I disagree with this to an extent.
I use QB Acc at 11 for me and 21 for CPU. I see ratings still mattering. Every QB does not suck. I use Pryor, I use a better QB - I see and feel the difference. I don't think I'm "lowering" their ratings. I'm challenging the accuracy skill and increasing the scope/impact of small mistakes. The CPU can still make plays on lower settings. I use the same type of sliders for both user and CPU and guys are losing, having good CPU players make plays on them both offense and defense, tight games, varied game flow, etc.
I throw on the run with Terrelle Pryor - and he's like a poor man's RG3. He can do it...within reason. I wouldn't take him and his 75 TOR and 75s Acc and try to throw it deep on the run (actually, I wouldn't do that with RG3 either - I'd buy time, let him set his feet and throw, otherwise, I'd just keep running...and that's what RG3 seems to do in real life).
I used RG3 a couple times (enough to know I want him in my Oakland CC *sigh*) and didn't have much issue throwing on the run with my set. The passes were "close enough" and if I make the right read on the run, threaten the LOS and make defenders run to me, I can find space. Did it with Pryor, pushed the LOS, the safety came up to challenge me, and I zipped it by him, that step or two making the difference (his ballhawk move failed).
Tried it in the Super Bowl, but Carlos Rogers wasn't having it - he made the play.
To me,
Lower sliders = stressing that skill more. Low QBA will stress/test a QB's accuracy ratings more. You may need higher acc to complete tougher throws - or use the L-stick, or have to keep that QBs feet firmly on the ground, or just not throw that pass, while a better S/M/DAC QB could make that throw off his back foot or while on the run, or fit it into the tight window. Lower sliders seem better at exposing who sucks at something because these players are already questionable, then you put stress on that skill with the lower slider, and that player will break more. 75 vs 65 is still a difference, but both are bad enough to where it's a difference between serviceable (75), and waiver fodder (65) - it becomes a "lesser of two evils" and you'll want to upgrade it. I find that realistic.
Higher sliders = highlighting top talent more. On higher sliders, elite players will be VERY elite because they'll rarely fail at whatever. However, a 90 vs 95 will still create a sizable mismatch because failures are dramatized more and when 95 whatever executes, it's going to be big regardless of slider set. Average players will suffer more vs top players at higher slider settings. Elites will require someone equally elite to stop them. Higher sliders let me get away with a 75 Acc QB because his mistakes aren't that big. To me, this should be reserved for the higher up guys.
It's just how you see the game.
I see the NFL as where any given week, an elite player can falter, most of them aren't perfectly executing every time, but even the mistakes are still "good enough" to beat average players. To me, lower sliders bring that out, especially in the more technique areas. I see the NFL as a place where average players can have a great day even against their "betters" because they had a good breakfast, slept at a Holiday Inn last night, whatever. The vagaries of human performance when tasked with doing a precise set of actions in a quick amount of time 60-70 times.
I think lower sliders capture that more.
By the definition in the post, I guess I go no higher than "mid range" sliders, and that's mostly for tackling - an area where there is not as much performance variability as some think, at least according to data I can find. It's an area where 80% success is BAD. For skills like that, I find midrange is about as low as I can go to be able to "accept" the game playing out that skill "correctly".
I guess, ultimately, my thinking is:
If a skill is defined by small differences - I go with mid range. 50-60 area.
If a skill is very important/defining - I go with very low - 0-20 area. I want the players who make mistakes in defining skills to screw it up more than the ones who tend not to flub things so badly.
If a skill has a low chance of success - I go with low/very low, no higher than about 25. INTs is an example of this. The WORST interception rate against an offense is 4.4% (KC). So the Chiefs throw 96% of their passes without a pick - and that's the WORST rate.
If a skill has a high chance of success - I go mid range to possibly to the edge of high range. Tackling falls under this as does WR catching. These are areas where 20% failure is about as bad as any starter-level NFL player gets. Below that should stand out like a sore thumb. Top end should also stand out, like Willis' 98% success in tackling last year. You should see that, too.
Simplifying it to higher vs lower also removes the impact of interactions. I could probably make 75 QBA work with enough interaction elsewhere - like going with maybe 15 WR Catching. But then I would feel like I'm "fighting one unrealism with another" instead of getting both "semi-realistic" with production coming out plausibly.
There's also the factors of speed threshold, game speed, penalty sliders, user ability, user scheme, personnel interactions, fatigue, injuries, DPP, etc.