What you miss is that in a volunteer environment, the game isn't zero-sum.
I'm a kde user, but I'm VERY glad gnome is there, because were it not there, arguably a good 80% of the effort put into gnome would simply disappear from the desktop, since kde doesn't fit the gnome style of "there's only one right way and it's ours, so no configuration option could possibly be needed since we're already doing it the only possible right way, and if you think differently, you're simply wrong."
What's worse, probably 80% of the other 20% would bring that "only one right way and it's ours" attitude with them, and disrupt kde's "people are different and work differently, so let's make it an option so if they don't like the defaults, they can change them" philosophy. If they won that debate, it would kill the biggest reason a lot of people use kde, especially those like me who have never seen a default desktop they were happy with and don't expect to ever do so either.
Obviously that's painting with incredibly big and inaccurate brush strokes (as well as assuming they'd all end up with kde for simplicity of argument, they wouldn't, but with your argument take to it's logical conclusion there'd only be one left to switch to) and is thus simplistic at the level described above, but the point remains, there's all these different projects because people are volunteering their time and effort to what they're interested in, and if the number of projects was somehow artificially constrained, it would actually result in very little more work getting done on the remaining projects, both because people would simply find other things to do with their (after all volunteer) time, and because if they didn't, it'd dramatically increase the friction factor on the remaining projects as people spend all that gained time fighting with each other instead of working on their own separate projects.
So as a kde user who LIKES all those extra options, since they allow me to configure a desktop that I actually enjoy, I'm EXTREMELY glad all those "only one right way and it's our way" gnome folks have their own project to work on, and can leave kde to its "confusing mess of options". =:^)
But there certainly are people who "just want the defaults to work so I don't have to mess with it", and others who get horribly confused by all those extra options. For a good portion of both of those sets of people, gnome's going to be a more appropriate option than kde, and that's fine. I wouldn't want to take that away from them, and since all those folk putting so much energy into gnome wouldn't actually do a lot to benefit kde if somehow it became the only viable alternative anyway, it's not as if it hurts kde much, either.
In fact, the alternate implementations give people a chance to see the same idea from another perspective, and the competition is useful and healthy.
Meanwhile, there's the freedesktop.org compatible standards effort, where it /does/ make sense to work together, which is why X-based apps can ship a common *.desktop file that "just works" with both major desktops and several lighter alternatives as well, among other things.
The same thing of course applies to distros. I'm a gentooer, again, because I like having that extra level of control over my computer and the OS it runs (while still automating most of it once the choices are made). But it's not for everyone. But if all those folks working on all those distros were happy working on the same distro, don't you think that's what they'd be doing? It's volunteer (either of the distro devs directly or of the people paying them), and they work on what they believe in. Try to force them to all work on the same distro and it's simply not going to work, they'll find other stuff to do with their time/money, and to the extent that they don't, most of the would-be added effort will be spent simply overcoming the extra friction of all that extra disagreement, so that single one or a very limited few distros won't be much better for it after all, and the community as a whole would be a whole lot smaller and poorer as a result.