Wikis suck because they (as of yet) are never worth the additional complexity over regular web pages. They have a special syntax that can’t quite do exactly what web pages and links can do. They make some things easier, but others harder, the primary example being linking to a page within the wiki and a page “external” to the wiki.
Another major problem is that they are almost always focus on using templating languages rather than complete programming languages. This makes working with data sources (json, xml, recent statistics) a pain because the document is dead as soon as it is written. There is no way to express template generation within the language itself.
In order to overcome this wikis would need to be much more of an ecosystem.
Ability to create interactive visualizations that link to historic as well as current data.
At the bottom it should all be compiled into “regular” web pages so no special software would be necessary for hosting or interoperating with other static pages.
Rather than collecting words on a page, as in an encyclopedia, people using this kind of tool would collect data transformations, references, aggregations, and higher order functions related to simplifying these things.
Daniel X's Blog 