My interest in WebVR is almost entirely for it's discovery and delivery advantages. Downloading, installing, configuring and running a native application certainly will always be the more reliable path to consistently high performance. If you want to sit down and devote some time and effort to playing "Elite Dangerous", native is clearly the way to go. But, for a quick demo or cat video snack experience, are you really going to install my app? I expect most people would greatly prefer the convenience and security of the browser sandbox for the small experiences.

I'm also not a big fan fancy browser demos. People have this fair speciation that anything in a browser will run well on their 5 year old laptop. This is basically true. Except for when you get into 3d content. There's also an expectation that things load relatively quickly. Rich games with tens to hundreds of Meg's of content do not load quickly. Then they continue to not load quickly everytime you go back to that page. (If not everytime, then sometimes, which is even worse)

I do agree that there should be a fast and easy way to experience VR, uh, experiences without having to download native installers for every one. I'm not sure what way that will be. I am extremely unsold on the browser being the solution. But if it works out then that'd be pretty nice.

I think the browser will be powerful enough for some simple VR experiences that consumers will want to engage with for <5mins. At least at the beginning. I've been working with the webVR stuff and have gotten a consistent 75 fps on an MBP, which is the DK2 limit. Not a 5 year old laptop, but also not a heavy gaming rig that Oculus expects people to own.

Maybe webVR won't be the way to get AAA titles in VR, but it will certainly lower the barrier to 1) development of VR content 2) distribution. I think webVR can do a lot for the adoption of VR without "poisoning the VR well."

Is that based on anything or is it just a gut feeling?

Unreal Engine in asm.js looked pretty impressive to me. If we can get VR to work on the Galaxy Note, albeit with custom hardware access, then Firefox on a desktop doesn't seem unreasonable.

Having low latency, no frames dropped running natively is hard. Exceptionally hard. One of the biggest issues is actually drivers and the near infinite number of hardware permutations. It's still a major issue with the SDK. Now let's add a super heavy, bloated web browser in between the OS and pixels. I've yet to play any decent WebGL demo that didn't have minor hitches or dropped frames. VR is merciless. You need to make your frame 100% of the time or it's a majorly jarring issue.

You didn't address the Galaxy Note part of the comment.

Also, I think you're exaggerating. I'm sure http://timeinvariant.github.io/gorescript/play/ would work fine with VR.

Relevant xkcd...

http://xkcd.com/1367/