Before getting into things, let me be clear that I think there's a pretty broad middle ground and that the "sweet spot" depends a lot of factors including, but not limited to GM skill, player skill, competitiveness, play goals (where they fall on the GNS scheme).
OTOH, rules light games, while significantly easier in the sense that you can learn the game in a much shorter period of time, are often much, much more difficult to run at the table and the experience will vary much more because of the DM.
IOW, a more extensive ruleset, run by an average DM will give a better result than a less extensive ruleset run by the same DM. The less extensive ruleset doesn't allow the DM to rely on the system, and makes it that much easier for the DM to make mistakes.
I'd argue that the Rules Light games are less DM friendly than the rules heavy ones. Sure, you can learn the game faster, but, because so much of the game relies on you, the DM, to make it run at the table, rules light games force DM's to play amateur game designer at the same time as they are trying to run the game.
I could not disagree with this more. Over 35 years of gaming, I've done various degrees of rules-light and rules-heavy, with the highest number of gaming hours logged between all D&D/AD&D editions (except 4E), Hero System, Storyteller/oWoD, and Shadowrun. At various times, I've bemoaned the ailments of "imprecise" or "vague" systems and at others, I've cursed rules-lawyering players who wouldn't let things be. One round of Phoenix Command, at a con, cured me of any interest in truly "realistic" game rules. I understand the lure, but it isn't what draws me to game.
There's a minimum amount of rules for "completeness", but most games can fit their core rules onto a single page -- or less. For D&D 5E, it's "determine your bonus (stat + optional proficiency + misc mods), roll a d20 and add bonuses, compare against the DC (AC is a special case of DC), if you roll equal to or greater than the DC then you succeed/hit/whatever. Some effects are not binary and typically use another die to determine how well an action succeeds (damage die). The rest of the page can be used for showing the DC chart, how to calculate AC, opposed checks, etc. You could summarize the character creation and advancement rules onto another page, maybe two. While a handful of additional rules exist, the vast majority of everything else in the PHB, DMG, and MM represents either guidance for applying those core rules, exceptions to the core rules and general guidance, lists of powers that open new things that can be done with the core mechanics, and/or reference tables to find bonuses.
That's not to say there's no value in anything beyond the most core, say, 32 pages of rules -- one of the things that keeps me with D&D is the extensive lists of monsters and spells, both of which could be considered "application". The point I'm really making is that there are a lot of pages devoted to stuff that a "good DM" should be able to wing, IMO. It doesn't
really matter whether attacking through cover grants a +2 to AC or disadvantage on the attack; the math is pretty similar for most DCs. It actually doesn't even make much difference whether it's consistent between attempts, other than the basic appearance of whether the GM knows what he's doing. Likewise, the long-term implications of a +2 situational bonus vs. a +3 situational bonus are pretty minimal, unless your group has really selective luck and misses by 1 a lot.
IME, when there are a lot of rules applications provided, there's a tendency by all involved to ensure they're following the rules as closely as they can, even if it means slowing down play when the details aren't going to change the result. In AD&D (and other games of that era), we rotated GMs with some regularity -- there were generally two primary GMs in any group I was part of, with others occasionally taking turns. Each felt free to tweak rules applications as he saw fit, based on story flow and setting lore. Yes, there were the occasional rules arguments or attempts at rules lawyering. But, the GM felt free to say, "suck it". Those GMs that made generally good rulings and kept the game moving in a fun and engaging way had standing groups. Those that failed to entertain weren't as in demand and would either improve or disengage.
Clearly, a new GM won't have the experience to make fair rulings. That's part of the process of becoming a good GM. So, having sufficient rules applications help a GM to understand how to apply the rules to situations. Too many applications can overwhelm the newer GM, though. Too few, and the GM has to learn too much by sink-or-swim. The truly newbie GM can only handle so much, but their ability (and need, typically) to digest additional rules applications increases quickly. At a certain point, however, a GM will gain sufficient comfort with the basic rules and the particulars of applying them to various situations that there is not really a need to have as much spelled out. At this point, additional complexity in the rules are either ignored by the group or provide fodder for arguments, depending on the group. So, I'd say low-moderate complexity benefits the novice GM. Moderate-to-high complexity benefits a proficient GM. But, low-moderate or even low complexity benefits an expert GM.
But... Players are the other side of the discussion. For the first part of the curve, the same thoughts apply: as much complexity as they can handle to gain an understanding of how core rules can be applied. What about the expert player, though? Do they benefit from a return to lower complexity rules? Well, it depends.
If the group is meeting for the "game" factor, then increased rules complexity means more opportunity for game mastery. Personally, I
love highly complex strategy games; the more moving parts and obscure rules, the better, and I get a huge thrill from using strategies that bring together subtle rules interactions in surprising ways. The gamist RPG player will, of course, be more inclined to enjoy heavy rules and good for them.
If the group is really about the story and focuses on things like pacing, character depth, etc. then too many rules will just get in the way. Once you have enough tools to do what you need, the rest is clutter. In fact, a group that gets together primarily to tell interactive stories may never need to get to the moderate-to-high level of complexity.
My personal belief is that the middle road is going to be most applicable to most groups: neither too complex/gamist nor too simple/narrativist. Only very experienced groups (not just GMs) will benefit from either very complex or very simple systems. Those that do are likely to only benefit from one or the other. They are both a niche of a niche. The extremely light systems have the advantage of being cheap to pick up and run for one-shots or other special case games. High complexity systems demand system mastery that only comes from focus and dedication.
D&D 5E is probably about as complex as I'd care to see a mainstream (that being a very relative term) system to go. In fact, it's a bit too complex in some areas, for my taste. I'll openly acknowledge, though, that I would consider myself to be a more than proficient GM and I favor the narrative aspect of RPGs, preferring to keep system mastery in the realm of board games. That's just what I want out of an RPG.
