(Gravestone: The title used to read "- by 2030". It was a bad title. I'm surprised to find that I might actually believe the 2030 target, but the factors that get us there are weird and cultural, and I've mostly only presented the much more legible technological factors here, so never mind 2030 for now. I'll try to examine that case properly later.)
This post was written as a resource for a set of project proposals in VR communitybuilding and transformative social technology design, but it should be of interest to anyone who'd like to have a rough idea of how our world will start to change very soon.
I should first make it clear what kind of devices we're talking about: The VR headsets of 2025 (or potentially even next year) will be qualitatively different from what we have today:
- Apple and Meta are known to be releasing headsets in 2023 with
- face tracking, the ability to capture and convey the full facial expressions of the user, including,
- gaze tracking (which may potentially obsolete the use of mice in most kinds of work)
- "passhthrough", the use of external cameras to allow the user to see and interact with the external world while they're operating in the virtual world. (Meta's Quest 2 already partially does). At some point, there wont be much of a functional distinction between virtual reality and augmented reality headsets.
- Potentially, varifocal lenses, which solve an eye strain issue, and potentially, if we're lucky, will avert any role VR might have played as a cause of myopia.
- VR displays are on a trajectory to hit the maximum perceptible level of visual detail (60 pixels per degree) by 2026. There are plausible cases that next year's apple headset will already be there (although Apple's headset wont be immediately transformative, as it seems like they wont be able to build enough to meet demand immediately, unless the plan has changed recently).
- They're getting lightweight. "Pancake lenses" will reduce the weight and volume of optics. Unlike today's high-end headsets, it will be easy to forget that you're wearing anything at all, comfort will be less of an obstacle.
I think that adds up to a generally preferable modality for computing, functionally equivalent to having an extremely large computer screen that also weighs nothing, fits anywhere, and moves with you. The prospect of continuing this practice of coupling a relatively small screen to your keyboard then hunching over it all day (I'm describing laptops) seems like it will be difficult to continue to justify. Desktops on standing desks hold up a bit better, but...
There are crucial practical activities that just aren't possible without VR:
- Many kinds of exercise games. Some people don't have access to nice places to go running or cycling, or they aren't able to find joy in it. I have found that exercise games make it much easier to push myself far beyond the point I'd normally go by projecting meaning onto the actions, EG: fleeing, chasing, fighting, or cutting virtual wood.
- A much better modality for any work that requires manipulating 3d objects in space. CAD work, for instance (which I'd expect to make up an increasing proportion of the GDP as additive manufacturing improves, as manufacturing automation increases, or just, as production processes generally become more accessible to more people.)
- Any sort of remote work: Having a sense of presence in a focused office (or at least, a sense of being in a busier place than the home) is important enough for getting focused work done, that people will sometimes pay for seats at coworking spaces even when they're working alone. VR can grant you the focus of being immersed in an office without a commute, without the fee, and the office will be fantastically better than any real office, it can be linked to other offices in non-euclidean ways, it can be restructured and rearranged cheaply, it can look and behave in impossibly magical ways to support impossibly magical vibes, and also you can't get respiratory diseases in a virtual office.
- Connecting with others remotely in general is a much better experience in VR than it is over Zoom! Reasons discussed more below. This is where most of the social transformation is going to come from.
So I think basically everyone will end up having or wanting or needing one or being hounded to get one by their friends (I have never seen so many people so willing to buy their friends multi hundred dollar devices), once we all start to notice that VR is really strongly preferable over text chats or video chats, for meeting new people, hanging out, or talking things through. I'm surprised and not totally sure why VR is already so preferable for me over video chats, but it probably has a lot to do with the following technical advantages:
- Spacial, directional audio, being able to implicitly convey your focus of attention by facing, and walking around the space, allows group conversations to organically split and merge. This is extremely important for anything aspiring to work in the way of a meetup, party or conference.
- Some body language. Pointing, gesticulation, moving around. (Full body language should arrive with camera-based tracking, and can be had today with special equipment.)
- Non-disadvantage: Although face-tracking will start to deploy next year, lacking face tracking isn't as big of an issue as you might expect, as the information that facial expressions convey is mostly already transmitted in speech. Don't believe me? Check out Meta's surprisingly good expression-recovery-from-sound thing. I admit that it's not so effective that I'd actually want to use it, but it's an illustration of how much voice conveys.
- (Believe it or not, video chats could eventually catch up with VR on these features, with promotion of the use of headphones, specialized compression algorithms, or filters, but it hasn't yet, so I'll still mention them.)
