I think a lot of this depends on what you place the most value in. While agreeing that the things RMS/GNU/FSF value are worthwhile and important, I place a higher value on "quality" and wide adaptation.
You understate things a bit. In the view of rms every aspect of a software project is a sacrifice on the altar of the GPL, if necessary.
It makes sense to have that policy when your goal is eradication of copyright, which is a goal quite a few people do share with the FSF. The trouble is, that's a destructive goal, and software is a creative enterprise, so things are fundamentally at odds and weird results happen, such as the refusal to allow a modular structure in GCC to prevent undesirable uses. A strange definition of freedom indeed!
The FSF's goal in essence is copyleft. Copyleft requires copyright law in order to work. As such, the FSF certainly does not intend to eradicate copyright, very much the opposite is true.
I'd go even further than that. For RMS he'd rather a problem not be solved using computers at all, if solving the problem with computers would require or even just enable non-free software.
The most recent example of this was the debate over refactoring tools in emacs that used an AST generated by gcc. RMS was against this, even though all the code involved was free, because it would mean exposing the AST generated by gcc to external processes, some of which might be non-free. So he'd rather people not have refactoring tools at all, than enable the possibility of a non-free IDE existing that uses the AST from gcc.
