I didn't say that. I said it wasn't actually broken to begin with. Especially because, as Chaosmancer mentioned, a lot of the monster statblocks don't have traits the PC races have.
I was going to reply to Chaosmancer on this point, but two birds one stone.
Bit of D&D history: I remember reading the opening chapter of the AD&D 1e DMG back in the '70s. There was a section giving, IIRC, six different ways of generating ability scores for PCs. The two methods out of the six that I saw get used the most were '4d6 drop the lowest, six times, assign those six scores to the six abilities as you wish', and 'roll 3d6 12 times, assign the highest six to the abilities as you wish'.
But what was all this for? Because the game assumed that the average 'adventurer' was 'better' (had higher ability scores) than the average of the population, and these methods produced on average higher scores than the population as a whole.
But in order for this to be meaningful, we have to know what these 'better' scores are better
than, right?
So how did your 'ordinary' person generate their ability scores? Just assume all abilities are 10?
No. The bell-shaped probability curve was the baseline. Specifically, assume every person rolls 3d6 for Strength and keeps whatever they roll, then roll 3d6 for the next ability, and so on. No assigning scores, no 4d6k3, just get what you're given. That's the baseline population, against which PCs are 'better' than in terms of probably having higher scores, and these six recommended ways are how to get those 'better' scores.
The Judges Guild produced one of the largest products for AD&D 1e: The City State of the Invincible Overlord. I had it. I read it.
Every person in the city had six ability scores rolled in order on 3d6. They weren't assumed to be all 10s, nor were there five or six standardised stat blocks for Guard, Cultist, Thug or Bandit. Each was rolled, because that's how people's ability scores were generated in the metagame of the D&D worlds.
Since then, it was realised that this was far too much work! Why not just have a handful of statblocks for guards, bandits etc? Why not have a Commoner statblock for when we just can't be bothered to completely roll up some totally unimportant dude? And since the average of 3d6 is 10-11, why not just assume they are all 10s?
But these helpful DM shortcuts
do not change that 3d6 baseline! Just because 10 is the average on 3d6, this doesn't mean that every single NPC actually has 10 exactly in every ability unless you actively change it!
And when the creatures in the monster manual were finally made the same way as PCs, with all six ability scores, the creature stat blocks for, say, Drow, didn't mean that every Drow had the exact same scores!
The creatures in the Monster Manual are just as much DM shortcuts as the all 10s Commoner. The DM has certain requirements for monsters in a scenario, and the MM gives useful statblocks to reflect this.
So ANY statblock in the MM is not a statement that ALL lizardmen (or whatever) have these exact ability scores. They are what they are to provide the challenge the DM requires.
The PCs aren't going to be fighting average lizardmen, they are going to be fighting
warrior lizardmen, who have already put their high rolls is Str even though lizardmen do not get a racial bonus to Str. Or indeed, those who were the tribe's warriors are the one's born with (rolled) high Strength scores. Those warriors may also have something special about them (like a bite attack) that is non-standard for lizardmen, in a similar way that casting 1st level cleric spells is non-standard for humans.
This was illustrated in 3e, when occasionally a MM may give rules for creating a PC from a 'monster' race. It gave rules for Drow and Snirfneblin and so on, and the racial adjustments to ability scores didn't map directly to a 'monster' Drow statblock.
So, yeah, the rules for creating a PC race are applicable to NPCs of that race too. But other things may further modify those NPCs, just like further rules (like class and level) can modify PCs.