JPEveryone can kind of imagine a version of the apple headset that is, size and shape and weight-wise, a regular pair of glasses. If you want a little window into how deeply irrational and faith/vibes-based many people are about technology, look at people who are convinced that the "just normal glasses" version is just around the corner / possible at all.
Like, the only way to describe the technology that would have to exist to deliver that is "mythical". Moore's Law completely broke peoples' brains, ie specifically by making people believe that thermal, mechanical, optical, ergonomic engineering are all just functions of Moore's.
Dan Hon@jplebreton I don't think people understand exactly how far we are away from fitting that amount of compute and battery in something that size, without even having something bigger than an original iPod and ultra low-latency high bandwidth wireless link to it from the glasses. It's wild.
@danhon The question isn't even "how far away are we from that" it's "is this even possible", we are up against some totally inflexible laws of physics but peoples' mental models are still calibrated around "computers went from filling a room to fitting in your pocket in ~60 years".
Marian@jplebreton @danhon Are there really any physical limits to how small you can make VR hardware? To me, it seems like it's mostly an engineering problem, not a physics problem.
@marian42 @danhon engineering realities (die size/density, power consumption, battery capacity, heat dissipation) are ultimately constrained by the laws of physics, and the huge strides in transistor density from 1970-2010~ happened when we didn't really have to worry about most of those concerns. now we do. there are eg battery breakthroughs coming down the pipe, but their yields are nothing like the orders-of-magnitude yields of the Moore's Law golden age.
@jplebreton @danhon that seems reasonable. But at the same time, we are far away from the physical limits for computation. If VR moves from a niche toy so something that everyone uses (I don't necessarily predict this), it will become feasible to put much more resources into miniaturizing the hardware, just like what happened with smartphones in the 2010s
@marian42 @danhon "we are far away from the physical limits for computation" - that's an interesting statement, actually. i guess it depends on which limits you mean - like yeah in terms of raw compute available we can probably scale up absurdly (and burn down the rainforests building it, etc). but when space/volume is the constraint, i don't think it's a given that "VP but in a pair of glasses, runs for more than 5 minutes, doesn't burn you" is a thing that can physically exist in our universe.
@jplebreton @danhon I was thinking of something like this: https://en.m.wikipedia.org/wiki/Bremermann%27s_limit
For example, the human brain is much more efficient in terms of energy and mass usage than any computer we can build, proving that we are far away even from practical limits. I'm not thinking of using extremely high amounts of energy, like you suggest, but rather extremely low energy usage.
@marian42 @danhon the brain is amazingly efficient yeah, but it's also not doing hardly any of the things we ask a VR headset to do. our entire civilization is built on avoiding using computation wherever possible / using as little of it as possible, eg to get a spoonful of food into our mouths, and externalizing the computation we can't avoid in various ways (counting with stones, the abacus, etc). if we built a brain-like computer it wouldn't be able to do most of what we need a headset to do.
@jplebreton @danhon I'm not saying the brain is a good architecture for a CPU in a VR headset, it's just an example that shows how much compute is (physically) possible with a few 100g of matter and 20W.
@marian42 @danhon i'm imagining theoretical far future humans playing a Zachtronics game where they can arrange any physically possible configuration of atoms to create a device that can turn energy into a properly simulated + rendered AR scene at 100fps and like, there is minimum spatial and energy footprint there. we don't know what it is rn, but it's not a given that it's <= size and mass of a pair of ray-bans.
@marian42 @danhon the biggest (read: most accessible) squish factor is obviously "how could we reconfigure the problem space / cut corners to get the result we want with an energy + matter configuration that is far far easier to attain" - that's the big unknown to me, and maybe there is absolute magic out there waiting to be discovered. tbc i hope we find it!
@jplebreton @marian42 put this way (and JP let me know if I'm out of place): the brain *can* be a turing-complete computer, but it resorts to doing it by hand (system 2) or... creating tools to do it.
But what we *want* is that turing complete super fast computer, like JP's Zachtronic's example. I take your point though that reaching JPs minimum spatial+energy/computer footprint is "just an engineering problem", but history is littered with attempts at "just"s :)
@jplebreton @danhon This is where my intuition disagrees with yours. Since we're a factor of 10^30 away from the Bremermann limit, it seems plausible that we could achieve a 10^3 improvement compared to Apples current hardware in the scenario you describe.
@marian42 @danhon ah, i'd forgotten the name of Bremermann's limit (and what terms to search for it by), thank you for reminding me. if i'm understanding it correctly though, it's about theoretical bounds not real engineering, eg it understandably doesn't offer any answers for what the energy requirements (construction or operation) of its earth-sized system would be, nor how we might arrive at its design without way more computation than it itself is capable of.