First, I've had some opinions on this "one language" ideal After PERL what?. We have a variety of languages for a good reason: the languages express different kinds of things. Shell command languages really have little to do with markup languages, database manipulation languages or "general-purpose" programming languages.
One of the C#/LINQ issues is to collapse some higher-level predicates into the programming language. The problem is that predicates mean loops and loops mean proof of termination and those proofs are impossible in general.
Further, there is the difficulty in optimizing those predicate loops. Oracle has had years to work out cost-based optimization, and it is still looked on with suspicion.
If you have no persistent data, how do you do optimization? You have to know something about the data and the algorithms. This makes general-purpose predicates not terribly useful.
When looking at C# from a distance, I have to ask about innovation in the Windows world. Are there barriers or is it my personal bias?
I think there are barriers.
First, and foremost, everything in Windows world must play with the obscure and rapidly-evolving Windows OS API's. The lack of clarity and stability in the closed-source API's is a barrier to innovation.
Second, you are competing against Microsoft. If your idea is good, it will show up in a competing closed-source MS product. While good for your idea, the innovation is effectively clamped off once it becomes a closed-source product offering.
Third, almost everything has a large fee associated with it. The tools, the platform and the libraries all cost real money. This is, I think, the reason why the open source folks have an edge -- they can work for free.
I'm happy that I don't have to sweat the details of this .NET, C#, LINQ, CLI, Managed Code world. Oracle's OAS, OC4J is complex enough, and much of it is open source. I can't imagine how I'd make a complex closed-source environment work.