Ability Score Generation and PC Dynamics in Basic, 1e, and 2e

I’ll try to keep this brief-ish despite the enormity of the title. Reading the 2e PHB/DMG has gotten me thinking about how the early games use character generation to teach the participants about the game’s superstructure.
 
Basically, the main thing (not the only thing) that I’m mulling over is Ability Score rolls. OD&D was 3d6, 6 times, in order of Abilities. Basic does the same. B/X offers a Referee-side option of allowing a new *set* of rolls if the scores the player has are really bad – so not a new roll for the bad stat, but a new array.
 
Now, Gary himself comments on this procedure in the Ability Score Generation section of the 1e DMG (p11). He says it’s important to create viable characters for an ongoing game, and then, second sentence, glosses this: “While it is possible to generate some fairly playable characters by rolling 3d6...” He’s talking, of course, about OD&D (and Holmes Basic). This is classic Gary: assume your reader knows about some whole other thing and hare off on it. Basically, he thinks 3d6 in order takes too much time or leads to shonky characters in unwanted Classes. So he offers 4 methods in the DMG (and more in Unearthed Arcana) for generating scores.
 
Methods II, III, and IV are 3d6 systems, premised on creating ranges of scores or stat arrays for the player to pick from.
 
But you’d think Method I is the default, right? And that is...roll 4d6 six times, drop lowest on each, and assign as desired. Exactly the same as in Fifth Edition! Though Race modifiers are both more limited and more punishing in 1e than in 5e (where you can break the stat array at char creation, pretty much), the general effect is the same: solid characters with 1-2 15+ Abilities, per Gary’s comment in the PHB.
 
This is not how 2^nd Edition does things. Cook gives 5 generation methods. Method I is 3d6 in order, and Method III is 3d6 but arranged to taste. Method II and Method IV are 3d6 twice, keep best – so like the multiple stat Methods in 1e. Method VI is a weird combo of point array and dice. Method V is 4d6 drop lowest arrange as desired – the same as 1e Method I. Cook explains (2e DMG p10) that this creates a more heroic breed of adventurer; not ordinary schlubs thrust to greatness, but heroes.
 
Cook offers a three paragraph critique of this Method in the “Disadvantages” section, easily the longest. He highlights the risk of “super characters” which become difficult to challenge, and suggests that “ability inflation” will make high scores less exciting for players.
 
You’d think, from reputation, that this would be the other way around: Gygax encouraging gritty low-power play and 2e blowing this up in favour of superheroism.
 
(This isn’t even the only place this inversion happens: 2e Thief skills start lower, Weapon Specialisation can actually disadvantage Fighters compared to “Weapon To Hit Vs AC”, etc. Some of the changes are simplifications – like Specialisation – but some are also pretty obviously about reducing the power curve, as with the Ability Score generation discussion.)
 
What are Gygax, Holmes, Moldvay, Mentzer, and Cook saying about characters via their organisation of material and commentary on it? (Well, the Basic authors don’t comment on it, but they recapitulate in principle.)
 
Gary, it seems, had concluded by 1979 that the “real” D&D could be best enjoyed by having tough and capable characters from the offset. Cook, on the other hand – with the benefit of having played through the 80s, through the UA/OA era (and he wrote OA!) – preferred 3d6. He implicitly prefers the “man down the street” adventurer, as he puts it. 2e is not designed, at this point, to be a superhero game – it’s designed to be a retroclone! It’s designed to endanger the characters. Or at least, those are some very basic presuppositions exposed by the information design.
 
Why, then, do we think of 1e as lethal and gritty and 2e as superheroic? All kinds of reasons, really. At some level it’s just about product ranges: 2e went heavy on splats and settings and dramatic story concepts for its adventures. 1e was narrower. 2e also cleaned up its PR – no more devils and demons and half-orcs and assassins, and the PHB cover is brave knights, not thieves robbing a temple. Stylistically 2e offered a different image.
 
I think the actual adventures matter, too. I mean, from like 1989 to 1994 or so, the only really good adventure designer TSR had was probably Ed Greenwood (at least, who was actively putting out adventures). But more than that, the basic style of adventure has mutated so far between 1979 and 1989: not just in the “big” things, where we slam Dragonlance, but in the basic building blocks. No grinning devil faces or giant frogs ruined player’s days in modules from, like, 1990. The assumptions about lethality and problem-solving had significantly morphed, even where the game was still much more lethal and difficult than 5e, say.
 
Which I guess leads me back to this comparison between Gygax and Cook: Gary, from shifting Ability Score generation to ramping up class power levels for the UA Barbarian and Cavalier and so forth, spent a lot of 1e working out how to make stronger and more heroic characters, in the midst of the lethal, gritty Edition; Zeb Cook, inaugurating the superheroic and smooth edition, wanted to rein characters in, to make the world dangerous and strange. The development arc of 2e vs 1e was in one way a continuity - *of Gary’s later vision*. But an alternative 2e is imaginable – one less run from the front office and more from Zeb Cook’s intuitions. Perhaps we see that 2e best explored, not in the harsh world of Dark Sun (start at 3^rd level, it’s so harsh!!! Oh wait we barely support actual play in this setting), but in the magnificent body of adventure work by Bruce Cordell in the second half of 1990s. There’s a game there to be reclaimed: polished, loads of extra modular optional rules, loads of ideas, loads of material, but building on the earlier traditions of D&D.