I honestly believe most of the Low Oxygen stuff is legit. I have worked some of it into my process, but I have too many other limitations to adopt it and test fully. For some the change in process may not be worth it, but that doesn't mean it's bunk.
Is there empirical evidence that showing that it doesn't matter at the homebrew scale? Sure lots of preferential and anecdotal evidence, but I haven't seen numbers against it.