[codehusker]

Andreesen is right about the past, but we're near the inflection point where the curve smooths out.

8K video (near IMAX quality) can be compressed into a 500mbps stream[0], so gigabit should have decent lifespan. If we lay fiber to the premises then we can increase speeds in the future without digging everything up.

[0] https://www.extremetech.com/extreme/130238-8k-uhdtv-how-do-y...

[mncharity]

So if VR next year is 2K per eye, and is hypothetically cloud rendered, then that means gigabit would support 16-ish people? Sounds good. 2K per eye was at CES last month, and is badly needed.

But 4K per eye prototypes exist, and will almost certainly be at CES next year. And will ship, maybe next year, maybe the year after. That's 4-ish people on gigabit? Hmm. Doesn't that seems a little bit tight to be confident that "gigabit should have decent lifespan"?

[momerath]

Cloud rendering is a fantasy for games on flat screens, because the latency is too high. To maintain 'presence' and not get sick in VR, the motion-to-photons latency has to be _consistently_ <20ms.

360 video for VR is certainly a bandwidth-hog, but I think that could well be offset by most VR content being game-like, where, though the game might weigh in at 40G, you download it once, and spend 40 hours in it, vs a 4k movie at the same size, which lasts 2 (and which you'd likely re-stream if you watched it again). In other words, widespread VR use, even with next-gen hardware, could actually lead to a reduced demand for bandwidth.

[mncharity]

> Cloud rendering is a fantasy for games on flat screens, because the latency

One approach I've seen mentioned is sending a larger-than-used field of view (and resolution), and then locally deciding, with more recent orientation data, which portion to use. Also depth segregation of the scene, and sending multiple copies of nearfield, selected by recent position.

These approaches might result in sending more data than the displayed video.

I don't necessarily disagree with your suggestion of gigabit adequacy. Though after hearing similar suggestions so many times over the decades, about everything on Moore's law curves, and then having them almost always be wrong, I'm... leary of this form of suggestion.

But one way such estimates fail, is being hit by an "oh, we didn't expect that one". So I'm brainstorming (well, merely sort of musing) about potential surprises.

Hmm, surprises... One advantage of a single video stream is the system always knows what's needed next. The user may turn it off, but not much else. The above are perhaps examples of needing to send speculative content, which acts a demand multiplier. The future equivalent of web page preloading.

As people acquire automated assistants, one thing they may do is speculative exploration and data gathering. When a user's eye pauses on a github project, not just download everything about the project, to produce the desired pithy little briefing popup, but also other pages associated with the repo authors (still alive? any replacement project?), news articles, related work, and so on. Automation of the 'github project evaluation dance', which in the fine-grained node.js ecosystem, is frequent. So what is now a few bytes of web link, and a rare on-demand textual mouse popover, followed by slow manual surfing, might be become an immediate massive demand spike, and pervasively common? One potential of VR vs 2D screen UIs, is that while screen realestate must be severely managed, else clutter, VR may permit vastly greater inclusion of speculative "some related stuff", blended as low-cognitive-overhead ambiance. Once upon a time, a web page had the bandwidth demands of a few lines of ascii email from a dumb terminal - no longer.

[elcct]

But you assume only one stream will be played at a time which is unlikely.