Many of the proposals to "fix" Wikipedia of late have seemed to take as a premise that what we've done is wrong. I, personally, disagree. I think we've got a pretty good encyclopedia. It needs work, but it's good enough to go public with, which, thank God, since we went public with it. Sensible users can use it well.
But if we really do want to speed up its improvement (which I can take or leave, but everyone else seems desperate to take it)...
Why don't we lock new article creation in the main namespace entirely for three months? Or six months? Demand that people fix existing articles.
Anything that's absolutely vital that comes into being in those months will still be possible to write about in a few months, so there's no real rush. And a lot of the crap that we create by reflex will not get created and be pleasantly forgotten about. (Brian Peppers, anyone?) And we could easily make the red page text read something like "On XX/XX/XXXX suspended new article creation until XX/ XX/XXXX in order to better work on existing articles. If this is an important topic that has developed since we made this decision, you can probably find information on it by looking at existing articles on related topics."
We've suggested doing it for a day here and there. The heck with that. Let's do it for a long period of time so that the culture of fixing what we have becomes entrenched.
Or, I mean, we could decide that everything we've worked on this far is actually crap and create drastic proposals for how we could start over.
-Phil