A few years back, politicians in India were fighting over whether to build yet another lavish temple to Lord Ram. On one side, people were complaining “he’s only a Hindu deity! The government shouldn’t favor Hindus at such great public expense.” The other side had a strong counter-argument “No, Ram is not just sacred to Hindus, Ram is god of the whole world!”. Since that was the position of the ruling BJP party (basically Hindu nationalists), it prevailed.
Just thought I’d give an example of an impenetrable echo chamber Westerners might understand because they likely don’t agree with the BJP that Ram is the god of their world 
The PSF paid Sage Sharp to begin with to help write the new CoC. And now pays their company for ongoing CoC enforcement training. It’s not a good look to be so reliant on a single source (well, in some political circles it may be the only “acceptable” look - but, no, it’s not really true that everyone believes “Ram is the god of the world”, or even that everyone “decent” does).
The training itself appears to consist of a 4-hour Zoom course. In matters involving conflict resolution, I can’t imagine that it’s possible to do more than scratch the surface of the potential difficulties in what amounts to a 4-hour video call. Conflict resolution is about the most difficult of all issues a person can tackle, because it can involve that worst of human nature, at both individual and group levels, in a context that provokes the worst of human nature at both levels.
I expect (but don’t know) that most cases that come up are of a “slam dunk” nature, not really particularly nuanced. In which case, fine. But perhaps outside bona-fide experts could/should be brought in for cases where it’s unreasonable to expect well-intentioned but lightly trained volunteers to successfully navigate truly challenging issues. That’s a kind of compromise with this depressing, but realistic, comment recently made on the latest LWN article:
I always got a kick out of seeing this stuff likened to a corporate HR action, because that so blatantly begs the question: of course it is. It’s - by design - purely dictatorial, one-sided, opaque, and unaccountable. At a PyCon (or other face-to-face meeting), concerns for imminent physical danger may amply justify harsh snap judgment, but bans for posting online appear to drag on for months behind the scenes.
“Stop pretending” seems excellent advice to me regardless. But I don’t share dberlin’s conviction that it’s essentially impossible to get an approximation to “a reasonable fair process” without vast effort in all cases. Then again, I expect dberlin is a lawyer, and I’m certainly not.
Please don’t tell me that we must take the same approach to all cases. That’s never been true. For example, when David Mertz got booted from Discourse:
So they’re already quite aware of that visibility does matter in practice. It always will in the real world. “Don’t pretend.” Practicality beats purity. “Simple process for simple cases; complex process for complex cases” is thoroughly principled in its own way.
Another possibility was mentioned long ago by @willingc, who pointed out that Jupyter’s processes allow for mediation instead if all parties agreed to that in advance.
Etc. A major frustration here is that many ideas have been suggested across years, but nothing visible ever changed as a result. Is that meta-problem eternal? Remains to be seen.