I'm obviously biased, having put time and effort into this particular one, but I think it may do better than the previous attempts. Two big things that I think help it:
* It's built into all modern browsers, either in production (Chrome) or early-access/beta (Firefox, Safari) with the exception of IE9. And with Chrome Frame, the lack in IE might not be a huge deal.
* It's as low-level as successful APIs like OpenGL and Direct3D (obviously being based on OpenGL). This means that it gives a stable basic API that people can code to that will work across browsers, across platforms. They picked OpenGL ES 2.0 as the baseline because it's the 3D API used on mobile devices, iPhones/iPads and Android devices being the most important. While this low-level nature can make it a bit of a pain to program if you want a higher-level API (like, say, VRML's or X3D's) it also means that people have a solid baseline on which to build higher-level APIs. There are a bunch of good ones already, for example GLGE, SceneJS or SpiderGL.
* It's built into all modern browsers, either in production (Chrome) or early-access/beta (Firefox, Safari) with the exception of IE9. And with Chrome Frame, the lack in IE might not be a huge deal.
* It's as low-level as successful APIs like OpenGL and Direct3D (obviously being based on OpenGL). This means that it gives a stable basic API that people can code to that will work across browsers, across platforms. They picked OpenGL ES 2.0 as the baseline because it's the 3D API used on mobile devices, iPhones/iPads and Android devices being the most important. While this low-level nature can make it a bit of a pain to program if you want a higher-level API (like, say, VRML's or X3D's) it also means that people have a solid baseline on which to build higher-level APIs. There are a bunch of good ones already, for example GLGE, SceneJS or SpiderGL.