A comment combining haptics and latency, based on a real-world example. Bear with me here while I go through it. Microsoft introduced a gliding text cursor animation that leverages graphics cards, and Looks Cool. Problem is, it completely messes up touch typists by adding lag to their internal feedback loops, as the drawing of typed characters now lags behind what is typed. Disabling this animation can be done on a per-Office app basis or systemwide, but keeping it disabled (along with other animations as side-effects), and that it's on by default or even exists in the first place, are issues if you can type fast. Usual user interface features-that-shouldn't-be morass, and it's quite amazing how word processing continues to get worse thanks to teams of dedicated developers, none of whom can apparently type. https://superuser.com/questions/989951/is-it-possible-to-disable-smooth-movement-of-the-cursor-in-office-2013-windows-1 https://support.office.com/en-ie/article/turn-off-office-animations-9ee5c4d2-d144-4fd2-b670-22cef9fa025a etc. you get the idea. Now, in taking this up with Microsoft, I got to go through multiple levels of support who wanted to 'see the problem' and I had to explain, again and again, that no, screen sharing wasn't good enough to see the problem, because the network and refresh rate was simply incapable of being fast enough to even show the animation, never mind how I reacted to it as a touchtypist, and did they know any touchtypists they could try it out on on a local copy of the apps and operating system? "you actually need to send the whole screen within the latency+reaction time." and that clearly can't be done in this case. Current technology simply isn't good enough. "latencies where the laws of physics are the only significant factor" and this is a deliberate design counterexample, sitting on your desktop right now. L. (I think the nottingham draft should be informational. It sounds more like a W3C aspirational text. And the grammar in the first para of section 2 is terrible.) Lloyd Wood lloyd.wood@xxxxxxxxxxx http://about.me/lloydwood ________________________________ From: Joe Touch <touch@xxxxxxxxxxxxxx> To: Stewart Bryant <stewart.bryant@xxxxxxxxx> Cc: IETF <ietf@xxxxxxxx> Sent: Wednesday, 20 March 2019, 13:34 Subject: Re: Finding the appropriate work stream for draft-nottingham-for-the-users On Mar 19, 2019, at 7:31 AM, Stewart Bryant <stewart.bryant@xxxxxxxxx> wrote: > >Just to be clear, those are not numbers I plucked out of the air. > >1TB/s is needed to do holographic teleconferencing. I would argue that being able to have a fully immersive experience including shaking the hand of your business contact, or hugging a parent or child without burning carbon must be a good thing that many people will want to do. The haptic component of the experience requires surprisingly low latency. I’ll take bets against those numbers - with a caveat. You can’t deliberately break the system in the wrong place just to kick the numbers up. The following is a chart from the early 1990s that shows the BW needed for a variety of ways of interacting. The latency budget is 100ms, i.e., intended to provide “realtime” haptic latency, some of which is consumed by propagation delay (depending on the distance between endpoints). https://www.strayalpha.com/interactive-web/ When the chart was developed, we were shifting from HTML to icons and ISDN-ish speeds were enough. Now, nearly 30 years later, we’re pushing 300dpi displays and everyone is happy enough with 100M-ish speeds, as predicted, but only if you actually need to send the whole screen within the latency+reaction time. If you push to 3D, you’re still just a factor of 2 above those numbers. But we basically stopped paying for 3D movies (http://collider.com/3d-movies-are-dead-again/), so I have doubts about anyone caring about remote holography. Instead, I’ll bet that in about 15-20 years, when our screens are 1000dpi, we’ll actually start to want 40G speeds. Tb, not so much. Joe
