I like D as a language to learn more about lower-level domains. It tends to be (imho) easier to read than C code and there are lots of interesting projects going on over at dsource.org
Some people seem to dislike the fact that there are the two standard libraries, but I think that just adds to making the learning curve less steep (i.e. you can choose whether you want a closer-to-the-metal or a more high level flavour of the language). Reading code written in different standard libraries in itself can also be quite illuminating.
Well, the author of this article is Andrei Alexandrescu. If you were a serious user of the C++ language you've probably read his "Modern C++ Design" and other stuff he's written -- so he has as deep knowledge of C++ and its stdlib as you can get, thus I judge his enthusiasm for D as in a way higher class compare to a a blog entry by someone who has just found Haskell/Erlang/or similar (but still somewhat theorethical).
D2 certainly sounds refreshingly simple. Having said that I don't fancy rewriting 125 KLOC of Python in even D. My performance critical code uses Cython. I'll have to try to use e.g. PyD (or some newer replacement) to write some code in D to interface with.
In a typical WEB app however there's usually not much stuff you can say: hey, this is totally independent CPU-bound code running for a significant amount of time on a limited set of data that can be written in a lower-level language.
(Oh, I suddenly feel so dirty for upgrading a server from 16 to 32 gigabyte to allow for bloated Python data structures!)