Lyxen
Great Old One
I am well aware that the PHB advises players which abilities to put their highest scores into. According to what you have said previously, doing so is not required: a player following that advice is power gaming seeing as it will make their character stronger in play. Or can you explain exactly what the difference is between 'building a reasonably efficient character and not gimping yourself' and power gaming? Is it a 14 instead of a 16? On what grounds do you argue that, given that a race + standard array is very likely to result in a 16* (inevitable for standard human, nearly certain for a half-elf, or must the latter avoid following the PHB advice when it comes to their floating +1s?)
And once more, you seem determined to find a fault there. I have explained to you multiple times what the difference is. And as for the race choice after rolling the dice, first it's easy to fix by asking people to choose their race/class first, and second, this is why humans are still viable for those who really want higher stats.
It is a strawman to introduce here 'asking the DM to enable options': we are discussing what should be the future core rules, not what a player might beg from their DM.
I didn't think we were, it's not the title of the thread, and I usually don't participate in threads about the future editions because it's so uncertain that it's utterly pointless and degenerates in 5e bashing 99% of the time. For me, it's just about Floating ASIs as currently available in the game.
Still, it does seem like the kind of hedging you have committed yourself to once you start special pleading between 'reasonably efficient' and 'power gaming' while refusing to accept anyone liking their highest score in their primary ability - as advised by the PHB - might not be power gaming.
And have I not also said that there is a wide range of behaviours here ? sigh
It's hard to understand then why you are not aware of the extremes of optimisation that are only possible using the 4d6k3 system, and why most optimisers use standard array or point-buy as a fairer starting point. You're conscious of the Anydice evaluation of the 4d6k3 average array, relative to the standard array, right?
The fact that 4d6k3 usually gives slightly higher rolls has little to do with optimisation, unless you allow people to add their ASIs wherever they want, in which case it makes it WAY better (which is another reason for not allowing floating ASIs). Don't confuse the two, optimisers are frustrated by rolling stats because of all these odd numbers. And the reasons for the optimisers (the true ones, not the uncountable copycats all over the place) to use standard arrays is that it allows them to get exactly what they want where they want, and to compare results.