Yes, it can be wrong in the context of a single game (although I'd assume that more often than not it still reflects accurately who had the better quantity of chances * quality of chances) but that's not what I meant.
The discussion about xG is usually between two camps: One who see football as a simple game (goals count, everything else doesn't matter) and those who see it as a complicated game. IMO it is the latter. Football's low scoring nature and the fact that goals happen so rarely means that results are prone to a lot of randomness. There are probably very few sports in which the chances of the better team losing are as high as in football. Those who really understand the sport understand this nature and get that the best you can do is maximizing your chances of winning, you can't force it. So when you review yourself and see what you can do better, it is important to keep that in mind and not think result oriented - as in a shot that doesn't go in was a bad decision/execution and one that goes in was a good one. Or a successful dribbling that leads to no goal was pointless.
So to speak, xG is a proxy war between these two understandings. And I'm sure Sumptner would agree with me that it's not a simple but a complicated sport with lots of luck involved - and if he doesn't, he doesn't understand the sport either