# Epistemic Bubbles in Adaptive Social Networks

**URL:** <https://forum.comses.net/t/epistemic-bubbles-in-adaptive-social-networks/11282>\
**Category:** Model Discussion\
**Created:** [November 16, 2026, 2:46am UTC](https://forum.comses.net/t/epistemic-bubbles-in-adaptive-social-networks/11282 "2026-11-16T02:46:07Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![omar173ahmed](https://forum.comses.net/user_avatar/forum.comses.net/omar173ahmed/32/2031_2.png) [@omar173ahmed](https://forum.comses.net/u/omar173ahmed)\
**Post date:** [November 16, 2026, 2:46am UTC](https://forum.comses.net/t/epistemic-bubbles-in-adaptive-social-networks/11282/1 "2026-11-16T02:46:07Z")

</div>

# Epistemic Bubbles in Adaptive Social Networks _(1.0.0)_

How can people come to agree with their neighbors while remaining divided across a social network?

Epistemic Bubbles is an exploratory NetLogo model in which people revise their beliefs and change their social ties. Agents start in a randomly connected network. At each time step, they move toward their neighbors’ average belief according to their openness to influence. A tie can dissolve when disagreement exceeds either agent’s tolerance. Agents can also form new ties with people whose beliefs fall within their tolerance.

To try the model, open Epistemic\_Bubbles.nlogox in NetLogo 7.0.4. Click SETUP and GO to run the saved settings. Then set mean-tolerance to 0.15, click SETUP again, and rerun with the same seed. Compare the number of connected components, the spread of beliefs across the population, and the average belief gap between people who are directly connected.

The model was developed as a final project for a Summer 2026 course on agent based modeling.

## Release Notes

Initial release of Epistemic Bubbles in Adaptive Social Networks. Includes the NetLogo model and documentation with settings and expected results for two reproducible runs.

Open “Epistemic\_Bubbles.nlogox” in NetLogo Desktop 7.0.4. Set the desired parameters in the Interface tab, click SETUP, then click GO. The model uses the bundled `nw` extension and generates its initial beliefs and social network internally. It requires no external input data. The model reports belief variance, the number of connected components, mean degree, and the mean belief gap between directly connected agents.

## Associated Publications

* * *
This is a companion discussion topic for the original entry at [https://www.comses.net/codebases/4a63f201-cfcd-40d7-a0bb-e860c97c8f63/releases/1.0.0/](https://www.comses.net/codebases/4a63f201-cfcd-40d7-a0bb-e860c97c8f63/releases/1.0.0/)

---

<div class="post-metadata">

**Author:** ![isabellaanderson](https://forum.comses.net/letter_avatar_proxy/v4/letter/i/9fc348/32.png) [@isabellaanderson](https://forum.comses.net/u/isabellaanderson)\
**Post date:** [November 18, 2026, 7:47am UTC](https://forum.comses.net/t/epistemic-bubbles-in-adaptive-social-networks/11282/2 "2026-11-18T07:47:28Z")

</div>

Thanks for posting this. One design detail worth clarifying: when agents form new ties, do they search the whole population for someone within tolerance, or only friends-of-friends? Global random search would let beliefs homogenize quickly regardless of network structure, while local search would make the component structure do more of the work. Also, is tie dissolution symmetric — does one agent’s low tolerance end the tie for both?
