Right, sorry Charles. Rob Manson's comment in the video about how native apps cause fragmentation was what triggered some of my comments, but it was definitely just a side remark of his. Allow me to try again, if that's okay.
The big question underlying the video is just how to address - and prevent - that very fragmentation of native/platform-specific apps. In other words, in the world of ambient computing - where a proliferation of devices with ever-changing form factors and ever-increasing interconnectivity is the norm - how do you really manage to "write once, run anywhere"?
In that respect, Java was the champion in the late 90s but failed to deliver (for whatever reason, I blame UI mostly but I'm not familiar enough with the history of Java). .NET came along as Microsoft's take on portability (MSIL), and the idea of WPF/E had us all cheering for a long time. But because Silverlight/XAML stayed a (mostly) closed platform, it was never really adopted universally on the Web, so now HTML5 is the dominant option for a "write once, run anywhere" de facto standard. And it will stay that way.
Hence my worries when the Web becomes fragmented with vendor-specific extensions that have no plans for standardization (whether Metro, Chrome, or anything else). I love taking advantage of close operating system integration, don't get me wrong, but the only way to make that profitable in the long term (in the absence of an OS monopoly) is to keep pushing cross-OS (e.g., Web) standards.
A classic example of this is GDI vs. HTML5 Canvas. Both (more or less) neatly express the need to draw free-form graphics across myriad devices - in ever-changing form factors - in a portable way. When Windows was universal, GDI was the de facto answer for doing graphics. But with the arrival of more devices that didn't run Windows, another approach was needed and evolved from Apple's custom Canvas tag into an HTML5 standard. As browser implementations then "caught up" with the standard (in other words, as Canvas became a standard way to do free-form graphics), we developers were suddenly able to use hardware-accelerated rendering - across any device that implemented the Canvas tag! Canvas thus subsumed GDI in a more portable way because the market reality changed.
Sure, taking advantage of the latest and greatest (whether hardware-accelerated rendering in the 2000s or touch in the 2010s) will always require non-standard extensions (or even platforms) for early adopters, and there's a great first-mover advantage to doing so (that's why, for example, I keep specializing in Windows rather than trying to learn all the intricacies of Linux, Mac OS X/iOS, or Android - when I need close OS integration I can only afford to do it with one OS). But in the long term, for those technologies to become commodities, we need a standardized way to use them.
So the more I think about it, maybe IE10 isn't so bad after all. To reword what I said to aL_, as an engineering professional I should respect standards unless there is a compelling reason to go vendor-specific, and even then I need to keep pushing for the standards to catch up.