The challenge, of course, is defining what "deprive others of that freedom" means. Does it mean you can't deprive other people of the freedom to have the source code to your work that extends the original work, or does it mean you can't deprive other people of the freedom to make private extensions to the original work? That's fundamentally the difference between the GPL and BSD licensees is what other group of people you want to deprive of freedom.
Arguably, the BSD license is more free because the existence of a private fork doesn't deprive anyone of anything; the original work is still freely available. But on the other hand, you could argue that some of those changes merely fix bugs, and thus are not rightfully new works, and should be available to anyone who has the original software. It's a fine line, and there's no absolute right answer.
The reason the public mocks nerds, of course, is that they argue vociferously over which license is correct, which takes time away from actually making the technology better, and is often seen as a waste of everyone's time. On the other hand, without those arguments (which expand the community's understanding of the licenses and their eccentricities), there's a possibility of critical projects choosing a license that is inappropriate and ending up stuck with it to the detriment of everyone.
For example, the FSF's decision to relicense GCC under GPLv3 created stagnation in its largest user base (the Mac community), with OS X users stuck at a much older version for years, until eventually Apple worked with the LLVM team to replace it with Clang. To be fair, in the end, everybody benefitted from a more modular compiler architecture that could better be integrated into things like IDEs, so the resulting platform is more capable than GCC ever was (or ever will be, in all likelihood), but the bad licensing decision meant that the teams couldn't take advantage of each other's work, which no doubt made that transition take much longer than it otherwise would have and resulted in a lot of duplication of work, ultimately culminating in GCC becoming an evolutionary dead end that's still a giant time sink to maintain (and that, no doubt, will continue to be maintained for many years, for no real reason other than because it exists and has to work).
So in spite of the public's belief that this is all a bunch of silly squabbles like Star Wars versus Star Trek, the reality is that there are real-world implications of these arguments, making them at least somewhat valuable (up to a point, anyway).