This is kind of very good, but in a strange way. All of this open standards stuff is great, the snag is Khronos keep open-sourcing prototype stuff on sourceforge and Micro$oft nicks it. Open standards is ideal for creating a real prototype, such as the suggested OpenGL on JavaScript. OpenGL is still limited to scene-by-scene generation, which is fine for 3D chatrooms, IMVU use a DirectX equivalent of this.

A “scene” is called a “scenegraph” in OpenGL and would allow for the scope of something like one Quake indoor room (in mapping terms this coud be a player area equivalent to about 6 metres square). JavaScript is a very limited language and has very limited scope for streaming in files in real-time, but it can handle a nice scene at a time.
To run from one Quake location/room/scene to another using OpenGL would mean the player takes an extended blink break each time, far from ideal. This is why the rather nice Quake Live (Zero, or whatever), uses little OpenGL and JavaScript but maximal other tricks, such as playing about with exactly how backgrounds are generated. One such trick would be ray-casting, which is much more efficient than ray-tracing, and can look pretty good, but limits the player to horizontal-only movement.

They are aiming for stuff done in a year, and browser integration could probably be quicker than that, although there are several open standards technologies that still haven't made it into browsers:

• “SVG” or “Scaleable Vector Graphics” which competes with Flash for animation, but not interactivity.
• “SMIL+TIME” which was supposed to really kill off Flash, but I still haven't found an API or IDE for it. Coding by hand broke my head.
• “MNG” and “APNG” which are animated image formats, I still see GIFs everywhere, despite the fact that the struggling AOL/Compuserve own it and have warned of suing people just for using them. GIF's 256 colours is looking unfashionally 8-bit these days, yet still no one can really put up a replacement MNG or APNG.
• “XSL-FO” which is a formatting structure for web-page objects. The design is very sensible, very useful, and very unfinished.
• The W3C (World Wide Web Consortium) has some other dust-gathering standards too, which shows that although they are the people to go to for good standards, people hardly use them. Even Google still uses HTML at 3.2 rather than 4.

Khronos are full of influential tech companies, they have marketing power and tech know-how. Teaming up with Mozilla is great for everyone to get things done, but M$ must have at least a conflict of interest, unless they're giving up on browsers.
There will soon be a lot more web games done in 2.5D (the isometric view style used by many RTS games). From marketing and distribution points of view, level 1 of one of these can be free and instantly playable in the browser, level 2 can be purchased and downloaded. This is achievable, but doesn't seem to be in anyone's business model. I think once some companies have made some decent money and significant market share, they and many others will go for doing proper 3D either as engines or transport servers. The rendering can already be done on the client on current PCs, XBoxes, PS3s and even WIIs could handle it. This leaves the bandwidth to be entirely data, no pre-rendering streams of anything.

I'm not convinced the current OpenGL is the way to do it. If Khronos' committees actually look at what their members are already doing in gaming and notice that all of the hard work can be done by open-source Linux nutters on sourceforge, they might bump up the spec of OpenGL to proper game-level 3D, at least to DirectX10. OpenVR never got anywhere, but the rendering part of it is what web users want now.

As ever, I'm working on stuff to fit in with all this.