Assuming a bog standard level 1 greatsword attack (+5 attack, 2d6+3 damage), normal attacking is superior when AC >= 20,
AC 19 is equal, -2/+4 is better between ACs 18 and 14, and -5/+10 is better when AC <= 13.
Power attack in forms that allowed chosen minuses to hit and comparable gains in damage have been around for a while in various types of d20 gsmes snd others for a long time.
Experience with them backs up your claims in actual play non-ehite room excels where the right assumptions get you amy results you want.
Those who typically evangelize the 5 -10 round here fall back when pressed on how you just eont use it when it's not superior and then ignore that part of the sample.
But, in actual play, a variable -×+2× will mean a lot more opportunities for the feat to give you extra. It eont be the same yield, but then, the -1 to -4 gains will be all coming from cases which would have been-0+0 so anything gained is more to the average. Its all gain, no downside if applies eith the same perfect choosing.
But, back to those pesky assumptions, to me it's a bit bigger than it likely seems to some. I never count either random distributions or such. Too often the math pretends to offset high results with low ones and rarely does that play out.
For my analysis, instead of random ACs or ACs distributed by entries in MM, I tend to pay more attention to "that matters". The idea is that casual warm- ups and skirmishes are gonna be over easy anyway.
The fights "that matter" are the hard to deadly ones. Those, IMO, do not show anything like an "even spread" of ACs or a bell curve of defenses. The fights that matter and are hard to deadly tend to combine higher ends of both defenses and offenses.
So, anything that weights the outputs against say AC 10-12 in anything even remotely the same zip code as the results against AC 18-20 is gonna be a product of that assumption.
Winning the hard fights that matter counts for a lot more than squashing the trivial fights that dont matter in 2 rounds of never-ending cantrips and swings rather than 3 rounds.
The -×+2× flexibility is gonna apply in more of those "hard that mwtter" than the locked in -5/+10 do.