Modelling Science

The study of ontologies has the potential to teach us a lot about everyday science. Ontologies constitute a formal model of common knowledge in a given domain – e.g. Gene Ontology contains the concepts and relationships between concepts that are needed to describe gene products (Bada et al., 2004). The purpose of ontologies is often fairly prosaic: it is mainly about mapping data between databases. Nonetheless, the construction of ontologies is likely to be as fundamental to contemporary research as was the formulation of taxonomies in the times of Linnaeus. The study of ontologies, and more in particular their evolution, is greatly facilitated by the fact that many of them are developed and distributed on-line – often
using the same tools and platforms that are also used for the development of open source software. Hartung et al. (2008) have started to mine this source of information. So far they have aggregated statistics on the overall growth of ontologies and developed a tool that makes it possible to retrieve the full trace of the naming and linking history of individual concepts, but what is still needed is a method that allows for a visual characterization of the structural properties of ontologies as a whole.
In order to fill this gap I propose to use design structure matrices. In technical
systems, these matrices describe the dependencies between components.
MacCormack et al. (2008) have used these matrices to show that the distributed organization of open source software development is reflected in the structure of the resulting source code. The mirroring hypothesis, which they thus corroborate, is also known as Conway’s law. Yet, I suspect that design structure matrices may also shine a light on a law that is more closely related to the concerns of science studies – Stigler’s law of Eponymy.
Stigler’s law of Eponymy states that “[n]o scientific discovery is named after
its original discoverer.” In the context of ontologies the equivalent of this law is a statement about provenance such as “no concept of general importance is associated with its original submitter.” While the validity of this claim would seem unlikely at first sight as collaboration software is able to trace everything, it might still hold due to issues to do with the complexity of the ontology as a technological design. The fact that there is only a limited amount of ideas to which scientists can devote their attention influences the organization of science (Klamer and Van Dalen, 2001).
Consequently the structure of the knowledge of a given scientific community is likely to exhibit a certain degree of modularity – which is probably reflected in the structure of ontologies. There are conflicting hypotheses about the ease and frequency with which components are exchanged in such a modular structure (Von Krogh et al., 2009). Murmann and Frenken (2006) argue that highly connected components will be hard to exchange while Gosain (2007), seemingly in contradiction, argues that changes to components are very likely while the abstraction of the component
remains stable. If we think of the concepts in an ontology as components in a system, following Murmann and Frenken we would expect that concepts become stable as they become embedded in the system – a feat that is unlikely to happen straight at the first submission – and following Gosain we would expect that it is not so much the actual concepts that will be widely referred to in the scientific community as a whole but some sort of Platonic abstraction which again is not necessarily associated with the original submitter or discoverer of underlying concept. 2

References
[Bada et al., 2004] Bada, M., Stevens, R., Goble, C., Gil, Y., Ashburner, M., Blake, J. A., Cherry, J. M., Harris, M., and Lewis, S. (2004). A short study on the success of the gene ontology. Web Semantics: Science, Services and Agents on the World Wide Web, 1:235–240.
[Gosain, 2007] Gosain, S. (2007). Realizing the vision for web services: Strategies for dealing with imperfect standards. Information Systems Frontiers, 9(1):53–67.
[Hartung et al., 2008] Hartung, M., Kirsten, T., and Rahm, E. (2008). Analyzing the evolution of life science ontologies and mappings. In Data Integration in the Life Sciences, pages 11–27.
[Klamer and Van Dalen, 2001] Klamer, A. and Van Dalen, H. P. (2001).
Attention and the art of scientific publishing. SSRN eLibrary.
[MacCormack et al., 2008] MacCormack, A. D., Rusnak, J., and Baldwin, C. Y.
(2008). Exploring the duality between product and organizational architectures: A test of the mirroring hypothesis. Technical report, Harvard Business School.
[Murmann and Frenken, 2006] Murmann, J. P. and Frenken, K. (2006). Toward
a systematic framework for research on dominant designs, technological
innovations, and industrial change. Research Policy, 35(7):925–952.
[Stigler, 2002] Stigler, S. M. (2002). Statistics on the table. Harvard University Press.
[von Krogh et al., 2009] von Krogh, G., Stuermer, M., Geipel, M., Spaeth, S., and
Haefiger, S. (2009). How component dependencies predict change in complex technologies. In Proceedings of European Academy of Management Conference.