In Epistemic Networks, Is Less Really More?

Abstract

We show that previous results from epistemic network models (Zollman, 2007(Zollman, , 2010Kummerfeld and Zollman, 2015) showing the benefits of decreased connectivity in epistemic networks are not robust across changes in parameter values. Our findings motivate discussion about whether and how such models can inform real-world epistemic communities. As we argue, only robust results from epistemic network models should be used to generate advice for the real-world, and, in particular, decreasing connectivity is a robustly poor recommendation.

Conformity in scientific networks

“…30 More generally, the likelihood of the various possible outcomes tends to vary across network structures. These effects are complicated because network structure influences the likelihood that an epistemic network arrives at true and false conclusions even without conformity (Zollman, 2007(Zollman, , 2010aHolman and Bruner, 2015;Kummerfeld and Zollman, 2015;Rosenstock et al, 2017), and that same structure simultaneously affects the influence of conformism. To tease these effects apart, consider various network structures under a very low conformity value and a very high one.…”

“…The actors involved will all come to correctly believe that action B is better, or else to incorrectly believe that action A is better. The parameter values of the model and network structure will determine how likely these two outcomes are (Zollman, 2007(Zollman, , 2010aRosenstock et al, 2017).…”

“…Previous authors have delved into the usefulness of network epistemology models, considering robustness of results especially. See Frey andŠešelja (2017a,b);Rosenstock et al (2017). As they point out, sometimes small changes in these models lead to significantly different results, raising questions for real world applicability.…”

Endogenous Epistemic Factionalization: A Network Epistemology Approach

“…In simulations of the model as described so far, with all agents treating all evidence to which they are exposed identically and using Bayes rule to update their beliefs, groups of agents always end up in a state of consensus. That is, either all agents come to have a very high credence that action B is, in fact, better, or else, all agents come to falsely believe action A is better, and because they stop testing B they never adopt more accurate credences (Zollman, 2007(Zollman, , 2010(Zollman, , 2013Rosenstock et al, 2017). In other words, because of the mutual influence agents have on each other, stable polarization does not emerge.…”

Scientific Polarization

“…discuss subtleties of how groups can polarize.27 The probability of correct versus incorrect convergence varies based on parameter values. SeeZollman (2007Zollman ( , 2010;Rosenstock et al (2017) for more.…”