Can you show me where in 5e D&D, the general population is generated by 3d6 in order? AFAIK, the general population is given flat stats same as any other monster. There is no D&D chargen system for general populations in 5e.
Since you either missed it or forgotten it, here it is again:-
RPGs can be thought of as if they were species of life. Instead of cats, dogs, hamsters and so on, we have D&D, Runequest, Stormbringer, 13th Age, and so forth. Instead of mammals, birds, insects, and others, we have Fantasy RPGs, Sci-Fi, Superheros, and other genres.
And, just like living species, these RPGs evolve; they change over time to suit their environment, and may die out if they no longer meet the requirements of the consumers.
In the same way that you can study the DNA of life-forms to understand the past of a particular species, or see if some are closely related to others on the Tree of Life (or even previous forms of the same species), we can study RPGs in the same way. We can see which games are direct ancestors/descendants of which other games; check their metaphorical DNA.
D&D 5E is the current phenotype of D&D. Earlier versions still exist, and there is little doubt that the game will continue to evolve (unless it fails to adapt to the changing needs of the player base and just dies out).
In this context, the characteristics of D&D 5E can be traced back through its ancestor editions. We can see where certain changes happened. Like with life-forms, those characteristics either change or stay the same. If they don't change (and that change may be appearing or disappearing), then they stay the same.
Let's take Armour Class. It has evolved over the editions in that 'higher is better' now when 'lower is better' is what it used to be. But it has not changed in one crucial way: it models armour as making the wearer harder to hit, when most species games model armour as having either no impact on how easy it is to hit the wearer or makes it even easier to hit the wearer, protecting the wearer by reducing or eliminating incoming damage.
What the heck am I banging on about this for?
One of the things that has remained unchanged over the existence of every incarnation of D&D is the 3d6 bell curve as the background against which ability scores are measured. It's right there in the metaphorical DNA of 5E, even if it isn't directly outright stated in any 5E book. We can study what is written in 5E and clearly see the truth of it by studying ithe continuing impact of the 3d6 bell curve.
At its core, the general population is modeled by the 3d6 bell curve. By that, I mean that the game assumes that every NPC member of every playable race has ability scores as if they were randomly rolled, in order, on 3d6, and then any racial modifiers applied.
The population as a whole was assumed to be as if fairly rolled randomly on 3d6. As players, any PC we make is conceptually assumed to be one of that population. It's as if we looked at the population of the game world and said, "I want to play that one!"
Although conceptually we are choosing to play an existing person, and that person's ability scores are on that bell curve, in order to actually do that then the DM would have to randomly generate ability scores on 3d6 in order for every single person in the world! The players can then choose which one's to play, presumably playing the one's whose ability scores were on the high side. In this theoretical bell curve population, around half will have higher than average stats and half will have lower than average, and it's likely that the people who are most likely to be adventurers are also most likely to have higher than average ability scores.
Now, that's a lot of work to end up with 4-6 PCs! So Gary Gygax realised that you can approximate this concept without the DM needing to generate 7 billion NPC character sheets! The players can roll stats themselves, using one of various methods that would skew the 3d6 bell curve to get higher than the average results for PCs as compared to that unmodified bell curve.
But the crucial thing to remember is that whether or not you are using '3d6 twelve times arrange any six to suit' or '4d6k3 six time arrange to suit' or later point-buy or array, every single resulting PC is conceptually one of the population that (probably) got lucky on that 3d6 bell curve. Every single PC, no matter how they are actually generated at that table, is conceptually one of that population that was generated on that 3d6 bell curve.
The result is not only that any method must be within 3-18 for any stat (before racial modifiers), but that every possible combination of 3d6 six times is a valid character concept. In any population, any combination of six stats of 3-18 (before racial) is a possible PC, even if it is unlikely. Six 18s would be very unlikely, as would six 3s, but each is a possible result of 3d6 six times and therefore a possible person and a possible PC concept.
Six 18s (or six 3s) would be very unlikely, and we know exactly how unlikely. Six 10s is also unlikely, although a lot more likely than six 18s or six 3s! The chances of six 18s or six threes are 216 to the power 6 (or about 1 in 5 x 10 to the 74th power, if my calculator is to be believed). However, how many times the DM has to generate an NPC before he actually rolls those six 18s may be fewer or greater than that mathematical expectation.
The point of all this is that the valid idea space of any (pre-racial) set of ability scores remains any combination of 3-18 six times. Any method that takes any of those possibilities away is also taking away valid concepts attached to those scores. Any method that limits your starting pre-racial scores to (for example) between 8 and 15 has taken away every concept that included a score of less than 8 and/or more than 15. Such a method therefore certainly removes valid character concepts. In comparison, any method which allows any result of 3-18 in any score (4d6k3, for example) has not taken any concept away.
Therefore, a statement like 'this method lets me play any concept I want' can only be objectively true if that method allows any combination of 3-18 for six scores to be generated by that method.
Point-buy does not allow that. Rolling does.
