In this paper we offer a way to measure the knowledge base of an economy in terms of probabilistic entropy. This measure, we hypothesize, is an indication of the extent to which a system, including the economic system, self-organizes. In a self-organizing system, interactions between dimensions or subsystems will unintentionally give rise to anticipations that are properly aligned. The potential reduction of uncertainty can be measured as negative entropy in the mutual information among three (or more) dimensions. For a knowledge-based economy, three dimensions can be considered as key: the distribution of firm sizes, the geographical locations, and the technological classifications of firms. Based on statistics of these three dimensions and drawing on a unique dataset of all Dutch firms registered with the Chambers of Commerce, we are able to refine well-known empirical findings for the geographical dimension. Counter-intuitive, however, are our empirical findings for the dimension of technology. Knowledge diffusion through medium-tech industry is much more important for a localized economy than knowledge creation in high-tech industry. Knowledge-intensive services tend to uncouple economic activities from the regional dimension.