Someone sent me a question regarding how to get started doing CE development.  I will preface the answer with the simple fact that I have been doing this since 2000 and so don't have the freshest eyes on the getting started question.

So I will take a general approach on the problem.  Any time I want to learn a new technology I start out by coming up with a sample problem and then start digging into the technology to see how I could solve it.  I tend to buy a lot of books and follow along writing code. I learn through my fingers better than any other way.

So if you don't have a copy of CE already download a copy of the CE trial: http://www.microsoft.com/windowsembedded/en-us/products/windowsce/default.mspx

Then try building an image (run on the emulator which is included).  Get a handle on how a modular OS works.  Go hit a book store and find a book which matches your learning style and try some code.

Only then if things look promising would I buy a spark kit.  I picked the ARTiGO because it has the interfaces I needed to test out the technologies I wanted to play with (USB for web cams, Serial port to interface to Micro-controllers and debug, fast CPU for image processing).  The ARTiGO has one downside in that it doesn't have any general purpose IO pins readily available.  So in the future I might switch to a different board.  I realize that casually switching boards involves an expense that most people can't lightly take (I am using hardware from the SPARK project team) so try to plan out your needs before you buy a board.

I would recommend making a matrix of features you need, prioritize them into three buckets (Must haves, nice to have, bell & whistle haves).  Then break down the board options to match those.