- Most people don't realize this (and that's a large part of the problem), but when you have an audio call without audio isolation (without wearing headphones), the system silences you when other people are talking in order to prevent audio feedback. This disrupts natural turntaking negotiation and makes very lively conversation very difficult (impossible?). VR headsets don't have this problem.
- It's basically impossible to make eye contact in video chats. It's basically possible in VR (Eye tracking is coming, but head tracking is kind of close already enough). This is subtle but it's probably pretty important for establishing a sense of connection!
- Not having to stream a load of video in and out (streaming relatively compact encodings of gesture or facial expression instead) often makes the data connection of VR chats more reliable.
In sum, VR gets us much more natural online conversations than prior mediaforms.
Textual online communities are good for their own reasons and wont go away, but humans are not natives to that format, in text, we have difficulty judging tone, processes don't come naturally, and it definitely makes it difficult to make friends: You can't see anyone, on some level you find it difficult to relate to them as real people. VR lets you to feel the reality of distant people. It's much easier after sharing space in VR to feel that you've met the person, and that they're a part of your world.
We highly recommend this People Make Games documentary about VR Chat if you want to see the extent of the vibrancy of VR as a social environment. I hope that near-future VR culture will be less awkward than this, hopefully the semiotics will be more self-aware and evolved and refined but personally I'm already vibing.
And we should expect VR offices to be productive: Immersion, in online social environments, is not a minor thing. The sense of being transported to another place, being bodily present with others, is important for committing with your whole self to the people you're with and the thing you're doing together. It's important for connecting with people, it's important for working with them: It's much much easier to focus on getting stuff done when you feel bodily present with people who're also doing it, who visibly expect you to do it too, and who are perceiving whether you are doing it.
So, yeah, I think it's gonna happen.
And, I can't write this up fully yet, but, there's stuff we should be doing about this. The things that are built in this new place, the standards we adopt, are going to be highly consequential. Social technologies determine the shape and the character of the communities that form under them, how they will interact (EG: How well AI research organizations will coordinate!) and how they will spend their time, and that basically determines everything.
We've been gathering technical principles, components and designs for a Social VR Engine that we'd expect to compete well against other projects and improve online social health. I kind of expect us to seek funding pretty soon. I will probably write publicly about those designs, as anyone attentive enough and apprehending enough to copy and deploy these specific world-improving social technologies would sort of necessarily be an ally as a result of being that way and doing that for us (but most likely no one will).
If I've gotten you curious about VR, if you decide to check it out, I strongly recommend visiting us at the EA VR group.
There's some ambiguity as to whether eye tracking will obsolete mice in general. Eye tracking seems to have an insurmountable limit on its accuracy, the eye is always bouncing around somewhat, seemingly uncontrollably and (currently) unconsciously. I assume it has to. It may be possible to learn somewhat better control through practice, but I've been unable to find research on that. I expect some limit to hold.
However, head tracking may be able to make up the difference, when it counts: https://precisiongazemouse.org/ I haven't tried this, it's not clear from the video demo whether it's preferable to a mouse. It seems to be using an eye tracker that is below optimal accuracy. I don't know whether the upcoming headsets are going to be near optimal, but I'd expect them to be closer. It's definitely going to be possible to accurately point and click on sufficiently large things without lifting our hands from our keyboards, which is pretty cool.
Source: A report tells of Sony providing a pair of small 8K OLED screens for Apple, which is more than enough detail to satisfy even the sharpest human eyes.
An aside: The report also tells of a mysterious, third, lower detail AMOLED screen. I seem to be the only one who thinks this, but I'm pretty sure the third screen is going to go on the front of the device, so that the headset can display a (potentially stylized) rendering of the wearer's eyes. Being unable to see the wearer's eyes when they're using VR is actually a pretty strong social obstacle to normalizing it! So we should expect Apple to care a lot about that! Note that the front of the device seems to be black (like a phone that is turned off). It absolutely could be a screen. The only alternative theory I've heard: that the third screen is going to be put behind the others and used for peripheral vision, seems really weird to me from a manufacturing perspective. I don't know how they'd make that work.
Why hasn't VR penetrated CAD already? Because adequate VR headsets could not be bought for any price 5 years ago (every headset was either too low res or too heavy to want to use all day, iirc), and because CAD software is sophisticated and will take a while to adapt! Why didn't the industry prepare in advance of the arrival of today's headsets? Because tech forecasting skills are both rare and difficult to identify. But it is beginning to happen! Consider McLaren Vector Suite for instance.
