Good faith roleplay can still have problems.
For example 5e more or less makes Int a dump stat. So it is easy to have a party of entirely or vast majority idiots.
But how many DMs police the group to tell them "None of you are smart enough to think of that" or "You all are dumb. Roll to see if you are smart enough to think that"?
This almost never happens. But the smart player running a dumb character is rarely punished for accidental metagaming at best and purposely power gaming at worse. And it all can be in good faith.
How often do DMs in good faith move the focus of a challenge not due to the character's activity but their attributes?
How often does accent, height, age, rank, attire, language proficiency, race, beauty, or attitude come into the success or failure of a freeform roleplay?