At one level, the only thing necessary to improve Wikipedia's writing style is to make it a priority. We're good at those -- just look at the way we got rid of nonfree images.
The difficulty, of course, is that writing style is inherently a subjective issue. We're not nearly so good at those -- in the current, rule-bound climate, it's much easier to enforce objective standards.
Our rule-boundedness is relevant in another way, as well: though it's loved by pedants and petty bureaucrat wannabees, it's absolute death to the truly intelligent and creative writer/editors who could really make our content sing -- and in more areas than just a more-readable writing style.
So while there are good reasons for our tendency towards firmer and firmer policy -- there's no way a project that's as big as Wikipedia now is can get away with as much informal freewheelingness as it had when it was younger -- it's a trend which has to be intelligently resisted, lest the cure turn out to be worse than the disease (as of course there are signs that it already has).
I don't agree that copyediting is inordinately difficult. It's hard, to be sure, but then, doing research and selecting and accurately presenting facts are hard tasks, too. In one sense, copyediting Wikipedia is a highly desirable task, as it's one you can do from the comfort of your easy chair (just as much of Wikipedia used to be written), without all that tedious sourcing and referencing and second-guessing about notability. I know some people who are great at it, and it wouldn't take much to convince them to dig in and make massive improvements to any Wikipedia articles which could use it.
Isn't there a Wikiproject or something where people who like to copyedit can hook up with articles which have been requested for copyediting? If not, perhaps there should be.