Would you care to make this precise and give it a specific probability?
I've dabbled in studying technological adoption (telegraph, phone, radio, TV, internet etc).
If I had to make this case, I would define parameters as:
What % of people use this medium every day/week/month? A more interesting sub-question would be to compare users of multiple platforms or users that "leapfrog" and use newer platforms without using older ones. For example, many people in developing regions use mobile internet but don't own PC. Mobile gaming is close to making more annual revenue than console and PC combined. And some crypto apps like Axie Infinity have reported high number of users who do not have a fiat bank account.
What % of people would cite this medium as a means of contacting friends/family/spouses?
What % of people use this medium as primary means of making new friendships/relationships? For example, we've recently passed a threshold where 40% of newly married couples met online, which a strong indicator of social habits shifting online.
I'm not in tune with the VR industry, but it's not the worst bet if you think VR adoption will follow the same rapid growth trajectory as mobile platforms.
Sure. 96% probability that more than 50% of all voice conversation in the OECD will take place in virtual spaces, through VR by the end of 2030.
(Edit: I answered this too quickly. Making beliefs precise is hard.)
I might be able to give a higher number if we could weight conversations by meaning/impact/depth of connection, but I don't know if I want to have to think about how we'd measure that, or be hooked onto spending a lot of time adjudicating between different survey formats.
I'll bet against this, my $1k against your $10k, for implied 90% odds.
Very open to it. It occurs to me now that I know very little about the cultures of many OECD countries: It includes countries that have much stronger food cultures (hard to share meals over VR), weaker media cultures, and larger aged populations, and also the "OECD" might be expanded at some point. Should we just focus in on the US.
I trust we'll be able to agree on the rest of the operationalization, or a mediator, and exemptions, when the time comes.
Thinking about this has made me realize how nearterm crypto prediction markets that depend on staking are going to be. If had to stake the money for 8 years I would have just been like "lol no". But any mechanism I can think of for extracting debts from a person's 'main wallet' after a long time period are basically going to be incompatible with anonymity or possibly even financial privacy or exit rights. I'm not sure that cryptopunk pirate types will ever build such a thing. In the least, the system itself would not have a very cryptopunk or pirate character.
To take this bet, say "book it" and whether you want it to be about OECD countries or just about the US.
Disappointingly, realizing my actual beliefs are softer and broader. When I said "most", I was thinking "most of the places that I think we should be paying close attention to and operating in (ML research)," but I really must not have wanted to expose and discuss that assumption, probably due to some crappy dynamics that I am getting tired of dancing around...
There might still be a bet we can make, but it would presumably be about the parts of the world that I think are not strategically relevant, it always is, and I would have to study them even more, despite not being authentically interested in them. Would you really make me do that?
No. As a result, I strongly downvoted the post.
Understandable. I wish I'd put more thought into the title before posting, but I vividly remember having hit some sort of nontrivial stamina limit.
This seems way too high to me. I think you should break things down (more).
Some conversations/breakdowns worth considering:
I also expect some inconvenience from actually doing work within VR and switching in just for meetings to be annoying and not very useful.
My assumption was that most of these categories will continue to depopulate (or host fewer conversations) in response to technological change, and note that the distribution of conversation is already heavily concentrated in bars and parties, and bars aren't doing so well right now (should I assume we'll get better covid vaccines by then? But will the bars come back?), and they'll face extremely intense competition come VR.
I'm hopeful that physical schools will depopulate, but it's possible that's where my prediction will fall down, depending on how the myopia stuff turns out, and depending on whether education stuff can develop fast enough.
I think things will return basically to pre-pandemic normals and I'd guess that they mostly have already, although with more remote work.
I think it's extremely unlikely that daycare, preschool, elementary schools and high schools will have a significant share done virtually.
I could imagine office work and university education going largely virtual, but 2030 seems far too early for most of it with very high probability. Even if we had perfect VR now, I still wouldn't expect more than half to go virtual by the end of 2030. Things will probably move much more slowly than that.
I would be interested in what the market would think on a Metaculus question.
Occurred to me that this kind of has to assume that more than more than 50% will even own a VR headset by then, which seemed a bit bold, so I decided to check against smartphones ownership statistics. Yeah, smartphones have > 85% adoption. It'll need to happen faster this time, but I expect it will.
