Also, I'm not certain that glove is reflected in the 1B's own stats, but instead in reduced throwing errors for the infielders. So I'm not convinced that pjf's example tells me that glove does not matter at 1B, although it might be range that reduces infield throwing errors.
I've considered the possibility of this before, but there is really no evidence that fielding percentage for the other infielders is based off anything but their own intrinsic glove + arm acc. I think it's just a glitch that 1B commit the wrong number of errors
It's probably debatable how much more plus and less minus plays correlates to actual winning
That's a preposterous statement. Without getting into another novela about it, the summary is that plus plays skew batting average towards the pitcher and minus plays skew batting average towards the hitter. Plus plays convert to fewer baserunners, which convert to fewer runs against, which convert to better Pythagorean expectation, which convert to actual wins in the standings. The opposite is true for Minus plays- they convert to more baserunners against, which convert to more runs against, worse Pythagorean, more losses. It is a fundamental principle and not debatable, I'm not sure what exactly prompted your statement or what has you skeptical