Similarity in research activities increase the likelihood to generate little divers but related and thus easier economically exploitable knowledge, but it also increases the costs of double research activities. Diversity in research activities increase the probability to create diversified knowledge considered to be the principle driver of economic growth. But high diversity also means that knowledge possibly is less related and more difficult to exploit. Both similarity and diversity incur costs, which regulation measures aim to minimize. This paper models this tradeoff to shed light on the effects of benchmarks in basic science. A multi-actor model describes the outcome of inventive activities by researchers in basic science creating novel knowledge. Benchmarking has an impacs on knowledge diversity and similarity. A Monte-Carlo simulation suggests a long-run total cost curve depending on the degree of knowledge sharing among actors. A sensitivity analysis supports the robustness of the model. The results offer room for discussing possible policy implications.