Personal Echo Chambers

A study by Eugine Leung and Oleg Urminsky showed up in my feed today through Richard Hahn by way of Stefano Puntoni (thank you all!!). "The narrow search effect and how broadening search promotes belief updating", published in PNAS in March, examines how the formulation of a search query influences the search results (obvious so far) in a way that reinforces prior beliefs that may have shaped the query.

Too abstract? The caffeine example is an easy and compelling one: if the query is "What are the health benefits of caffeine?", which you would ask if you believed caffeine is good, you are likely to get search results that confirm your beliefs (good for alertness, mood, memory, stamina, etc.); similarly, the results for the query "What the health risks of caffeine?" will not change your dark views on coffee (cardiovascular risks, sleep, urination, addiction, etc.). The balanced query "What are the risks and benefits of caffeine?", which does not indicate a preconceived belief in either direction, will provide a more balanced perspective and would be useful to those who do have strong prior beliefs. The paper details clever study designs to disentangle various effects and clearly show the "narrow search" effects. I recommend looking at the supplementary materials.

Not in the paper, but one observation that has been made countless times: unfortunately, the folks who would benefit most from information that could update their beliefs are the least likely to even WANT to see it. They are likely to become very upset if any form of "broadening" was applied to their search behind the scenes.

Confirmation bias is one of the strongest forces that prevent humans from exploration and open-mindedness. There is an evolutionary heuristic aspect to it: if you are in a group of people that all share the same belief(s), you don't want to be the lone skeptic, it's exhausting and socially dangerous. It is less costly to conform and confirm. All this can become decoupled from facts and reality, a dynamic I described in a Harvard Business Review article more than 20 years ago. That dynamic has only become more potent since then, and now a runaway train in the age of sycophantic AI chatbots whose reason for being, it seems, is to figure out your beliefs to reinforce them.

Related article: link