Jon confesses that he’s cheating: he wants to search for my quotes and wants to use XPath for that. Well, guess what? he’s approaching the limit of what XML can do for you. You would like to ask “give me all the quotes that came from Stefano’s blog” and, know what?, what you are using an implicit inferring model that is hardcoded in your markup. As I’m sure Jon is starting to note, that sounds hackier day after day.

That’s why I was so attracted by Simile and accepted the job: I started already to understand that XQuery was going to work well for querying multidimensional hierarchical data structures, but that the more dimensions (namespaces) you add, the more relational things start to become… and at this point, you find yourself asking for a different way to search this information.

In my quest for data emergence, XML was going to be, ultimately, a rock-solid wall. Even a humongous and superfast XML database with XQuery cream on top would not be enough to let data emerge (well, the kind of data I would want to see emerging). Jon wants to search for my name, not for my blog URI… well, if that simple of a problem shows already weaknesses in the approach, imagine what you will have to face when you don’t even know the schema of the data that it’s in your database and you have no control over it.

So, what Jon is doing is implicit semantic searching using XPath. It’s a first step and feels refreshing as the data looks much more morphable, but it’s not even close to be a real solution to the data emergence problem and to the ability to find hidden trends, for example, in second order effects of massive data bases. That is much more relational than a semi-structured (say content-oriented) XML document is.

So, how do we go from here? hell knows what’s going to work, but one thing is for sure: XML is not good enough.