Metascience Symposium Recap
Joe Bak-Coleman, a computational social scientist, studies how groups make decisions under uncertainty, with a special focus on how digital technologies shape collective behaviour. Speaking at the Metascience Symposium, he explained that his career began with research on swarming insects and schooling fish, but his curiosity about collective dynamics eventually led him to examine human behaviour—especially in online spaces.
In his talk, Bak-Coleman warned that much of mainstream meta-science suffers from decontextualization: abstracting away the specific history, norms, and goals of a field, then applying the same problems, solutions, or metrics universally. While this can seem efficient, it risks creating reforms that are ill-suited—or even harmful—to certain disciplines.
He contrasted the idealised modelling common in physics with the messier reality that psychology often embraces. Ignoring such differences, he argued, can produce overly prescriptive “fixes” to science, guided more by metrics like replication rates than by whether a field is genuinely advancing knowledge.
To illustrate, he shared a personal example from a large replication project in social psychology. His team was tasked with replicating the first statistical claim from a paper—a preliminary, almost incidental result—while overlooking the richer, more innovative parts of the research. The replication succeeded, but the exercise stripped away the study’s context and missed its core insights.
For Bak-Coleman, true meta-science must remain embedded in the content of science. Progress isn’t just about universal measures; it’s about understanding what each field has actually learned and how reforms are serving the people doing the work. Without that grounding, he cautioned, we risk reducing science to a flattened version of reality—much like judging an ant colony only by its first pheromone trail.
Author: Assemgul Kozhabek