Study finds growing social circles may fuel polarization

Methodology, Data Quality, and Causation Doubts

• Many commenters can’t access the paper (broken DOI) and are reluctant to trust a popular writeup without seeing methods or distributions.
• The headline claim that close friends doubled conflicts with other surveys on friendship and loneliness that show the opposite; some suspect a data-aggregation or definition issue.
• Several question using the average number of close friends; a skewed distribution (a minority with many friends) could raise the mean while many remain isolated.
• Skepticism that parallel trends (more “close friends,” more polarization) imply causation; multiple people argue this is at best a shared-cause story, not “friends → polarization.”

What Counts as a “Close Friend”?

• Strong suspicion that the meaning has shifted: people now count online-only or shallow ties as “close,” inflating numbers.
• Many distinguish between deep, in-person support (help with crises, physical presence) and digital “chat buddies”; the latter may not reduce loneliness and can even heighten it.
• Some note post‑COVID pruning of weak ties and intensification of a few relationships, which could raise reported “close friends” while making others friendless.
• Others note that technology lets old ties persist at low effort (group chats, Zoom), complicating any time-series comparison.

Social Media, Connectivity, and Polarization Mechanisms

• Strong consensus that social media and smartphones are a key common factor around 2008–2010, whether or not they act via “friend count.”
• Mechanisms discussed: algorithmic feeds optimize for engagement and outrage; exposure is skewed toward extremes; misrepresentation of “the other side” (perception gaps); drama is rewarded.
• Several argue that high connectivity plus ranking/voting systems creates huge, homogeneous online tribes that behave like “monsters,” driving real-world political conflict.
• Others emphasize economic and structural factors (financial crisis, housing, inequality, late-stage capitalism, information overload, foreign interference) as major co-drivers.

Centralized vs Client-Side Moderation and Ranking

• One major subthread blames centralized moderation and recommendation (social feeds, search, chatbots) for creating ever-larger, ideologically uniform groups.
• Proposed remedy: ban server-side ranking/moderation on large platforms; move filtering and ranking entirely client-side, with user-chosen or third-party algorithms (analogous to adblock lists).
• Pushback: most people won’t or can’t curate algorithms; scale and data volume make client-side ranking impractical; spam and abuse still require some server-side control; de facto, people would just subscribe to a few popular filters.
• Supporters counter that even partial decentralization would limit mob dynamics and restore individual control over exposure.

Friendship Graphs, Group Dynamics, and Polarization

• Several map this to network theory: denser graphs produce tighter clusters; more/better-matched friends → more homogeneous groups → stronger in-group norms and out-group hostility.
• Others see friend growth as a symptom: once giant homogeneous communities form, they supply more like-minded “close friends,” while weaker cross-cutting ties (neighbors, casual acquaintances) wither.
• Commenters reference older work on small-group conflict and Dunbar’s number to argue that expanding beyond a certain relational capacity naturally drives hierarchy, dogma, and “groupthink.”

Broader Diagnoses of Polarization

• Long conceptual list offered: fragmented realities; epistemic closure; outrage economies; moral absolutism; purity spirals; identity built around enemies; collapsing shared norms and identities.
• Multiple people note declining attention spans and text-based, dehumanized discourse (short posts, “dunking,” performative beefs) as making nuance and cross-tribal trust harder.
• The thread itself hosts heated arguments about “far right,” Nazis, and recent politics—used by some as a live example of how quickly discussions become moralized, existential, and polarized.