I know Carmack was specifically talking about 3D visualization, but I was responding to the parent comment's point about the value of data simplification that 2D brings. I don't usually quote myself, but that's what I meant by:
I'm carefully wording this in terms of UI design in 3-D because I agree about the physical issues of eyes and focus.
I specifically brought up the car's knobs and wheels and buttons because it's an example of a fairly complex but widely adopted and successful non-2D interface. My point was that I don't think there is an inherent limitation in the human brain's ability to comprehend or think in 3D, and that I think it's more of a training and acclimatization phenomenon.
And I'm saying I think it's not actually the brain thinking in 3D. I think it's thinking in terms of body position and movement. Which is something that takes place in a 3D space, but isn't actually a generalized 3D comprehension.
One way to check would be to have people reproduce things they know via touch in visual contexts. E.g., can people draw their steering wheel? Given how their tongue knows their teeth, can they do a 3D model of that without looking at references? If we hand people a box where they can put their hand in but can't look, can they model the shape just as well as they could by looking at it?
For me at least, these are very different kinds of knowledge. If I'm, say, feeling around and working on the back of a server I can't see, actually looking at it is a very different experience.
I would wager that unless you were a practiced artist/designer, you would have trouble replicating a 2d UI from memory with much accuracy. Our brains don't work like that in 2D either. You could probably tell me that ABC is in one corner and XYZ is in another, but you probably couldn't tell me the relative font sizes and widths and heights of boxes and cards. Unless you're a frontend developer or a UI designer, in which case you're trained to see and remember those things.
There are plenty of examples of people who are able to fluidly navigate 3D systems. RC helicopter pilots are a great example. The helicopter's controls are relative to it, not you, and it takes a while to retrain your brain to think in those terms. But you can see plenty of videos of stunt pilots pulling off tricks that clearly require an excellent understanding of the controls, the mechanics, and the space around the helicopter.
Sculptors are also an excellent example. Even with a reference, I'd bet that it would take you a really long time to replicate an object with clay. However, I once watched an artist friend carve a block of butter into a Buddha over a one hour dinner. I think that would pass your test for having an effective 3D mental model.
Speaking not as any authority on the topic, I'm arguing that our brains aren't usually asked to deal with 3d problems, so we don't have developed mental muscles for it. Except for the people who do deal with 3D regularly.
And I'm saying I think you're overgeneralizing. I don't think there is general 3D capability to be developed. E.g., I doubt a sculptor (who takes years to learn to do it well) swap chairs with a helicopter pilot (ditto) and have them both do reasonably well.
I of course agree people can learn to develop all sorts of specific skills. But I don't think specific skills that operate in 3 dimensions are an argument for VR being particularly useful for a general audience.
I'm saying there is no general 2D capability either, so the existence of a general 3D capability (which I doubt it exists) is irrelevant to the question.
And that is also separate from whether VR is useful for a general audience. VR becomes useful for a general audience the same way a smartphone or a computer does: by covering enough usages that the vast majority of users will find useful applications. That's regardless of whether those applications use a 2D UI projected into the virtual environment or a skeuomorphic 3D UI.
I'm carefully wording this in terms of UI design in 3-D because I agree about the physical issues of eyes and focus.
I specifically brought up the car's knobs and wheels and buttons because it's an example of a fairly complex but widely adopted and successful non-2D interface. My point was that I don't think there is an inherent limitation in the human brain's ability to comprehend or think in 3D, and that I think it's more of a training and acclimatization phenomenon.