There exists a fundamentally false assumption in centrally resposited information sources - specifically those that are erected by a vast community of semi-anonymous users. That is, they assume there is one possible explanation or at least a best or “optimal” explanation for a thing. The biggest proponent of this fallacy is of course, Wikipedia. The essential premise that it is authoritative by way of democracy or worse compromise.
Democracy is a poor measure of determining truth. Appealing to the masses is intuitively convincing but demonstrably wrong - this is especially true when attempting to isolate entirely subjective metrics such as “quality”, “worth” “beauty” etc. Wikipedia exercises this fallacy on occasion to settle disputes - putting to vote reality. This is the primary critique levelled at Wikipedia, that everybody often isn’t as smart as somebody when it comes to a specific subject.
Curiously, where democracy is needed it missing. The kinds of things I read on Wikipedia are, at best, what the author(s) of the document thinks of prime relevance to the subject - there exists little avenue for the demands of information seekers to be addressed; information flow is manipulated entirely by the suppliers.
Often Wikipedia articles diminish their own definitiveness by conceding to angry (but wrong) users. To allow information that is demonstrably untrue simply because sufficient quantities of people believe it dilutes the value of the affected articles. Articles on wikipedia are not consensus reality, they are compromise reality - they are the explanation of things that trigger the least violent emotional reaction in their readership.
The truth, as presented by Wikipedia is that endorsed by those valiant enough to partake in the discussion page and article editing voodoo and none else. Think of what this means for simple language barriers - factual reality is impeded by the mundane block of differing languages.
If one seeks the consensus reality - as is typical for scholars, one must be as comprehensive as possible yet discriminating at the same time. To calculate the bias of information sources as precisely as possible, so it may be least understood if not compensated for when summarizing or deducing. How is this possible when but a single version - or version lineage is allowed for a given subject?
Contrast this with say, the blogosphere. People’s perceptions of the world, assertions and beliefs are far better understood if one to aggregate all relevant blog entries on a given subject than if one were to read the appropriate wikipedia entry. This is enabled because each blogger maintains their own artefact of perception - they record it without interference or oversight. This allows extreme bias to enter very quickly, but when analyzing opinion or perception eliminating bias is undesirable - accounting for it is the goal.
What is needed, then, is an intermediary between the highly structured wikipedia - with its rigorous oversight, stringent quality standards and logical navigation scheme and the highly networked, cohesive and comprehensive blogosphere.
We need information neighbourhoods, not information kibbutzes if we are to understand the emergent properties of collective intelligence. Every person’s schema must be simultaneously partitioned and cohered to attain a properly normalized consensus.
To that end, the semantic web provides an avenue.