Why would I? They're completely different cases unrelated to one another.
There's a marked difference between having one particularly strong stat, and having high stats across the board. There's also a marked difference between clicking the re-roll button two or three times until you see something acceptable, versus sitting at that roller for hours, even days, to get the 'ideal' roll. Conflating the two just makes it look a bit like you don't understand the discussion.
Would it interest you to know that the average expected roll spread of rolling 4d6 drop lowest six times is: "16, 14, 13, 12, 10, 9" Which is, in reality, very close to the standard array. If you need me to provide the maths, it can be arranged, but I'd urge you to trust that I don't provide false information.
If you think it's not a counter argument to note the extreme waste of time taken to roll even three 17-18 stats in one roll set (a chance so small that measuring the percentage it to three decimal places still registers as zero percent), Rag, then by all means, go and roll a six 18s stat spread with normal rolling. Come back when you have done this, and then tell us how long it took you to get that extremely unbalanced character. We can review, and let others decide if that balance of time is worth ignoring as a consideration. If it's something that someone might feasibly do to start a character in a game, then surely it's no real stretch for you to do it to make your own case in this discussion stronger, no? Go ahead... Assuming you have a REALLY quick hand and eye, and it only takes you one second to roll, check, record and move on per roll, and that you never eat or sleep or take any breaks... I'll see you in circa fifteen years.
I argue in favour of a direct point setting character generation for a completely different reason: The creation of older existing characters that individuals might want to remake, or give new life to in this game.
No, that's why modders wrote an autoroller for that in BG2.
I'm puzzled why a modder would ever do that.... if you're going to do that, then surely it would be faster and more efficient just to write a mod that lets you set your starting scores?
