A good spot up shooter doesn't necessarily have low off-the-dribble ratings. See Steph Curry. But I would really like them to split up ratings for shot types (assisted vs unassisted, off the dribble vs catch and shoot, wide open vs contested, etc) and use detailed data from STATS and SportsVu. I guess that's what all the sig skills (spot up shooter, corner man, dead eye, shot creator etc.) are supposed to do but I don't think they're applied with deep statistical analysis and there's no granularity with them. There's also the whole problem of raw FG% causing certain players to be overrated-- ie, the Kyle Korver (or Klay Thompson for that matter) being rated higher in 3-pointers than Steph Curry, when their shot difficulty and volume differ drastically.
The pick and roll maestro/roller sig skills are also done somewhat incorrectly. They shouldn't give magical shooting bonuses in those situations. Rather, for the PnR roller, there should be some minuses to catching, moving, and shooting immediately after setting a screen, offset by high hands, mobility, and good finishing ability in traffic for good PnR finishers-- or at least the sig skill should reduce those penalties.
Same goes for the PnR "maestro". There shouldn't be a simplistic magical shooting boost, but a good screen should allow someone with good off-the-dribble shooting or driving ability to take advantage of the screen. With the sig skill, the pass should have more accuracy, take a better angle, or have more zip with the sig skill (I think the first is currently true). Proper ratings probably also make the skill unnecessary.
The Currys of the League run the PnR well because they need to be hard hedged or doubled, and the Lins and Rubios of the league can run the PnR because of their dribble penetration. They all have in common the ability to time accurate passes, but the PnR doesn't work for them just because they get some magical shot bonus. In fact Lin and Rubio are mediocre and abysmal shooters, respectively.
While I like Sig Skills in differentiating player play styles, they are best used for things that are hard to quantify and are intangible. They've kind of turned into simplistic, binary shortcuts to get players to play a certain way without developing more nuanced ratings and getting them to work right. They're better than nothing but not always an ideal implementation.