My suggestion for the meantime is that we concentrate our efforts on understanding how FSX works as best as we can. Let's be able to decode all of the file formats 100%, completely understanding what they do and why they work the way they do.
I notice a lot of people saying they would like to contribute but that they don't know how to program. Although the art would honestly be the majority of the labor in this project (as it was in the development of FSX), art is useless if you don't have an engine in place to work with. So use this opportunity to learn! Admittedly you probably won't ever be able to implement a complex physics or graphics system unless you spend some serious time on this. However, some basic programming knowledge and computer science can get you far, especially in the field of technical art.
For example, if you are a talented scenery designer, start taking a look at the BGL format. Start understanding how the sim stores your data and steps it has to go through to get displayed. Learn about geodetic coordinate systems and take a look at how the simulator represents the earth (a good place to start is a paper by Adam Szofran on global terrain technology). Aircraft designers could start learning about what the data exported from max looks like and 3d rendering API's use that information.
Honestly, throwing together a game engine is not THAT difficult (relatively speaking, of course). You'd be amazed what you can throw together in a few weeks using OGRE, ODE, etc. Representing the world the way FSX does is the real challenge here; not many other games require such a precise, programmable way to represent the whole earth.
I guess what I am trying to say is to focus on how FS specific things work, not how to throw together a simulation engine. For the mean time we can work hard to document the internal workings of FSX, which is exactly what I am working on right now (and I wish many others were too!). That can also serve to improve the state of FSX addons right now. FSX is far from being dead as a platform. Heck, it's just now finally being utilized to it's full potential, 5-6 years after its release!
There is so much more out there to learn, it will only serve to improve the state of simulation if we can get it out there in the community!