I'm new + Design Ideas

I was thinking about what it would take to replace X when I came across Y.
I read the original paper by Mark Thomas and viewed his PPT brief and I was
very impressed.  After following the progress of GGI, Berlin/Fresco,
DirectFB for a few years I came to the conclusion that they're all doomed
because they don't provide a PRAGMATIC approach to replacing X.  OTOH I see
a lot of potential for Y.

I just wanted to mention a few things.

1.  The obsession with a 3D desktop is utterly absurd.  It seems like
everyone is doing it these days.  WinFX, Project Looking Glass and now
DirectFB.  WHY WHY WHY???  All it does is increase dependencies (need
opengl), bloats the codebase, reduces portability, and makes things
unnecessarily complex.  Eye candy is cool, but a 3D desktop demands *way*
too much while giving *way* too little.  The ability to rotate and twist a
G. damned window is cool for the first week but after that everyone will be
sick of it and turn it off.

2.  Trying to make Y run on every display target under the sun is a huge
mistake.  Just look at GGI - it runs on top of just about everything and in
turn each of its targets will run on top of other things which run on top of
other things.  The reason for this is because nobody knows how to write
low-level drivers for recent video cards, so they target things like SDL and
DirectFB instead.

I really think there are only three important targets:  (1) FBDev, (2) *one*
high level library such as SDL, and (3) native low-level drivers.  Anything
beyond these is a WASTE OF TIME and just increases project complexity.

3.  Real credibility comes when we have native drivers.  Does anyone know
how difficult it would be to "copy" the XFree86 vesa svga drivers and
possibly the nv driver?  I'm not that good at low-level programming - even
if you gave me a whole bookshelf of hardware specs I don't think I could do
it.  Is it possible to recruit talented people to do this work?  All it
takes is one or two people who know their shit.  For example I remember a
few years ago when the linux USB project was moving at a snail's pace - then
suddenly Linus Torvalds himself came in and wrote a whole working USB
codebase in about three days to replace the old, buggy codebase that had
taken someone else two years to write.  What we need is a "Torvalds event"
to get things off the ground with regards to graphics drivers.

4.  Does anyone have any resources on mouse programming?  Does Y use GPM
when targeting FBDev or does it use its own native code?  What is the
disadvantage of using GPM versus native (aside from adding a dependency)?
The reason I'm asking is because I can't find any info on programming
/dev/mouse except via GPM.  How do the "pros" do it???

Mack