In so doing, you get into "ivory tower" discussion, divorced from real-world practice. I agree that it is simpler, and sometimes clarifies things, but it is also the "spherical cow in a vacuum" of gaming.
Wait, looking at the systems as designed is "ivory tower" while whatever set of individual and idiosyncratic changes are made at any given table is "real world." This, to me, is the pit trap of starting from the position that RPGs are meant to be homebrewed -- it discards the idea that the base game is actually a real thing and that you must change it for it to be complete. While this is true of some games -- 5e in particular is an incomplete design because it bakes in "insert your GM here" as a large part of the glue in the system, but games like B/X are very complete. If you run B/X as the book says, there will be few areas you need to adapt and the rules will absolutely work as designed. This concept that you're supposed to not like the base game and change it until you do is, to me, a crutch to avoid actually looking at how games work. This argument continues into claiming looking at games as designed is "ivory tower," an anti-intellectual dog whistle statement, while games as changed are "real world," despite there being no rhyme or reason to changes other than personal preference.
Shouldn't changes be rooted in an full understanding of the system as designed, rather then the approach that says the system as design is a mythical construct to be ignored in any discussion?
I raise a point of complexity, and it becomes "stonewall any discussion..." Adversarial DMs were and are a pretty common experience - Gygax's own writings imply adversarial relationships at times. You want to dismiss that?
That's because this is what you're doing. And saying "complexity" is the same as the gamut of idiosyncratic changes made at individual tables is quite a leap as well. What you're proposing here is that there's no definable baseline because of these changes, that the game as played is some creation of the game as designed and all of the possible changes that people might make to it. This is trying to shut down discussion, because it introduces the easy argument for any point that people can and probably have changed the system so that point is now moot.
I think the point is that humans cultivate things. Talking about failure modes and flawed play is just as important to cultivating actual experiences at tables.
Sure it is, but not to talking about how systems as designed work. These points are great for talking about how people can fail, and ways to avoid that failure, which, oddly, is often using the system as designed rather than with the changes they've made. Or selecting a system that does what they want to start with. What you're introducing here is that people can fail (trivial and banal point), therefore those failures moot any discussion of performance as designed. If someone can fail, then the system is a failure, essentially. This isn't at all true, because people quite often do operate these systems without failing. Discussion of how the system works without failure modes is very valuable, and shouldn't be discounted because people can fail at it. This is like saying all pizza should just be considered burnt because people can fail at making pizza and burn it. If I want to talk about the best ways to make pizza, I'm not looking at cases where people burnt their pizza -- that's a clear failure at making pizza, and I'm talking about how to do it well.
