More to the point, the sites that are big enough for this to even be an issue (such as the cited Google Docs) have a business interest in keeping that code proprietary. If you drew a Venn diagram of sites complex enough that minification prevents tinkering and sites with any likelihood of GPLing that code, you'd have two independent circles.
I just want to point out though, that optimized & minified code isn't really that different from say -- a blob of JVM code. JVM can be de-compiled, and sort-of be brought into a human readable form. This is a tab bit harder with x86 - but we're forgetting obfuscators here. Anyone who really don't want you to peek into their code will use an obfuscator.
Ultimately though -- I think the defining aspect of whether X code is open or closed, is the a measure of the ease with which I can take that code and modify it (perhaps make some enhancements to it). This is either difficult or sometimes practically impossible with x86, obfuscated JVM, obfuscated and/or minified JS, etc.
You seem to be arguing that because X is worse than Y, we shouldn't worry about Y. Or that because Y is hard to change, we should ignore its problems. The FSF will never agree with that, though.
I joined the FSF right about the same time. When it comes time to renew my membership, sadly what comes to mind is not the good things that the FSF works towards.
It is stuff like this that makes me seriously question whether I should continue to give money to the FSF money or if perhaps there is someone else who might make better use of it.
Edit: strange, no I can see the article. Maybe a bug with the URL before or some kind of site hiccup.