Citation Agents: A model of collective learning through scientific publication 1.0.0
Simulates the construction of scientific journal publications, including authors, references, contents and peer review. Also simulates collective learning on a fitness landscape.
Generates distributions, including power law distributions for papers per author and citations per paper, as seen in bibliometric data. Generates networks, including co-authorship and collaboration. Performs heuristic search on a Kauffman NK fitness landscape, effectively making the scientist authors a form of swarm optimisation.
Described in the paper:
Watts, Christopher & Nigel Gilbert (forthcoming) “Does cumulative advantage affect collective learning in science? An agent-based simulation”, Scientometrics.
Release Notes
This version was used to generate the data behind all charts included in the Scientometrics paper.
A later version, which runs in Netlogo 5.1, was developed for our 2014 book, “Simulating Innovation”. Download it from here:
http://www.simian.ac.uk/resources/models/simulating-innovation
A version in Excel 2003/VBA is also available from the book’s site.