I don't think it should be decided by the furthest edge of a line as it's impossible for an attacker to know if they are onside by a gap as marginal as 1cm. If they knew that it was the width of a ball, or something like 20 to 30cm, then they could play to that.
Whichever way we look at it, the problem is always the degree of variability of decisions being open to interpretation.
Cuases of this can include the fundamental definitions of what constitutes on/off/where/when lines should be set etc.
If we're going to go for as accurate as possible, then it almost doesn't matter what you choose to measure from; as long as it is possible to set a rule/measure that can be consistently calibrated, you're heading in the right direction.
My feeling is that the shirt sleeve rule is too variable as some shirts may be adjusted,some players have longer arms, some wear long sleeves. There is no definitive point for everyone.
And I'm quite OK with players not being able to tell to within 1cm/whatever - that's part of the skill of playing the game - their judgment. And of course, if things could be know to be reliably exact, then we shouldn't have any qualms with the truth.
There is of course, a way to ensure you are onside. (Check). But to push the limits, as in any competition, there is an event of risk required , and sometimes you'll get it wrong.
VAR probably has improved the game from a technical correctness point of view, and will occasionally suffer from issues that either can't be addressed perfectly, or are part of the learning curve.
But in getting more precise, it highlights the areas where there is a lack of consistency in its own processes. And when that inconsistency is down to the humans who implemented it not having thought things through well enough (perhaps deliberately so), then that's what grinds gears.
The huge human reaction to perceived unfairness.