99.99% of the time, rejecting a suggested edit that changes code is the right thing to to. The competence of the reviewers isn't really the issue. The author should have just used his original account to make the edit and this would have been a non-issue.

He probably didn't realise he had two accounts.

The incompetence of the mods is relevant because it is one of the main criticism of TFA. It's a nice example of the Pete principle.

If a mod doesn't understand an answer he should just pass, not vote according to what's right most of the time. They are incompetent, both regarding LuaJIT and as moderators.

How many LuaJIT experts do you suppose there are reviewing edits on Stack Overflow?

He used to be there. You can't find a better expert.

I'd expect mods familiar with Lua to recognize him, and he should have mod powers, to begin with.

He doesn't even recognize that he's using two separate accounts, but other people are supposed to recognize him and his accounts should have mod powers?

Again, if he had just used his original account to edit his answer, he wouldn't have needed any of that.

Yes.

He's a world class expert in compilers and high performance, low level programming on modern hardware.

This addresses the core point of the article, being that the SO "meritocracy" is badly implemented, and conflates heavy site usage with maturity and technical ability.

Edit: If not a mod status, a VIP status. SO would gain a lot if its staff were to individually recruit people as talented as he is.

This seems more like an argument that SO's "meritocracy" isn't perfect, not that it's poorly implemented. Having two accounts is an edge case.

You're using the fact that he had two accounts as an excuse, but it really doesn't hold up. This was an edit with inarguable merit. Yet users rejected it because they were using the heuristic "changes to code is bad" as a substitute for understanding the edit. That's bad moderation. If you don't understand a change, you shouldn't vote on its worthiness.

So instead of taking the one in a thousand chance that its the wrong thing to do (which, from what I've seen, is generous) they should let edits languish for days or weeks until someone sufficiently knowledgable shows up? That just doesn't sound like a worthwhile tradeoff.

If something should not be edited through another account, then the software should just not allow it.

If it allows it, and delegates the check to moderators, then these moderators should either reject it if they are sure the change is wrong. In all other cases they should allow it.

It's really that simple.

You have no idea what the spectrum of edits is on SO. Why do you think your heuristic is obviously better, despite making no argument for it?

The heuristic they use is something along the lines of "only accept edits that are grammar fixes, equivalent thereof, or rewordings". Content changes without the authors consent are considered poor form.