Current data processing is suffering from the bane of the many data silos. Data is locked up in a hierarchy of containers and can hardly be connected in new ways. Once data has been grouped into relational database tables or object oriented classes, it’s difficult to regroup it or set up new relations. Regrouping would either break existing code or mean duplication of data. New relations would entail schema changes and be limited to connections between containers on only a few levels of abstraction. Products like Microsoft’s WinFS, dynamic languages and database refactoring tools are trying to overcome this lamentable state of data processing while having their own perspective on it.
This, of course, does not mean there is no value in current concepts and technologies any more. Relational databases and object oriented languages are very useful and will continue to be so for a long time. However, the general need to go beyond them to solve the data silo problem should be obvious.
At the heart of the data silo problem are the basic and mostly unquestioned concepts of data container and container reference.
For those of you who followed my musings on associative information processing or processing data by getting rid of data, here´s a new stab at explaining a completely different basic way of dealing with data. The above is an excerpt from a new paper I wrote on a concept called Pile (www.pilesys.com), which I find quite interesting (although still being in its infancy).