Niche vs Mainstream

February 18, 2026·34 min·2 clips

Recommender systems can discriminate by showing women lower-paying jobs and burying niche creators under mainstream content.

The central question is structural. Recommendation usually lives inside the platform, so the host asks what happens when the platform and the ranking algorithm are split apart. SMORES is the tool for testing that idea. The guest describes the Simulator for Modular Recommendation Ecosystem as a way to study how decoupled recommenders might behave before anyone treats the idea as obvious. The app-store analogy lands cleanly. Netflix-style recommendation stops looking like fixed plumbing and starts looking like one possible choice among different recommenders. The experiment compares niche and mainstream recommenders. The niche recommender concentrates on a specific movie genre. The mainstream recommender follows wider demand across the platform. Utility is what makes users move. When people stop seeing the genres they want in a recommendation list, their utility falls; once it drops below a set threshold, they pick another recommender. The episode does not oversell the result. Users who move toward recommenders that fit their taste see better utility, and niche providers can benefit when their content becomes easier to find. Provider welfare gets real attention here. The guest notes that providers buried by a mainstream recommender may do better when a recommender is built around their kind of content. The ending turns toward user agency in smaller online communities, including Mastodon instances and subreddits, where users may care about what appears, how it ranks, and what counts as important. The tone stays empirical. Instead of declaring open recommender markets better, the conversation asks how to measure them.

As heard by us

A measured, research-minded look at modular recommendation systems and the case for more algorithmic choice.

Data Skeptic uses recommender systems to ask a clean product question: what if the algorithm were not owned entirely by the platform? The episode frames today's Netflix-style model as an in-house default, then explores a more modular future where users or communities might…

Read the full review in PlayNext →

Why you'd press play

If you want the politics of feeds with a technical edge, press play.

Read the full recommendation in PlayNext →

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from Data Skeptic