Wikis provide this property because any redundancy could be reworked (though see ReworkingProblems).
One particular kind of redundancy exists when the writers of similar information do not know about each other. If the information is a Wiki, sooner or later someone will add links connecting the two isolated communities.
Redundant comments can be merged.
Redundant topics can be merged.
If the whole web were a wiki, redundant websites would be merged. No more would you hunt for a certain topic and find 200 web sites, many of which duplicate the same information. You would search the WWW (world wide wiki) and find 200 WikiPages, many of which point to the single canonical WikiPage dealing with your query. There would be 20 duplications of the same information, not 200, but more importantly most of these duplications point to the canonical page that you are looking for.
(not that a WorldWideWiki? would be better than the WWW; it is good that the web provides other modes of information than a Wiki)
Benefits of LessRedundancy:
(re: "redundancy isn't automatically lessened" in the change summary field; no, but overall I think there will be less redundancy in a wiki-based system as compared to if the same information were maintained in a read-only WWW based system; that is to say, I expect that as wiki catches on, there will be less redundancy in the future than there is in the WWW today (in relative terms; in absolute terms, there will be more information and hence more redundancy) -- BayleShanks)
LessRedundancy is similar to OccamsRazor. Occam's razor applies to ideas, to ontological objects, whereas LessRedundancy applies to written representations of information.