Cognitve Complexity: A Suggestion for Improving XML Designs

Perhaps this article can inspire some thinking about "bad XML design".  
  The last article I posted concerned the syntax itself.  This one  
points to the challenges of abstraction and context as well as the  
bottom-up-ness of implementation (which I find in FOSI implementation  
is the right way to do it so one might contrast XSLT design with FOSI  
design).

http://cacm.acm.org/magazines/2011/8/114944-cognitive-computing/fulltext

So the question:  how can XML design and implementation improve  
distributed systems for cognitive computing?

Try not to use this immediately to market RDF and the semantic web.

len