I haven't noticed a decline in the level and quality of reasoning and documentation that answerers have been providing since early May of this year, though such a decline is certainly possible.
What I have noticed is a tremendous improvement in the level and quality of documentation that answerers have been providing in the past 36 months compared with what they tended to provide in the period before that. I think that the proportion of answers that are backed by useful, confirmable citations to authoritative references—linked and unlinked—is much higher on this site today than it was three or four years ago.
It's not that the answers given in 2011 are incorrect or even (in general) less reliable. But I think that in the early months of this site's existence, the prevailing notion was that answerers who knew their stuff fulfilled their role more than adequately by answering questions briefly and correctly, without the need to cite any authority beyond their own expertise. (Some knowledgeable participants have cited authority and investigated questions in depth from day one, but others seem to have seen less reason to adopt so painstaking an approach.)
The greater stress on third-party authority in more-recent years is perhaps a reflection of the participation of so many non-expert answerers (like me): If you don't possess an academic background in linguistics and an intimate acquaintance with advanced grammatical analysis, you have to look things up to ensure that you aren't misapplying a rule or repeating a baseless factoid that you picked up at random somewhere. The worst answers propagate erroneous information; and some supposedly authoritative reference works undoubtedly are erroneous themselves on some points. But at least when you point to a source of information, you enable better-informed participants at the site to review it and challenge its conclusions.
As for knowledgeable answerers who don't provide documentation, I imagine that people who have complete command of the subject must find it extremely tedious to have to dig up a reference to something they consider elementary and obvious. I remember seeing, some months ago, a comment from an EL&U participant saying that we are lucky to receive contributions from some genuine experts, and that we should gladly accept those contributions on whatever terms the experts choose to provide them. I agree that their expertise is of tremendous value to the site, though I think it is even stronger when yoked to documentation.
In the end, what makes an answer on this site helpful for the poster and for subsequent readers is a combination of things, some of them not especially compatible: accuracy, clarity, succinctness, comprehensiveness, and documentation. The closer we get to producing answers that possess all of these qualities, the better we serve our community.
