FormerlyHemlock
Hero
Yes, it does. Perhaps I haven't made clear why it follows, but it follows.
You've agreed that the characters that are played are the ones that tend to be above average.
The question becomes what defines 'average'. I'm arguing that average is defined by the table expectation of what a 'good character' looks like, which is going to be circularly defined by being 'above average', thus creating a feedback loop.
Suppose you are playing 3d6 straight up, then average is above an average score of 10.5. A character then with mostly 12's and 13's is an above average character, and players at that table will then have the expectation that a player ought to play that character rather than throwing it away or rerolling it. Suppose that group however tires of the randomness in 3d6 straight up, and decides to play 4d6 drop the lowest. This initially works better. There are fewer players with unplayable characters, fewer requests for do overs, and so forth. But you'll still have the problem that one guy in the group got the character with the equivalent of like 46 point buy, and you are playing a character with the equivalent of 15 point buy - perfectly normal results for something as random as 4d6 drop the lowest. As more and more players begin to find playing with higher stat characters normal, the ones that are left out of the goodness feel worse and worse about their substandard characters - even if a year or two before they would have been mostly content. That's because a year or two before, a character with no 16's was normal. But now, most players have the expectation of a higher stat array.
So what tends to happen is when a player rolls up a new character, if his scores are now mostly 13s or less, the rest of the group (and the DM in particular) takes pity on him and says, "Dude, that's just bad luck. Have a do over." And the more emergency do over's you have to call for, the more the group starts thinking, "Gee, this 4d6 drop the lowest method isn't working out." Because whenever you allow do over's in your preferred gambling method, it tends to create cognitive dissonance. The method no longer feels as fair, and in particular I would argue that do overs are attacking the illusionism that makes dice rolling fun for everyone. So they come up with a new rule that they agree is fair and everyone should live by, like say "4d6, drop the lowest, but reroll the ones." And then, the process repeats itself, because the new method generates higher averages and with it higher expectations about what a good character actually is.
Looking back, I see that process either playing out or had already played out and reached some sort of extreme that produced such high results that, along with a bit of judicious cheating, no rerolls were ever asked for in just about every group I was involved with that used dice rolling.
Let's break this argument down to its abstract logical form.
As I perceive your most recent argument, you're arguing that A => B. (Player jealousy leads to played characters being above average.) A & B => C. (Escalating spirals.)
I acknowledge that B. Not because A, but because Darwin.
!A & B !=> C.
In the post that I took exception with, you simply stated that B => C. That doesn't hold without your hidden assumption, A, which I don't share.
I've been away from (A)D&D for a good couple of decades until recently, so I'm not contradicting your observation that A (player jealousy drives escalating stats) at your table. I haven't seen any sign of A so far though, nor is it plausible to me that you can't hold very firmly to a baseline by grounding the game in the reality of what the stats actually mean. In short, I don't believe that it is in any way impossible to impress upon players that a 13 Str is rather high compared to most people, e.g. by requiring them to generate five NPCs on 3d6-in-order (donated to the DM for future usage) for every new PC generated on 4d6-drop-lowest. You're asserting an impossibility which looks prima facie non-impossible to me, and which does not seem to be impossible in practice.
A year in to 5E, and my players are still rolling 4d6 drop lowest exactly per PHB (sometimes in front of me, sometimes between sessions--I don't care), and some of them are using point buy, and the stat distributions look mathematically plausible with no evidence of cheating (I wouldn't tolerate cheaters anyway), and about half the rolled 18s that I know about were rolled in front of me. I see no evidence of creeping "ever-more-elaborate dice rituals" that drive averages continually upwards, as you've hypothesized. But I have observed that characters which are played for months at a time usually appear to be in the top 50% of what you'd expect from 4d6 drop lowest, and that neither surprises nor disturbs me.
If I roll six times on BrockJones.com and get:
9 15 9 17 8 12
17 10 9 10 14 11
13 15 14 16 7 13
9 13 13 14 15 10
15 12 9 9 16 14
10 13 10 12 7 11
and then I realize that row #6 doesn't look like most PCs that have been played at my table (he does look like two PCs, one of whom didn't get much screen time), that doesn't bother me.