It's very safe if you don't disable safety systems put in place (like it was explicitly done in Chernobyl to do dangerous experiments, even though they labeled it as "routine tests"). And if you don't ignore many warnings given in the industry (like TEPCO ignoring many recommendations of raising their tsunami protection walls many years before the catastrophe, like other plants closer to the epicenter that were unaffected).

So the chance is much lower than people think, if plans have measures to avoid such corruptions. And that's not counting the many safer designs we have nowadays. Many plants today don't even need backup electricity to automatically shut down by themselves.

Good engineers take the human factor into account and include backups in their designs to handle failure modes caused by the human factor.

What they can't do is work around politics or budgetary constraints which force trade-offs between safety and cost. Anything can be made as safe as necessary, if you're willing to pay for the redundancies.

When it comes to reactors on Earth, a very large proportion of the budget is diverted to bureaucratic paperwork, lobbying, and PR, which could instead be used to pay for additional safety measures.

There are some "human factor" problems that engineers cannot and should not be expected to resolve.

I mean, how can someone design a communist-proof nuclear reactor?

Make them pay to get in?

Wouldn't work. They know how to steal and can even make fiat currency outside of the free market context.

Not a bad idea though.

Sure, but that was a) 31 years ago, b) a reactor built without the safety standards we have now, and c) a reactor operated without the safety standards we have now.

As far as I understand, making a reactor built in the last 25 years go 'wrong' in any way like that is very difficult, and would take much more than negligence or incompetence.

Fukushima is an example of this – the reactor that had issues was over 50 years old, but all the modern ones were fine.