Combine a wiki with code assist, templates, syntax checking, relational technology, a query language, semantic mapping and Google Suggest. What do you get? An emergent language , that's what. A domain specific data entry language uniquely tailored to the information at hand through continued use and extension. The IDE features in the background are there to help in achieving consistency, but the user entering the data is really free to just enter it. This is the point. Make it easy to enter data and assist in making it valuable, ie. mineable. When it's hard or awkward to enter data, nobody enters it. In the background, the system constantly analyzes existing data entries for commonalities to be exploited as structural information. These are then used in code assist proposals to further their consistent use. Code assist could also offer templates for common constructs. Key elements to recognize in the data would be traditional relational concepts: attributes and associates. In...
Literate testing, JCite, Java, Excel for Java, Mercurial, and whatever else seems to long for a tweet.