Virtual Reality: still not quite there, again (2017)
karpathy.medium.com
karpathy.medium.com
A simpler explanation is that outside of games VR simply doesn't bring much to the table.
My favorite anti-VR-hype example involves sending a letter to people. Do I virtually fold paper into thirds for a virtual envelope and virtually lick the flap to seal it before exiting my virtual house to go to the virtual post-office to buy virtual stamps? Even in a game-world, would anyone want to, once the novelty wore off?
Heck no: I click "Send" on my e-mail client.
All the 3D parts of the workflow get removed, because they're tedious and tiring and not necessary for the goal. There may be some useful 3D visualizations for understanding things, but everybody will quickly realize that standing and crouching and waving their arms around all day "picking up" data-objects and and "dropping" them into virtual filing zones is going nowhere. Moving a mouse is so much faster and easier, even more-so with abstract filter/assistant tools.
It might just solve the obesity epidemic though.
> In 2016, Pontzer and colleagues published a study putting forward the constrained energy model: Energy expenditure does increase with more activity, but only to a point. Once physical activity gets really high, the body will adjust other components of the metabolism to keep your daily energy expenditure within a narrow range
https://physiqonomics.com/biggest-losers/
https://www.theatlantic.com/health/archive/2012/07/study-off...
> There may be some useful 3D visualizations for understanding things
> Moving a mouse is so much faster and easier, even more-so with abstract filter/assistant tools.
Exactly. The only thing I'm hyped for so far is the idea that one day, we'll have virtual monitors. First, we need good enough headsets. The technology has improved since OP's article, but there's more work to do.
Some of it is software. We'd have to do the awful work of implementing backward-compatible virtual monitor and window management. Look at X vs Wayland for an example of how hard it is to get things right.
Despite how hard all of this is, I suspect it's the way forward because:
* it'd be cheaper and more flexible than complex multi-monitor standing desks
* thousands of years after inventing writing, we still rely on writing 2D characters on 2D surfaces for language
* we can inset AR and 3D panels into manual windows the way we already do to reality with physical 3D displays
IMO, some of what you mentioned is contradictory. That's not a bad thing. For example, virtual monitor and gesture-based headset-only devices might have different use cases. It's like the difference between a an iPhone, iPad, and ARM-based Mac. All might share software and hardware components, but they're each good at very different things.
To my understanding, focusing on the virtual monitors first has a number of benefits:
* Simplify deliverables and OS integration significantly
* Help developers by providing cheaper multi-monitor setups
* Dogfooding
* Get certain technical parts (foveated rendering, etc) right before trying to nail new input methods
I want to be able to put on a pair of glasses and pull up a 3D diagrammatic overlay of anything I'm trying to fix around the house or in my car, complete with tutorials. If I'm in a hardware store, I want to be able to say what it is that I want and have a path to it highlighted. I want to be able to turn a depressing walk in a crummy downtown neighbourhood into a relaxing walk in the forest. I want to be able to stroll along in that forest talking to someone as though they were walking right next to me even though they are very far away.
This stuff is, technically, not that far off. I just hope a company that respects user privacy gets there first.
Games like Horizon Zero Dawn or Cyberpunk 2077 have diegetic AR HUDs that highlight some of the stuff you could do, and I’ve definitely felt the desire to have some of those features available IRL.
The AR products that eventuallg spur mass adoption are almost guaranteed to be backed by billion dollar investments, with the expectation of more user data than ever before.
I cannot wait to get mugged by the gingerbread man
The mugging man?
The mugging man.
Do I know the mugging man... Who preys on Drury Lane?
I think it could go either way, honestly. As far as modern efficient e-mailing goes, it will simply be nice in many ways to have floating apps in the room instead of windows on a monitor, once the comfort/productivity tradeoff improves. But also, maybe when I open a real letter, my glasses recognize it as a letter, scan it into my library and blow it up into a floating virtual page, with or without OCR, maybe with additional context and history. Maybe I can choose to open a UI to respond digitally, or maybe it physically points me to my paper and pen. I can do any combination of writing a real letter or a virtual letter and sending it digitally or using the oldfashionedletters.com plugin to send out 100 cursive wedding invitations supported by whatever digital back-end is activated when the recipients' glasses detect reception of the letter. Point is, there's a unification of both approaches and they each gain new superpowers.
You're talking about novelty wearing off, but I think that might just be the difference between good and bad skeuomorphic design. Your email scenario is definitely a plausible thing that someone would make right now, and it would absolutely suck and die, but maybe it's just part of an experimental trend that will exhaust itself like that period in the 90s when every new cd-rom you'd pop in would autolaunch a unique, horrendously ugly and confusingly skinned UI that looked like it was straight out of a Cronenberg movie. It suddenly becomes easy to make a bunch of dumb shit and it takes time to see what sticks, and there's a lot of silliness on the road to best practices... but they ARE buried somewhere in there, we DO eventually get to a fundamentally better way of interacting with computers.
The things you report (initial novelty, nausea, uncanny valley, little justification for the expense) simply are not applicable to anyone who plays VR for longer than a few Beat Saber sessions in front of friends & family.
Of course it's a niche, but there is demand. Especially in Gen Z, many of whom received Quest HMDs over the last few years and have been rabid adopters who will grow up to be committed customers. Lobbies for basically every game on offer are full of them (for better or worse).
I actually thought this is where the next big AI company would come from. A gaming company that created a game that also had a business use case. They would use the data from that game to train ML models that were capable of doing work. If someone were able crowd source some AI robot training through gaming, they would be paid by consumers and by businesses. Kind of like how captcha digitized books in the backend.
Instead of building a wall full of monitors, I think running on the entirely on device makes a lot more sense. Facebook just can’t pull it off, because they don’t have a full operating system like OSX to pull from, but they were right to but the processor in the headset.
Globally they’re at 6 — 10%. The rest is practically all Windows.
These companies have hundreds of engineers just on release processes and tooling.
it's also hugely relevant for sw dev. i don't quite get what this point is. macOS has always had small marketshare, but large profits from hw
That's not entirely true. SimulaVR has been working on a full-PC-power, standalone, Linux-based heart for quite a while now. They're just small and don't have Apple's "move the world" money.
Where Will Virtual Reality Take Us?