I don't think the concept is easy for people to understand at all. It's just darwinism really, but people think it's about punishment. Note that darwinism doesn't just reject bad ideas, it has a rather high and variable false positive rate. I.e. it's unfair, for example Ramanujan died prematurely despite being a genius and a gift to his species. But it is overall also quite effective.

If you can game the numbers there's no real skin in the game. And people do! Just sell volatility ("everything will always be fine") regardless of the conditions, it will work wonders, until you blow up. The difference is that if you blow up your own money you cannot continue the game, whereas if you blow other people's money - you can.

You seem to believe that "skin in the game" guarantees the survival of the fittest leaders. This is only true assuming that the selection criteria are good ones and ones that cannot be cheated.

As long as anyone is capable of gaming their numbers, it floods out honest people in favor of cheaters. It's very powerful at selecting for whatever your metrics are. When you have the wrong metrics or ones that can be cheated, it's very powerful in an extremely bad way.

In addition, the general public and the media are absolutely not good at this, or we would have far better leaders than we do.

I think you're mainly describing the optimal culture for the organization or workplace, but either collectively or perhaps even just the leaders, they must recognize the importance of their decisions and the possibility to cause deaths. That's why research like what you mention is important, it shows how organizations and leaders with "skin in the game" can achieve success.

Engineers create systems where individual vigilance is less disastrous, or where vigilance is augmented. The real world and tech world still does rely on brute-force responsibility sometimes, to cover the gaps in a plan where the details were left out.

Exactly, the complexity of the "real world" far exceeds that of what a typical engineer deals with. This requires more darwinian, rather than cerebral, processes, simply because human brains are too small to understand all consequences.

Nothing about safety modeling requires that we understand every risk in the model. I don't understand how you think that, e.g., aviation safety modeling doesn't have to deal with the real world or with real people who do stupid things. It's true that we're probably less able to control all the risk factors in some cases, though, but nothing prevents us from understanding them or doing what we can to make better systems for future pandemics. E.G. we could come up with a better plan for testing and perhaps better ways to share (or leak?) data on new diseases.

The fact is that the public are terrible bosses. We look for people to blame, we're inattentive and have no idea, generally, how to evaluate things on actual metrics. We have no idea which metrics are being gamed and the people who do are actively gaming the metrics and flooding us with cherry-picked information as well as disinformation, though the former is probably more dangerous and more widespread.

The average person in the public will be satisfied as long as we blame some random politician instead of looking for structural reform. E.G. maybe we should adopt Japan's culture of using masks in public when sick, require employers to guarantee minimum standards of sick leave, and change various hospital/ER waiting room practices.