Matt McAlister writes about problems with mashups and what they ought to be. Mashups are really an attempt for data owners to answer the question: what should we do with all this data? In the none mashup world, the applications and the data are to a large extent driven by the same business model and re-investment of revenue. Consequently, the investment in the data and the investment in the application is related and dependent. This is not the case in the mashup model in which there is a huge disconnect between the investment put in to the data and that put into the application. It is because of this disconnect that McAlister makes his observation:
[T]here's one element that seems to be missing from the mashup body out there: intelligence.
Mashups will deliver the occasional jem in which the fundamental concept behind the mashup transcends the need to invest heavily in the application, as long as it is well executed. However, in general this is not the case.
A corollary of this problem is the fact that there are so many mashups that do almost exactly the same thing. If each of these applications had also to deliver the data that enables them, there would be far fewer. This would result in less choice for the user and, consequently, an easier task of deciding which news/rss/mapping aggregator to use. I find it hard enough to keep up with posts announcing the release of a new news/blog aggregation site let alone actually spend significant time with them determining if I want to commit to them or not.
Don't forget: if you build a mashup, the data provider owns you.