But it doesn't quite have to assume that, it could also happen as a result of VR tending to play host to a greater proportion of the world's spoken conversation, per word, that the distribution of conversation will not be even over users and nonusers. Sad as that may be, it's plausible. It is easier to find people to talk to, online, and the venues are often better.
I don't think VR will be so good that it replaces text and other asynchronous social interaction, given text's many advantages. (There is a reason why we are here on the EA Forum, and not some EA Youtube Channel!) Similarly, it will be hard for VR to outpace the social interaction that people have with the people they live with IRL, becoming the dominant mode of "live" conversation. So "most social interaction by 2030" seems crazy to me.
But I think some scaled-back predictions would be much stronger. At least among the tech & EA circles, VR will definitely surpass zoom-calling and telephone calls by 2030. And I wouldn't be surprised if VR became the dominant mode of live conversation specifically among some subset of remote workers (like "young people living outside of urban centers") by 2030. I would also predict that by 2030 there will be a lot more remote workers!
Online community building
I think it's easy to underestimate the value of online community building due to it requiring collective effort to get off the ground. EAGT and EAVR sprung up at around the same time, and they're partly in the same boat. Insofar as there is or will be network overlap, we could benefit from each others' existence due to increased network externalities from a larger population. If either or both can demonstrate a highly successfwl community, I think the value of the more general model of online community building is proven right.
The reason a widely-used online EA office hasn't spontaneously sprung up yet is not because the community wouldn't benefit greatly from it, it's rather because it's a coordination problem. There have been numerous local attempts at getting something like this off the ground, but it can't sustain itself unless it reaches critical mass, so just trying to do it locally was doomed to fail from the start.
A space like this only has value to you if you can find people there that you like coworking with. And the people you like coworking with won't be using the space unless they can find people they like coworking with. So if you end up trying to use it at different times, you'll separately log on only to find an empty office and decide to log out again. Alternatively, if the network is small and you only find people you aren't enthusiastic about coworking with, you might end up logging out again before compatible people log on.
For projects with a steep Lindy longevity (future life expectancy proportional to age; probability of failure decreases over time), network externalities (the utility of the network per user grows with the number of users), and positive threshold effects (e.g. critical mass/escape velocity), it doesn't make sense to aim small--that's just planning to fail. Instead, it makes sense to invest heavily at the very beginning, when all the value seems the most uncertain and the project hasn't had time to prove itself. For opportunities like that, you can't expect to have the value demonstrated to you before you invest in it--you have to rely on theoretical understanding or guesswork.
While an IRL EA neighbourhood could be higher value, an online "embodied" community platform captures some of that value, and it can cover a much larger slice of the EA community. And, unlike temporary university groups, online platforms are always accessible regardless of changes in life-situations. Plus, the EAs who benefit from the online platform are more likely to be those people who counterfactually wouldn't have been able to have high-fidelity interactions with other EAs, because they're the people who are less likely to have access (or inclination) to an IRL community. So the counterfactual impact per person of online community building may be higher. Maybe.
Why regular video chat is qualitatively different
Can confirm. I think this is also why people underestimate the difference between e.g. Zoom coworking and coworking at the Gather Town. I call it the ladder of escalating interaction. VR is much better for this than Gather, and Gather is much better at this than Zoom.
When I log on to Gather, I can walk around and see what people are roughly doing based on their locations on the map. The fact that people are online means that they are at least weakly signalling that they are open to interactions. I can acquire this information without having to send them a message to check what they’re up to and how they feel about interacting, and without having to commit to interacting in any way. Thus, when I do interact (e.g. messaging someone to ask if they wish to chat, or joining a coworking pod with someone in order to work), it’s much more likely to be mutually beneficial.
Compared to Zoom, an online embodied community platform makes it much cheaper to meet other EAs and maintain a community. I frequently have 5-minute chats to briefly catch up with people when I notice they're online and I can walk up to them in social areas. It's much more like sharing an aspect of daily life. Now, technically, I could also just ask my FB or Discord friends for a 5-min chat to catch up, but the costs and uncertainties involved in arranging that are higher. And it's hard to meet new people this way.
This has sort of the opposite of the coronavirus pandemic. When forecasters called out for collective action to prepare for the looming pandemic before the threat was legible (the only time when it could have worked), they looked like fools to ordinary people. And when someone points out an amazing opportunity and calls for collective action to realise it before the value is legible, they may look like fools--or worse, people may accuse them of not being humble enough, not being realistic, and tell them to get off their high horse.
