There's no "application incompatibility". What exists is a difference on dependences between distros. And that's a good thing, since it gives, for example, a chance to choose between linking your app to the most stable version of a library or to the newest and fastest one.
The problem will come only if the distribution makers are unable to link the application themselves. Then the developer has two choices: link statically (with an increased executable size) or present a list of libraries needed to run the app (like in the output of ldd) and nag the buyer to install them. Make them three: supporting individually each and every setup. In any case, the developer is paying extra for the privilege of keeping its code kind of hidden.
Now to the open standards.
Sun's argument is, I quote, "that open standards enable substitution, choice and competition". It would be easier to support their case if they showed how open sourced standards do not enable those three factors. I give to them that their product does enable them, but where is the proof or even the case for their competitor's standards not being open? Or what standards are they even talking about?
If I knew what standards are they talking about maybe I would present examples of "substitution, choice and competition" - three things that in my book resume to having several applications implementing them. And that by itself would not be needed, since a proof of the *possibility* of creating those several applications would be enough for open-standarity.
On a last note one thing they don't seem to be confused on is on the fact that "open source" is not the same thing that "Free software" (with capital F). I bet a dime and two bottlecaps to Sun using some restrictive licensing.