement over Chernobyl, which killed ~60 acutely and up to 4000 long term from early cancer deaths.
The real death toll of fukushima is the extra climate change deaths resulting from the hysteria it produced. Radiation wise it was really not that big of a deal.
It is well known that most reactors (at least here in France) suffer from bad engineering and are unsafe. Existing ones suffer a lot of downtime due to various levels of accidents, while ones being built are often delayed due to serious issues (see also: EPR scandals in France, Areva scandals elsewhere in the world).
>What is your evidence that the tech "becomes safer with time"?
Sorry but this just comes off as an extremely ignorant comment. Modern reactor designs (gen3/gen4) are very different from the old Chernobyl/Fukushima reactors, have more safety features and often are passively safe in case of meltdown or coolant loss.
Likewise, Lehman may not even necessarily represent the tail risk of a financial crisis, after all the other banks didn't also collapse.
Extreme Value Theory tells us it is hard to calibrate estimates of tail risk but it's not as if we have no data and no grounds with which to do it in the case of nuclear power.
Flippant analogies that are actually flawed analogies upon close examination aren't helping. If there's reason to believe that modern nuclear plant designs could lead to large scale death with a non-trivial probability, far beyond the scale of Chernobyl which had a very small death count relatively speaking, then the case for this needs to be made by outlining the actual mechanism by which this can happen.
Regarding the significantly improved safety of modern plant designs, this was addressed by other commenters.
> What is your evidence that the tech "becomes safer with time"?
Why does there have to be evidence for this? Did all technology associated with nuclear safety stop improving at some point? Does it ever stop with anything?
Modern reactors are designed to be far more "passively safe" such that failures cause the reactor to shut down safely even if the operators do nothing.