My two .NET gigs have been enterprisey, line-of-business-type apps. What the whole Metro thing is going to mean for line of business apps is hard to say. WPF might continue to be the choice for things like that. I have a hard time picturing a Metro-style data entry screen, but I haven't really looked that closely at Metro. I've been figuring that it's kind of like PhoneGap for the desktop, which may be a completely mistaken assumption. Even if Metro replaces WPF universally as the preferred way of creating Windows desktop apps, I agree that the need for desktop apps in the first place is diminishing.
Prior to the big Windows 8 announcement, it was pretty clear that WPF was a high priority for Microsoft and, a pretty safe bet in terms of skills to learn. That's not the case anymore, and even if Windows 8 and/or Metro fail with consumers, the shift in resources for things like Visual Studio, etc. will could really hurt WPF (which needs better development tools pretty badly, based on my experience). That could hasten its demise and possibly .NET's as well.
The web will always be playing catch up to native desktop/mobile apps. And when web apps catch up to the desktop, there will be something else new that web apps can't do that desktop apps can better. (It might take 7 yrs to catch up -- like being able to play DOOM on the web).
NHibernate is (again, imo) arguably as good as or better than EntityFramework, yet (again, in my experience) no one tends to even be aware it exists.
I've seen this as a small part of the larger disappointment I have with "the" .NET community. Despite my love for .NET/C#, I have no love for IIS, web.config files or the broken Debug/Release configuration management in Visual Studio (1). Since I've worked with the .NET web stack, I've gone back to python, ruby and now golang and I missed them and the community (IRC, mailing lists, IRC, forums [that weren't written in 1995], IRC, etc) so, so much.
(1) (for a regular MVC project it's not awful, but for an Azure project it's very inconsistent, hard to control, and easy to screw up which is annoying with 45 minute deploy times, using something other than MVC or EF helps a bit as you can do more in the configuration either in code or XML rather than having to flex opaque VS options)
I think you can do amazing things with the .NET platform, but it mostly requires a certain amount of, for lack of a better term, fearlessness when it comes to what you build your system on. Using anything that isn't blessed is often an uphill battle where you are met with suspicion and derision.
That's the scary part. Microsoft is in charge of its development ecosystem, but it seems to be lying to its developers. They've become much more closed off about releases in the last few years, and it's worrying people.
It really doesn't have much to do with wanting to learn "one true tech." It's more about feeling misled.
The efforts by WebKit, Chrome, iOS and others simply changed the entire game, and in this game it is pointless to attack flash (silverlight 1.0) or flex (silverlight 2+)
That's what's worrying. The "best and brightest" just played follow the leader to nowhere.
Flash requires you to write it in Actionscript, whereas Sliverlight can be written in any .NET language and ported across. Just because they try to accomplish the same task (getting rich content in your browser) doesn't mean Silverlight is a knock-off.
And at the time Silverlight was developed, there was no major support for the kinds of things Flash/Silverlight did (and for some important things like DRM (think Netflix), there still isn't). It's not always possible to see the things the market is going to grab hold of.
I think complaints like this reveal what some people really want: to learn the "one true tech", stick with it and squeeze 10 or 15 years use out of it. That's a recipe for stagnancy be it from MS or an open source stack. People in tech just need to get used to abandoning their hard learned skills to stay relevant.
I realize there are real world employability fears that influence these choices, but for people like that being a generalist is probably a better choice.
To answer the question more directly however, WPF or any desktop stack is only a stop gap till web apps are more capable. It's definately on a downword trend.