Fig. 3: Scaling of opinion consistency in synthetic worlds.
Mean opinion consistency, μC, and the standard deviation of consistency, σC, as functions of the number of subjects, N, for regular random networks with node degree z = 50 and various levels of structural noise, β. The symbols and error bars show the mean and three times the standard error of the mean, respectively, obtained by running the model on 1000 network realizations. The solid lines show μC(N) and σC(N) obtained by solving the master equation (Eq. (6) and Eq. (S6) in the SI). a, b Results for a single seed opinion. In this scenario, both μC and σC converge to zero as N grows: There is a tension between opinion consistency and the world’s complexity. c, d Results for a fixed fraction of seed opinions, fS, when the number of seed opinions is NS = fSN; here fS = 0.01. In this scenario, the tension between consistency and complexity is prevented: μC converges to \({f}_{S}^{2\beta }\) (c) and σC converges to zero (d).
