Think about it, though. If you can feed a separate video channel directly to each eye, through goggles for instance, you can simulate a 3D environment stereoscopically. Similarly, you can simulate audio. With ForceFeedback, you can simulate tactile sensation. If you're clever enough, you can simulate smell (say, with the IsmellDevice) or even taste.
Ok, slam enough computer power behind it and enough programming prowess and you can build a completely immersive false environment that you, or at least your "avatar" (cf. AvatarAgent), can wander around in as if it was real. Or even better, you can wander around an environment that doesn't conform to RealWorld "physics" as we know it at all. AliceInWonderland.
Yeah, cool, but what's the point? Well, if you realize that almost all computer technology is built from a metaphor to something in the real world, it's pretty straightforward to actually use the real world analogs as the interface to the computer. Make the machine as "simple" to deal with as RealLife.
Except that doesn't make sense. As the AntiMacInterface admirably points out, it's best to invent completely new interfaces to map onto new technology. The analogy never holds. The limitations of one medium don't hold in the other. The first tractor used reins to steer, but eventually replaced it with the wholly new steering wheel interface we have now.
Another example, consider existing software that attempts to make the computer into a town metaphor. It forces users to physically crawl across "town" to the post office to pick up their mail. Instead, it would be easier just to run your mail program.
Moreover, it's difficult to navigate a 3D environment with only the WimpInterface.
Or is it? Games have been doing this for quite some time now. Indeed, the biggest 3D VirtualReality simulations are games, networked in multiplayer modes. UltimaOnline? is perhaps the biggest, not even truly 3D.
Indeed, it can be said that many online communities are virtual realities, at least in the imagination-scape of their members. Consider LambdaMoo with its main "rooms." It's purely text mode. Now that's cool.
Are virtual realities ever going to exist? Probably. Are they going to be the dominant metaphor for interaction with computers? Probably not. -- SunirShah
VirtualReality is typically some form of ObjectiveSpace?; when that ObjectiveSpace? maps to the problem domain, that can be a good thing, such as medical imaging or architectural fly-throughs. However, it's a mistake to assume all online interaction is best performed within VirtualReality. Instead, consider the higher-order spaces: SemanticSpace and SocialSemanticSpace?.
One problem with 3D VirtualReality has been that many people will suffer "motion sickness" when the displayed motion doesn't match one's internal senses of motion. This is especially a problem with head-tracking headset displays, but it can also happen with 3D games (like Doom or Quake) on a regular monitor.
More generally, people can be very forgiving of high-level "abstract" irregularities, but low-level inconsistency (like head tracking/display lag) is often unacceptable.
New media always take much of their initial content from the old. Also, VR is interesting partly because humans have some good 3D spacial abilities. You get an immense amount of information thrown at you walking through a forest, but you don't suffer from overload. People are good at dealing with data if it is in the right form, which generally is the form we evolved for. VR is adapting computers to people instead of vice versa. A spacial postbox metaphor may be better even if it is somehow less efficient.
Hitherto, the technology hasn't been in place, so I don't think we can conclude much from the current take-up of VR devices. That said, there may be an inherent contradiction in wanting VR to be both immersive and unobtrusive. Things like HalfLife? contain deep "real-world" models but are not immersive in the sense of needing head-sets.
They are arguably more immersive than, eg, films, both technically and practically. When I watch a film like TheMatrix?, I am thinking that I have already done that stuff myself, playing computer games.
While computers are easier to change/adapt, I think the human side will also need to change in order to get the full benefits of computing. Much of ordinary life today is an "unnatural" adaptation to the social/technical structures of modern civilization. (It has often been said that main purpose of schools is to teach children to fit into an industrial society.)
Reading might be a good example of the kind of adaptations people can make. People don't naturally read in the same sense that they naturally learn a spoken language. Most young children can be taught to read, however. Another human change is touch-typing--it makes a huge difference in one's ability to express written ideas.
Computers are also making vast improvements. Continuous speech recgognition has made extraordinary improvements in the last few years. Modern speech recognition systems will use context and parts of speech to determine appropriate words. For instance, if one says "the weather", the program will determine that "whether" is not an appropriate choice for the second word. (One would never say "the whether". One might say "whether or not".)
The role of computers in most people's lives is not yet determined, however. Computers may become more "appliance"-like to most people, and require very few adaptations. There will probably always be some "hard-core" individuals who want the "best" interface, and a large number who want the "easy" interface. (To some people it is important to be able to "cp *.c *.h ../prjlib/src" in one quick step, while others would prefer the simplicity of drag-and-drop. Some people would prefer to say "Copy my current source files to the project source directory.")
Cars provide a good example of this kind of interface choice. Some people learn to drive a manual-transmission car, and gain more control in exchange for more complicated operation. Other people like the convenience of an automatic-transmission car, and will pay in the extra cost and loss of shifting control. Perhaps computers will also keep multiple interfaces for different kinds of users.
A recent SlashDot article, "Natural Language CLIs?" (http://slashdot.org/article.pl?sid=00/07/16/1713207&mode=nested) may also be of interest for more ideas on human/computer adaptation.
As I'm reading OtherlandCityOfGoldenShadow, I'm increasingly disappointed with the ability with TadWilliams' ability to make meaningful interfaces in 3D. Most of the novel focuses on false worlds, like RolePlayingGames and the ilk. Functional VirtualReality is limited to rooms that are really big (and an awful number of these), connected together by hallways. Most of the, "Hey, let's do some work!" is done out of 3D. On 2D windows that pop up or off to the side.
Even the MetaVerse is wrong. In the MetaVerse, companies actually try to buy pixel space on the strip. However, I seriously doubt sites will be located spatially relative to each other in the virtual space aside from the handful of VirtualMalls? (and even then?). It's not practical or natural.
Can anyone point to useful 3D interfaces? This candyland, theme park stuff is all nifty, but it isn't going to change the world, I think. -- SunirShah
On social MUDs the initial building is all in a one area--since there's nowhere else to build yet--so the oldest places are centralized. Eventually there are popular places elsewhere, especially if it's not easy or transparent to build in the central core. Since it's harder to instantly transport in the MetaVerse than social MUDs, central building should be more likely in the MetaVerse.
You should probably read Diaspora, by GregEgan, even if it doesn't answer that question. It is about sentient software; emulated humans. He doesn't so much have 3D interfaces as, well, places. He also has 2 languages called Gestalt and Linear. The former is 2D (what vision became) and the latter 1D (what sound became). For me, thinking is largely linear, like speach, and I could believe that humans will use linear systems for expressing symbolic logic for as long as they are human.
I disagree about not wanting spacial locations, for the reasons I gave above. Walking through The Strip would be like walking through a forest. Much better than visiting a current Web search engine. At least Stephenson does have us all stuck with the regular polygons.
VirtualRealityInScienceFiction has become commonplace