D&D still has its market in the face of competition from incredibly rich game experiences. The gulf between video games and TTRPG's, and video-games with VR and TTRPG's, is not so huge. VR dramatically increases immersion, but until software can generate realistic human interaction on the fly, the table-top will continue to have its place.
When a game really does generate a realistic NPC, with all of the humour and weirdness and complexity of people I've known all my life, the world will have changed so radically that even the phrase "table-top" will seem quaintly prehistoric.
Oh yes, definitely, PC to PC and PC to NPC, NPC-NPC interaction are very important aspects of D&D which are hard to do in a game. Although using ventrillo or the like is the obvious PC-PC solution, and if you have a DM who can assume the NPC voices then that handles that.
But they already have bots which pass the Turing test now, today. And it should be possible to make an Orc interact with you better than some DM snarls and grunts that you often get. And monsters don't need to speak, or at least often don't. They do need to think though, and that's harder to pull off right. AI is getting there, slowly. And like in Neverwinter when a DM can assume full control over an NPC or monster that should be a decent stopgap solution.
The real value of a DM then becomes more in the world-creation, and NPC high-level planning and strategizing and motivating. And you don't need D&D rules for that, in fact they are a distraction towards a DM's true purpose. All the nuts and bolts of the combat engine could be hidden away behind a UI and some sliders to control difficulty.
Since there's no reason a VR game can't have virtual DMs or with players playing the villains or monsters if they want to, VR can pick up where MMOs left off and run with it. In a sense, Wow has already stolen D&D's lunch money, several times over.
My point is, once you move away from the tabletop and turn-based rules migrate into real-time sims, especially full immersion ones with presence, there is no need for an actual D&D rulebook per se, only the IPs of the monsters and campaign settings.
In a game loop, you can have a physically accurate and complete model of swordplay and environment interactions for PCs and critters, while mixing in magic which breaks those laws of nature, and the hand-of-god DM who is pulling some strings here and there where it suits him (or her). I don't see why the average person even needs to know how each game engine works under the hood, or even if they could properly comprehend a real-time simulation even if they wanted to. It's best to keep these design focii separate. At the table top, a game system lives and dies by its rules, and being rules-light while still powerful, flexible, and fun is a distinct advantage (5th ed vs 3rd/4th is a good example). 3rd and 4th were way more complicated than they needed to be to make a workable D&D game, and often got in the way of casual gameplay for the majority of gamers who are not crunch-fanatics unlike most of us on these boards. But in a computer simulation, there's no reason you couldn't have a skill system that reflects actual skill use (like Everquest, for example), and remove a lot of the nuisance from managing those crunchy subsystems. I hated assigning skills in 3rd ed, for example. And 4th was just : pick arcana and/or perception and/or stealth, and you're done, basically. The rest were useless. And skill challenges were a pathetic mini-game of multiple dice rolling that everybody hated. Both skill systems sucked, in different ways. 3rd edition was way too customizable and tedious, 4th edition went too far the other way and made it absurdly simple. 5th edition is slightly better, but a sim could do it far, far better, from both a usability point of view, and a crunch point of view.
A lot has been said about Wizard's inability to manage Digital tools endeavors, even getting a character creator seems beyond their reach (and for such a simple game too, for shame). I'm not optimistic about Wizards either identifying what their core property is in a purely digital context, in an era of massive virtual worlds and incredible technology. But they might. They did properly identify what people wanted out of a tabletop D&D ruleset. By asking people. Lots of people. And trying different things until they could boil it down. Simple rules. Fast combat. Simple fighters that just attack at the same table as complex maneuvers. Simple blaster casters at the same table as complex ones that do amazing things but require some thought. A challenge for PCs to survive. Flexible, smooth, versatile combat system with few moving parts and much less immersion-breaking game jargon to learn and use. Faster character creation.
But in a VR game, you can have both "make a quick character" and customize in the same character creation step, or at each level up. You could have fully randomly generated characters, worlds, NPCs, even adventures. With or without a DM. Am I regurgitating Neverwinter Night's press release? Yes! But that's an isomorphic, turn-based adaptation of D&D rules which are still a far cry from what the legacy of D&D could evolve into. The idea of an imaginary world run like a simulation, created by a DM with other people running around killing monsters and taking their stuff, is golden. Now Wizards needs to recognize what the fundamental idea of D&D is, and exploit it in a non-turn-based, non-human-run-simulation, context. Of course the DM should run the overall game, but not at the level of deciding if each sword thrust lands. That should be entirely abstracted away from players. You don't want to see the code running the Matrix, that's the whole point of full immersion. D&D should embrace full immersion, it paid off with 5th edition and it would pay off in VR. Cut away the legacy turn-based rules, and identify what is the real magic behind the idea of D&D. A fantasy world, with an independently run reality (a sim, if you will), yet with human interaction built in so they can use it as a playground. And that fundamentally intersects with VR. You could even say VR borrows a lot from D&D, except does it with computers instead of human imagination directly powering the imagery.
It's also a false dichotomy I think to say, a graphical representation can't be as good as a book or one's imagination. With better / easier content generation tools, that line will also blur until it disappears entirely (brain-computer interfaces).
