Philip Sandifer wrote:
We use attribution and verifiability because, in empirical fact, they are reasonably similar to truth. But in terms of actual use of Wikipedia as a resource, people depend on that isomorphism between accuracy and attribution. When that isomorphism breaks down, it poses a genuine problem.
A bit stronger than that, really. We use attribution and verifiability because the point of a Wikipedia article is to summarize external consensus on an issue, and attribution and verifiability are really the only ways of determining external consensus. They are often actually quite dissimilar to "truth", in that often we accurately summarize mutually contradictory information, such as two different academic fields' dissimilar takes on a subject, or religious and secular views of an issue.
More pragmatically, the reader has no particular reason to believe that a random Wikipedia editor (who could be anyone) has any particular personal insight into "the truth", so would prefer some sort of pointer to a more reliable source for potentially contentious statements, rather than "some guy on Wikipedia has investigated, and determined that all the sources are in fact wrong".
-Mark