I haven't been able to try the gathertown thing much, as my computer always became very loud whenever it was open. That computer has just died, so I might visit soon. Unfortunately coworking in VR is still not a very good experience.
But once it is, I think establishing continuously populated shared workspaces in VR will happen without any deliberate coordination, as soon as it's practical. I'd expect to start around 2025.
Maybe we should be deliberately coordinating about it anyway though! It's conceivable that the network structure we get without deliberate coordination will be much less humane/productive than one we could design - or it might be that they end up being exactly the same, this is something we really need to investigate and will hopefully discuss at length in EA VR's social tech design channel!
Oh. well, it's actually possible that virtual offices wont happen without coordination in 2025 despite the hardware being ready then, for the very dumb reason that the software still sucks, making social coworking software is extremely hard. There will be virtual offices, but there wont necessarily be virtual offices that people want to hang out in organically.
Incidentally, that's a very large part of what I want our group to be working on.
VR is better for events, talks, hang outs, discussions, imo. But when I need to work, I just can't do the headset. So I'm hoping you're right about them being much lighter and more convenient to wear.
This, on the other hand, is just plain false. :p
Ehh it's true on a certain level x] maybe it's just entirely the "feeling like they're in your world" thing. When I meet a person, I feel responsible for them. I pay much more attention to them and think about them more often. I usually try harder to like them. They enter my dunbar sphere.
Every time I've used VR (including latest ones), I feel sick and dizzy afterwards. I don't think this issue is unique to me. It feels difficult to me to imagine that most people would want to spend significant daily time in something that has such an effect and nothing in this post addressed this issue. Your prediction feels wildly wrong to me.
It's not unique to you, but I don't know how common it is either. It hasn't been trivial to find statistics on how many people get how much nausia and how long it takes them to find their legs. It's something I want to look at.
What were you doing in VR, btw? Were you using teleportation controls? There's a hope that good enough hardware will be able to just totally avoid the nausia triggers.
I've noticed that, at least for me, immersion and nausia are coincident. The more aware I am that I'm sitting in a room using VR, the less nausia. Every piece of entertainment software wants me to forget I'm using VR. We might only start to see solid trials of zero-immersion apps once VR is a practical interface device.
Trying to figure out whether passthrough will make it better or worse. Unsure. People don't get nausia from wearing sunglasses, do they? Would it feel any different to that? (maybe?)
Just as an anecoditcal observation. I've had two cases of people who would claim to feel sick and dizzy in VR when they tried it, but then when I put them in a cool social situation in VR they reported not feeling dizzy any more. It's like the dopamine of being in a cool/interesting situation (mostly social) seems to make the brain overcome the dizziness?
Theory: Seeing other people happily doing something is a stronger signal to the body about whether it's okay than proprioception error. Consider, seeing other people puke is gonna make you want to puke. Maybe seeing people not puke at all does the opposite.
This post made me think about whether most social activity in OECD countries is through messaging / social media apps. Unclear how to quantify, but doesn't sound implausibly true by some measures (!?).
Seems like a pretty excellent analysis, best I can tell.
I would like to add this prediction. The technology continually evolves, which is indeed interesting. But the people using the technology remain largely unchanged. So if one is bored on, say, Facebook, one will likely eventually be bored in VR as well, because it will be the same people doing and saying the same things, over and over again.
I'm interested in the claim that the networks (and so, the ideas) of textual venues are going to stay the same as the the networks of voice venues. It's possible, there's a large overlap between oral and textual conversation, but there are also divergences, and I don't know if it's clear yet whether those will grow over time or not.
Voice dialog can traverse topics that're really frustrating and aversive in text. And I find that the people I enjoy hanging out with in VR are a bit different from the ones I enjoy hanging out with in text. And very different in terms of who I'd introduce to which communities. The social structures haven't had time to diverge yet and that most of us are most oriented in text and don't even know the merits of voice and how to use them.
But yeah, I think it's pretty likely that text and voice systems are never going to come far apart. And I'm planning on trying to hold them together, because I think text (or at least, the text venues that are coming) is generally more wholesome than voice and voice could get really bad if it splits off.
The claim that people aren't going to change... I don't think that's true. VR makes it easy to contact and immerse oneself in transformative communities. Oddly... I have experienced doing something like that in a textual online community (we were really great at making people feel like they were now a different kind of person, part of a different social network, on a different path), but I think VR will tend to make that a lot more intense, because there was a limit to how socially satisfying text relationships can be, and with VR that limit kind of isn't there.