I specifically put no value judgement on "metagaming" in that post. I was simply pointing out that the poster you were quoting didn't "leave out" as you say, your #4 and #5, they are actually just other types of metagaming.
Where you say "If I'm playing a game, I'm going to engage with the mechanics of that game." Others could just as validly say; " I don't care if it's sub-optimal or not, If I'm playing a Roleplaying game I'm going to ROLEplay."
Many of the things you pointed out such as making decisions based on how XP is received for example, obviously tend to lead to decisions that make less sense for the characters and story than if one more strictly considers the characters motivations.
Some people care about, some don't, obviously.
I feel you've somewhat missed my point. It relates back to the discussion with [MENTION=6801328]Elfcrusher[/MENTION] about
player agency. I'm also not sure how you are both denying making value judgements and saying that decisions make less sense. That looks like a (negative or pejorative) value judgement to me. (And [MENTION=23]Ancalagon[/MENTION] said that the issue of retreating is a "bad gaming" problem. In agreeing with that, you seem to be sharing in the (negative, pejorative) value judgement.)
It's not a necessary feature of an RPG that mechanical concerns and "motivational" concerns cut across one another. When the game was invented, by Gygax and Arneson, they didn't. In many contemporary games (eg 4e, BW, DW) they don't.
The reason that they tend to cut across one another in 2nd ed AD&D is because 2nd ed AD&D adds basically no mechanics to Gygax's game, yet assumes that players will play characters with motivations very different from the dungeon-raiding that is the principal focus of Gygaxian play.
To apply this to retreating: if I, in playing my character,
don't actually know what my chances of retreat are, and have
good reason to think that if I retreat I will be hunted down, but
do know, roughly, what my chances of success in battle are, and furthermore
I know that battle is not certain death, then it is rational for me to battle. In other words, one consequence of having relative certainty around the combat rules, yet relative obscurity around retreating, is that fighting battles becomes a part of most rational character's motivations.
In real life, of course, people retreat
because they don't have the requisite degree of certainty about combat. If you make a RPG in which combat is as uncertain as retreat - say, Runequest or low-level Rolemaster - then you might get more retreating.
Or you could flip it around: make the rules around retreat as clear as the rules around combat - which Gygaxian D&D does - and you will get more retreating.
As thing stand in post-Gygaxian D&D, though, you are asking players to divorce their sense of character motivation from the actual mechanical levers available to them in the play of the game. How is that good RPG design?
EDIT to make this more concrete:
13th Age has a rule that players can declare a retreat, and the GM is obliged to narrate their narrow escape, dragging fallen comrades with them, etc; but the GM is also then entitled to narrate a significant story loss.
That's a mechanic that aligns player mechanical levers with effective character motivations.
In 5e, you could allow a player to spend Inspiration to escape without needing to make a check; we might add that players without inspiration canhave their PCs come along for the ride provided they're in the minority rather than the majority, but they are automatically reduced to zero hp as they flee, and have to be dragged out by their comrades.
Or, if you don't want to put it all into the metagame, you could have a rule more like the Gygaxian one (and Burning Wheel has a somewhat similar rule): once fighting characters are out of engagement distance (10 feet in a dungeon, 10 yards outdoors) then initiative is no longer tracked, attacks may not be declared without having to trigger a new combat sequence, and the evasion & pursuit system is engaged (indoors it's based on movement speeds and distractions; outdoors it's based on a moderately complex percentile chart). In 5e, this could easily be an aspect of the exploration mechanics, which would also make rangers especially good at leading effective retreats. Which seems to fit nicely with the archetype.
