When bootstrapping knowledge-rich specialized corpora from the Internet re-ranking documents by their density of Knowledge Patterns (Meyer, 2001) has been shown to be a reliable indicator of knowledge richness (Agbago & Barrière, 2005). However, it is an open question whether all KPs are equally useful in assessing the richness of specialized knowledge in a document, and whether KPs are universal across domains and text types. The presentation will introduce a means of measuring the usefulness, or "weirdness" (Ahmad, 1993), of individual KPs by comparing their distribution in a general language corpus with that of specialized corpora. It will also outline a pilot study in which web documents (representing the domain of Information Technology) are re-ranked by uniform and weirdness weighted KP densities. Evaluation is carried out by comparing the densities of knowledge-rich contexts in the two corpora based on a comprehensive list of IT terms.