Since the average age of this forum appears to be 40, I know of no better place to answer the questions I raise in this post.
Genuinely, why were page counts so high in the 90’s? Why was there so much demand for more rules, more setting detail across the entire scene, before the OSR’s push toward minimalism?
A very funny read. Nice! Also: horrifying!!
I feel like most of this boils down to problems like “marketability” and “product design.” A higher page count lets you inflate the retail price, which gives you more (theoretical) budget to spend on making the book look like a more desirable product: glossy hardcover, colour pages, etc… Which in turn makes it more attractive to potential customers, and in some cases may even be necessary for getting your product into certain desirable retail spaces like the Waldenbooks RPG shelf.
Similarly, Mike Pondsmith had proven that the formula of “game mechanics wedded to a setting bible” was apparently a profitable strategy; I suspect a large part of it is that you can expand your market by selling to both RPG players and just general fans of the media property your game is about, and maybe manage to convert some of those people into RPGers while you’re at it. Game texts like Teenagers from Outer Space and Mekton are as much exigesis on the narrative conventions of Japanese animation as they are RPGs. (For what it’s worth, I think that this is actually a woefully under-examined part of the history of “fandom” as a kind of knowledge production, and points at how much the history of fandom and RPGs are intertwined in ways that not a lot of literature on either seems to want to dig into.)
That being said, the OSR certainly didn’t invent “minimalist” RPG publishing either. The indie RPG scene of the 2000s ushered in by accessible DTP options and driven by communities like The Forge was way ahead of the curve on that point. Before that, there just weren’t a lot of viable options for wide-scope publishing/distribution/marketing if you didn’t play by the rules set down by the big fish in the supply chain. The best you could really hope for was hawking zines advertised in the back of industry magazines and buying a small booth at a convention.
Are you saying the majority of early cyberpunk books included equally granular setting details?
“How did Cyberpunk get off the ground” is an interesting question I’d never considered.
I was thinking more about R. Talsorian’s anime games–the generic ones like Teenagers and Mekton as well as the licensed ones. I’ve only read the 2020 rulebook, and that was decades ago: it may have been riding a cyberpunk zeitgeist but I don’t really know how much the overall product line went the way of setting cruft over other kinds of tools for play.
Another prominent example from the 90s is FASA, who arguably perfected this form (essentially bootstrapping a fandom–not for the games, but for the setting/product itself–by selling details of a fictional universe) with Shadowrun and Battletech, and did it better than almost anybody else at the time.
I don’t see this premise working, to be honest. Looking at games all over the board, there’s still a large page count, and now with added color graphics to ramp up the previous non-existing coffee book value for those who don’t get to actually play 
One of the founding pillars of the OSR was OSRIC, a retro-clone of the ground zero of moar rules. Basically every B/X retroclone got AD&D classes and rules added in supplements. Even more additional classes, spells and monsters sell decently well.
The desire for rules and setting detail and our capacity to produce it hasn’t changed.
Tenchi Muyo is a decent enough example, as we see that reflected quite a lot these days. It basically deviated from the idea that you publish your generic core book and separate setting book, but instead you got all that, plus a “episode guide” in an vague attempt at getting some anime viewers into their first RPG (this is pre-internet). IIRC there were quite a few BESM variants doing that.
Pretty much the same situation you have these days with don’t-say-the-word-generic systems like 2d20 or the “Year Zero” games–or their grand-daddy Unisystem.
Nothing really changed.