Epistemic diversity and industrial selection bias

Abstract Philosophers of science have argued that epistemic diversity is an asset for the production scientific knowledge, guarding against effects biases, among other advantages. The growing privatization research, on contrary, has raised important concerns philosophers science, especially with respect to sources biases in research it seems promote. Recently, Holman and Bruner (2017) shown, using a modified version Zollman (2010) social network model, industrial selection bias can emerge community, without corrupting any individual scientist, if community epistemically diverse. In this paper, we examine strength reinforcement learning which simulates process decision-making when allocating funding projects. Contrary Bruner’s probability success agents performing action given priori, our model industry learns about rate scientists updates each round. results simulations show even previous knowledge still able disrupt consensus. fact, more diverse easier move consensus opposite conclusion. Interestingly, also shows having random agent effectively counteract bias. Accordingly, consider allocation projects as strategy bias, avoiding commercial exploitation communities.