Remind me again why this is even a discussion?
Long ago in an edition far, far away there was a mention of rolling 3d6 for ability scores that has never been mentioned again yet is somehow relevant to the discussion of how to generate ability scores for PCs?
Okay, step by step:-
* whatever ability scores a creature has can only be understood in the context of the scores of other creatures in the world. To illustrate this, let's say you're picking up an RPG you have never played or heard of before. You skim the rules, and come across the stat block of an NPC. That NPC has a Strength score of 27. Is 27 good or bad? Is it weakest possible (ability scores are generated on 25 + 2d10)? Is it fairly weak (1d100 for each stat)? Average (1d%/2)? High (10 + 2d10)? Superheroic (3d6)? You cannot possibly tell, unless you have context. In D&D, now and always, the context is that the 'normal' population's scores are represented by the 3d6 bell curve, and the method used to generate the ability scores of PC's (whatever method that might be) is deliberately skewed to give higher results
in the context of that bell curve.
* although the underlying assumption was (and remains) that bell curve, games have evolved techniques to make the DM's life easier. The DM doesn't
need to pre-generate 20,000 NPCs whenever the party reach a new city(!), he can just invent 'reasonable' stats for NPCs the party meet on the fly; 'reasonable' in the context of that bell curve. These other methods of generating NPCs did not replace the background assumption of that bell curve, they
assume that bell curve
to this day! Even the stat block of the commoner in the 5E MM reflects that bell curve, as are the stats for every NPC in that section, where their stats are judged (good or bad or average?) in the context of that bell curve. In fact, the bell curve is
so ingrained into the structure of D&D that the writer of the rules assumes that
we all already understand this so he feels no need to explain it!
* since the general population is assumed to be the bell curve, then any valid character concept must fall within that bell curve, or be measured against it, with credible explanations for stats that go beyond it. This also works the other way: any arrangement of six scores from 3 to 18
is a valid character concept, for PCs and NPCs. Any PC stat generating method that takes away any of those possible scores from 3 to 18 also takes away the concepts that go with those scores.
* another consequence of the bell curve is that all creatures are
not created equal! Any method that makes all PCs equal not only takes away the vast majority of valid concepts but also has sacrificed realism on the alter of game balance.
Although sacrificing realism for game balance
is a valid choice, and a valid personal preference, what it most definitely is
not is a method that 'allows players to play the concept they want', since it denies the vast majority of valid concepts.
That is the source of this part of the discussion. It's not that anyone's preference is a problem, it's that some claims made about a particular method are demonstrably not true.