So the problem I see is that this then means that Pathfinder has been designed with the _explicit_ expectation that everyone is a character optimizer.
50%, yes.
But, they also eliminated or improved the bottom 20% of "core" Feats, getting rid of some trap options (Toughness and Skill Focus, I'm lookin' at you). They improved the classes that are most affected by poor optimization choices, so that they're "pretty good" out of the box.
Also, in Paizo's defense, 10+ years into a game system, I'm not sure "is aware of the balance of choices" is an unreasonable expectation of your audience. I would certainly have to look pretty hard to find a gamer who can't discern between "good choices" and "bad choices" pretty quickly - and decide if a "bad choice" is worth in for flavour reasons.
And regarding "new" players -
First, Paizo hasn't yet released Pathfinder Basic, so we don't know how they'll get new players up to speed on system mastery; Pathfinder, the core game, is not, I would say, an "introductory" game - not that it's particulary hard to learn "the basics", but there's a whole lot of spells, rules, feats, and pages to that book...
And second,
most (not all) new players are coming into established groups, that can warn them about "bad" options (even if the new player isn't sure why they're bad yet). If the group
is entirely new, well, presumably they're starting at first level, where the "bad guy improvements" havn't started (or at least, aren't as in-depth).
Do most 3.5 groups find a Balor a CR 20 encounter? Is a Marilith really CR 17? I'm saying those numbers are <more accurate> in Pathfinder - and it seems that the counterargument is, "Drat, that means I'm going to need to be more optimized!"