Modern Computer Interfaces are not Ideal. Modern desktops are particulary designed to look visually like Windows or Mas OS, bring nothing new in usability. Though, in 80-90 years a lot of good ideas had been born. And much of are still not implemented. Some ideas were described by Jef Raskin in his books. One of his ideas is a good cancelling system, to undo actions. There is no need to ask user if he either want to save his document or not. An application simply should allow to undo any action even if you changed (your own) document long ago.
I have been thinking about interfaces, since autumn, last year. I document my thoughts in locally created files from time to time. I think, my ideas can't be implemented in a single separate project, like new WM. It is necessary to create a new environment with the set of components. It must be a very integrated system. Also it must allow to easily edit one component, without changing anything more.
Corporations cannot write such system in the modern world. Microsoft have been blamed because it holds in Windows the browser implementation. But the close integration is very profitable for the user. So all hopes are directed only on Open Source world (community).
Already Open Source world had done plenty of applications. To write new standalone project from zero it is not simply madness, it is impossible.
Therefore, the first goal of the project is gathering documentation of algorithms and API have been used in OpenSource projects before. Further, it is necessary to reuse existent code as much as it possible.
Studying of somebody else’s code could be long and tiresome process, but it is worth effort. Sometimes it seems to write a thousand lines from zero is faster, than to understand one's thousand strange lines. But actually, trough understanding alien code, after all we'll get yet debugged code. So account must be taken of time not only to write but also to debug. And debugging in this case will occupy lion’s share of the work.