I'm thinking about abandoning Metacity support for my themes too. My next theme that I've started working on is not going to have a Metacity theme as I see no point in supporting it when I'm not using it and when it's going to be abandoned. A quote from the musical artist Pogo: "I don't make art for you, I make art for myself."

But when I think about all this fragmentation I also think about Elementary OS Luna that seems to do things right, especially with application design. I'm not saying that eOS Luna is perfect. Personally there are a few things about their interface that I just find a little bit annoying. Example: Docks may look nice but tend to grow annoying. I believe a desktop can work with 1 bar or panel. 1 click (mostly), 1 panel, 1 everything and always accessible are my thoughts right now.
"When it's done philosophy" is actually quite nice in this society that just shoves new things up our asses and then makes us throw it away a few months later so you can consume more and repeat. I'm not thinking about using a tablet or smartphone in the following 5 years for example. So why don't I use MATE, KDE, XFCE or Cinnamon then? Well, I feel that one extra push is needed to hopefully bring in some new good standards. I hope for CSS-styled interfaces and javascript for extensions and small apps. Actually I do use XFCE from time to time, so I have that little thing in mind, but I do hope they switch to GTK3 soon or I'll have to abandon that one too. CSS just makes everything so much easier for a designer and I see no point in not supporting it fully.
And some thoughts about that Cinnamon desktop. It feels very amateurish as of right now. I have to configure it for over an hour to make it look the way I want, and even then it's not perfect. Gnome 3.8 may bring a "fallback" desktop that may be similar to the old Gnome setup. I wonder if that will do it better than Cinnamon.
I can always complain, but I suppose if I want something perfect I'll have to do it myself. I've started to read up on javascript so I can try to do something myself.
We'll see what comes next. In 10 years we may need to create interfaces for holographic displays.
