I've gotten out my rulebooks for Advanced Dungeons & Dragons, 1st Edition recently, as I wanted to see just how much of the game I played 30+ years is actually what was in the rules. If I found omissions or discrepancies, then I wanted to see what the as-written rule actually did vs. what's been said.
First: no, what I played was (as was commonly done) some mashup of Basic and AD&D. We had fun, but we also contributed to the conditions that would result in D&D 3.X and its successors. We were punk kids with no wargaming background and no connection to the Twin Cities-Lake Geneva scene. We thought we were the shit, and when we weren't playing Car Wars we blundered our way through mega-dungeons that one of us designed on spare sheets of graph paper (where it was eight squares to the inch, not four) meant for use on a drafting table.
Second: While the Dungeon Master's Guide discouraged it formally, the rest of the game did not work properly if PCs were not rolled using strict 3d6 in order. Critical hits were not baseline, and THAC0 was a mistake; the matrices worked as intended when actually applied properly. Paladins, Druids, Monks, and especially Bards were meant to be rare (and Bards a reward for executing a well-planned long-term goal) and that didn't happen without said rolling method. Level limits actually have teeth when used, and the training rules by themselves reigned in bad behavior by players of many sorts.
Third: Gygax needed a better editor. I thought I had issues with phrasing for reader comprehension, and once I parsed what he wrote I got mad at how needlessly obtuse and complex his technical writing was. It's no surprise why many AD&D1e fans prefer to use OSRIC when running that edition of D&D. That said, when seeing how the machine operates his infamous rant about time-keeping makes perfect sense.
This is going somewhere.
While this D&D edition is the first to commit the sin of being books and not a boxed set, it is also the edition that many gamers would say is where the practical limit of ruleset complexity (and thus intellectual density) got discovered. Player-facing rules complexity is significant, but still not so heavy that it's a turnoff; the Game Master shoulders the bulk of the burden, one he can increase (and benefit players) simply by not using official character sheets. A pencil and a single-subject notebook is still more than enough.
In play, players can still operate on natural language and the fundamental feedback loop of "What Do You Do?", leaving the rules to the GM. This does mean that the GM has to spend time with the tomes, doing homework until he masters the rules. So long as the GM isn't the punk-assed power-tripper too many of us were when we were kids (i.e. be a normal adult), this is fine and a lot of the later editions' measures to limit GMs is not necessary at all. Mech Piloting is not a viable strategy by default; trying to do that doesn't give the full reward that later editions do.
While Greyhawk is mentioned, the rules don't specify a setting. They give you enough to imply one, but that is only implication. Instead, the Game Master gets told multiple times that setting decisions are his to make, and that he should bring those decisions forth to players when that information matters to their decision-making (and not just during play; if Dwarves are not in the campaign, players need to know at character generation. The tools to create verisimilitude are present, but it is on the GM to make it so.
If the Original and Basic D&D editions erred on too little substance to make use of liminality, AD&D's 1st edition erred on too much. (As we've seen, you can code AD&D into a videogame- as several examples show.) However, contrary to informed expectation, the rules of AD&D1e are not so inter-dependent that they cannot be changed at all without disastrous consequence.
It is by no means a perfect example, but it is an example of practical liminality that isn't destroyed. Damaged? Maybe. Destroyed? No. That came later.