I find both the PMEG's argument on one side and Sudanglo's manĝo argument on the other very convincing. I've always been a firm believer in the flexibility of the Esperanto to cover both meanings of korekta, but I find myself leaning to the PMEG's revised argument against it now.

The important fact that use of "korekta" as ~"senerara" is part of a minority words commonly used this way (e.g. suspekta, komplika) weakens its position as a general word-building concept. And if the method was applied to any action-y root, it seems like a great deal of ambiguity would arise.

And given that the rule only applies to a minority, it seems less like a rule, and more like improvisation before a good rule-based guide came along.

Whereas with the "o" finaĵo, where the meaning allows, takes on results of actions, the action itself, the manner of the action, the action as session (e.g. "kuro" for "a run" ) etc. as its meaning.

"Manĝo" as "meal" in my opinion then is eating as a session, "an eating". Seemingly inline with usual word building rules that apply to everything else. Any implied notion of prescribed times and courses are pragmatics we infer like we can with any word used often in particular contexts.

The hardest argument to counter is: why worry about "korektigi" when we don't worry about "manĝigi"? My counter being, that the reason I'm no longer going to use "korekta" for "senerara" is not because "korektigi" is ambiguous (though now I won't have to worry about the ambiguity), but because of this notion that its usage was more improvisational on a minority of words rather than based on a logic that can apply generally.

sudanglo:Really, Tom? Doesn't limiting korekta to corrective entirely follow from the view that korekt is a
verbal root

It may well follow from that view, and of course the root-class theory expresses that view. But that's not the same as wishing to limit the meaning of the word purely for the purpose of conforming to theory. It's about conforming to the norms of Esperanto word-building. The theory simply reflects those norms (rather well, IMO).

sundanglo:The potential ambiguity of korektiga (is it derived from korekti or korekta) seems to me a small price to pay for the usefulness of korekta

On this point I agree, and I don't personally agree with PMEG's shift in stance. I find its previous assertion that "tia ĉi vortfarado estas hejma en Esperanto de la komenco", and that such words have "frape montrata vivantecon" much more compelling. I'm not going to stop using korekta, komplika, veka, falsa, fuŝa, fiksa, suspekta etc in the way I always have done because of what seems to me to be a very minor risk of confusion. But I respect the opinion of those who want to make Esperanto more logically consistent with itself, and I do not accept that they are simply pandering to theory.