I have also seen articles decline in quality, but I wonder how big a problem this is. What fraction of articles does this apply to? And what fraction of articles are getting better at the same time? And when an article declines, does it stabilize at some quality level? If so, what is that level? Perhaps one even has something more complicated going on, like a slow overall increase in the level of an article, but with significant short-time fluctuations, just like global temperatures? I read wikipedia extensively, both at work and for fun, and at least in the topics I visit, the average article quality is very high. And in my fields of expertise, the error rate is also very small. I think this indicates that either the fraction of articles that tend to decline in quality are very small, or that the level at which the quality stabilizes is very high.
It's a fundamental problem for them, but one which they can do little about without changing their most basic policies.
I think it's also the reason for Wikipedia's success: More articles recruit more editors, which leads to more articles, etc.. Its predecessor, Nupedia, was written more according to your wishes, but because of its strict focus on experts and quality, it never got the network effects going that have driven Wikipedia's enormous growth. Wikipedia's success, and both in the number of articles and the quality it has achived (and its quality is, on average, pretty good) is quite the miracle, and if you had asked me, or most others, whether "an encyclopedia anyone can edit" could work, I think the answer would have been that it would get bogged down in trolling and sabotage. I guess most people are more constuctive than we give them credit for, and the "armies of editors" approach seems to be a very good strategy.
Lately, Wikipedia's balance seems to have shifted away from the initial inclusionism ("allow imperfect and incomplete articles, someone (not necessarily the same person) will improve them and add sources later") towards deletionism ("if an article isn't good enough (yet), delete it; if information isn't sourced (yet), delete it"). While the intentions behind this is good, namely getting more reliable articles, I think it might be counterproductive, as aggressive deletion policies probably hurt editor recruitment, and hence lowers the pool from which expertise can be drawn. I speculate that part of the reason for the slowdown in Wikipedia's growth the last few years might be this deletionism trend, though the fact that many important topics already have articles probaby is more important.