Bridgewater's approach sounds interesting, albeit not for everyone.

Ideally, you want to catch mistakes sooner rather than later. People don't want to admit they made a mistake; or people don't want to sincerely criticise others.

One way of fixing this is creating an environment where people feel safe about making mistakes.

AFAIU Bridgewater's approach is the opposite; everything gets exposed. e.g. Modelling different personalities that people have, or whatever, so that these can be taken into account. The upside is stuff like you can tell the CEO you think he did a bad job. The downside is just how exposed it feels compared to normal social/political interaction.