With the caveat that I'm not picking any trade-off other than the time cost involved in looking up a few incomplete answers to forum questions that catch my interest;

it's extremely difficult to evaluate industry (broad industry, not just nuclear) safety value on the basis of deaths that have occurred without a solid understanding of the deaths and other costs that can occur should standards be relaxed.

The analysis on various Los Alamos et al. National Laboratory incidents during the early atomic days reveals that things easily could have been much worse, rather than three dead greater numbers could have been killed and expensive facilities rendered unusable. Carrying live but "safe" nuclear weapons about came razor close to accidental detonation on US soil near civilian population centres on a few occassions - these make studies for whether safety procedures were justified in time and expense or perhaps barely went far enough.

I raised Union Carbide Corporation as an example of what can happen in an industry if safety isn't headed, such accidents can happen in many industries and some have the potential to "salt the earth" for many many years past an event that immediately kills large numbers.

    Timing Toast

    There's an art of knowing when.
    Never try to guess.

    Toast until it smokes and then
    twenty seconds less

suggests the pragmatic answer to the question you pose is to reduce regulation until an acceptable death threshold is crossed and then regulate a tiny bit harder.

This can be difficult to do in practice.