Simulating User Watch-Time to Investigate Bias in YouTube Shorts Recommendations
Selimhan Dagtas, Mert Can Cakmak, Nitin Agarwal · 2025
Read Simulating User Watch-Time to Investigate Bias in YouTube Shorts Recommendations by Selimhan Dagtas, Mert Can Cakmak, Nitin Agarwal (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2507-04534.txt.
Description
Dagtas, Cakmak, and Agarwal studied this question in twenty twenty-five: how watch habits steer YouTube Shorts suggestions. Short videos shape what many people see, yet the effect of viewing behavior on the feed is poorly understood. The team simulated viewers who linger or skip at different rates. Their data covered over four hundred thousand videos.
The simulations followed chains of suggested videos on world politics and specific conflicts. A large language model judged how well each new video matched the starting topic. The team found patterns of drift, where topics slid away from the start, and of widening, where themes grew more general. Fast skipping and long watching led the feed down different paths. The study speaks to platform openness and content diversity.
Catalog record
| CREATOR | Selimhan Dagtas, Mert Can Cakmak, Nitin Agarwal |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 17 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2507-04534