I've been developing with C# and Managed DirectX for the past 2 and a half years, and I have learned a lot about C# as a game programming language and Managed DirectX as an API, and here's my thoughts on the matter.
1) You can be crazy productive, and knock out prototypes in an incredibly fast amount of time. The more common tasks in 3D programming are very EASY to do, but there are a lot of generally advanced things that are NOT easy (i.e. skeletal animation).
2) The documentation and object oriented style make MDX a true joy to work with for C# fans and developers, or any fan/developer of object oriented programming styles. It's incredibly intuitive and robust.
3) Deployment SUCKS. It's not because any single thing is altogether difficult, but when you have several minor pains in the ass you eventually end up with a rather huge pain in the ass.
Bottom line is that my experience with C# and MDX are great for making small and simple prototype games. However, the scalability of the API is not linear and deployment is rife with headaches. This makes it somewhat difficult to create full-featured, next-gen games. For the hobbyist, it's great.
So where does this fit in with XNA? Well the bottom line is that the same issues I have with Managed DirectX exist with XNA. It's a great Language/API combo for hobbyist or casual games, but the API is going to need a lot of maturing before we see mainstream titles use the technology. Granted this isn't an issue for most people, but it is for me (I have more lofty goals). C# needs a killer app to get the respect it deserves in the commercial front, and I just don't see it coming with XNA in its current state.
I just caution people to remember what Microsoft's true goal here is: to make money for their shareholders. I have nothing against Microsoft at all, in fact I am very thankful for what they have provided developers with C# and the .Net Framework. But the bottom line here is that hobbyist and educational game development is a new market for them to expand into. With XNA they will make a good amount of money on the sale of subscription fees and consoles from hobbyists and educational institutions, not to mention provide a lot of content for Xbox live to help compete with the Wii's virtual console. So just take XNA with a grain of salt, and if it doesn't do it for you, give Tao a try.