Kaleda Denton was cooking a turkey when the idea struck: what if people don’t average opinions—they just follow the crowd? That simple question, sparked by a holiday dilemma, has now upended a decades-old model of how humans conform. In a groundbreaking study published in Proceedings of the National Academy of Sciences, Denton, along with Marcus Feldman of the Santa Fe Institute and independent researcher Jonathan F. Johannemann, has shown that people are far more likely to align with popular opinions than to compute an average—challenging the long-dominant DeGroot model of opinion dynamics. This shift isn’t just theoretical; it’s been rigorously tested across five real-world scenarios, and the results are hard to ignore.
For years, the DeGroot model has served as the gold standard in social science, assuming individuals update their beliefs by averaging the views of those around them. But that approach falters when faced with outliers—like the one friend who insists on roasting a turkey for 33 minutes per pound while five others say 15. Averaging those answers gives 18 minutes, a compromise that may make little sense in practice. Denton’s model, by contrast, treats conformity as a clustering behavior: people gravitate toward the most common opinion, not the mathematical mean. And when tested against empirical data, her model consistently outperformed DeGroot’s, even when constrained to use fewer free parameters.
The team evaluated both models using five distinct datasets reflecting real human decision-making. In each case, the conformity model provided a better fit, particularly in situations with limited data—where the DeGroot model’s sensitivity to outliers becomes a critical flaw. "The DeGroot model is really affected by outliers," Denton explains, "but the person who gave the outlying answer may have just misunderstood the question." This resilience in sparse data environments makes the new model especially valuable for understanding fast-moving social phenomena, from emergency responses to viral misinformation.
Beyond immediate applications, the implications ripple through fields like cultural evolution, public health, and network science. The researchers tested the model on various social structures—complete networks, static networks, and adaptive ones where connections shift based on accuracy—and found it robust across all. In adaptive networks, individuals who updated their connections to align with lower-error peers saw faster convergence on accurate beliefs, suggesting that social learning is not just about popularity, but about smart conformity.
Feldman believes this work restores conformity to its rightful place in the study of cultural change. "These papers place conformity in a central position in studies of cultural change," he says. As social scientists begin to re-examine long-standing conclusions through the lens of Denton’s model, one thing becomes clear: sometimes, going with the crowd isn’t mindless—it’s rational.