Overall, where possible I rather not have a rule and only create them as needed. Once needed it is created and enforced consistently.
>How would you remediate brilliant jerks? How would you measure and incentivize working well with the rest of the team/company?
These are areas I'm not proficient in and thus I would first need more study.
One thing is that I don't really think there is a choice between two options, but that both options are the same. The only difference from social pressure is that the rule system is much more complex and not written down because people cannot tolerate facing the rule system once written. Imagine a law that says attractive people get less time in prison, would that stand? That women get less prison time? That whites get less prison time? Yet this is what happens when judges are free to given sentences and prosecutors are free to pick what charges to bring.
What I advocate is any rule must be written down, and thus any unfairness like the above must be faced and either codified and thus open for challenge or be done away with by removing the subjective application of a rule.
To bring it back to your original point, you expressed a clear preference for a letter-of-the-law approach, and for rules that are always applied and always enforced. You place the burden of fairness on the creator of the rule instead of the enforcer. That's an understandable approach. I don't agree now, but I did at one time, so let's run with it.
How would you define guidelines/rules for code reviews specifically, or engineering culture in general? Would you? How would you remediate brilliant jerks? How would you measure and incentivize working well with the rest of the team/company?