Platforms are teaching people what gets rewarded

The strongest thread across this reading queue is not simply that platforms distribute culture. It is that platforms teach people what kinds of speech, work, refusal, and institutional design are likely to survive.

One preprint makes that mechanism unusually explicit: Petter Törnberg models the "attention price" of political expression across X/Twitter, Bluesky, and Mastodon, then asks which rhetorical forms each platform rewards. X/Twitter sharing rewards antagonism while penalizing respect and nuance; Mastodon boosting rewards reasoning, nuance, compassion, and collective expression; Bluesky reposting favors more neutral, lower-emotion language. 1 That gives this week's debate a useful frame: platform power works before a post is moderated, demonetized, or banned. It works at the point where users learn what gets attention.

The Törnberg preprint is useful because it refuses the usual shortcut: platforms do not only influence politics by exposing users to messages. They also shape the conditions under which political actors produce messages in the first place. If antagonism is a cheaper way to buy visibility on one platform, and deliberative language is less penalized on another, then "platform choice" is partly a choice of rhetorical economy. 1

That matters for media studies because it gives platform media logic an empirical handle. The paper does not prove that a platform causes a politician to become more antagonistic. Its claim is narrower and more useful: different platforms make different expressive choices more or less attention-efficient.

"Flattening the news" pushes a related point into journalism. On newsrooms' own sites, differences between outlet types remain visible. On Facebook, Instagram, and TikTok, those differences largely disappear. 2

The sharpest finding is about local news. Local outlets' civic and community-oriented work is especially vulnerable to platform selection pressures that favor softer topics. 2 That should unsettle any policy language that treats outlet diversity and content diversity as interchangeable. A country can have many outlets and still end up with a thinner public agenda once social distribution filters the work.

The MCN paper adds an important intermediary layer. Streamers do not only confront the platform's opaque algorithm. They confront managers, training documents, rankings, studio routines, and financial dependencies that translate algorithmic uncertainty into instructions for self-discipline. 3

The asymmetry is the point. Managers understand the algorithm as unstable and probabilistic; streamers are taught a simpler moral story in which effort, optimization, and professionalism should produce results. 3 That turns opacity into labor discipline. Nobody has to say "the system is fair" forever. It only has to be repeated often enough to make workers blame themselves when outcomes wobble.

Cai and Chen's study of copyright disputes shows a different kind of mediated platform governance. Chinese creators' self-narrative videos do more than complain about takedowns or infringement. They turn copyright conflict into public claims about originality, culture, identity, and the loss of platform neutrality. 4 The useful move here is to read creator resistance as cultural work, not only as a procedural appeal.

The non-use article is the quiet counterweight to the resistance papers. Young Korean adults who limit or stop using Instagram do not simply exit into freedom. Their non-use practices are pulled back by social expectations, platform-specific resources, interface habits, and the discomfort of being disconnected. 5

This is where "platformization" becomes more than a background word. If a refusal has to be constantly renegotiated because social life, cultural participation, and practical resources remain platform-bound, then non-use becomes part of the same system it tries to resist. The article's strongest contribution is not that leaving is impossible. It is that leaving has to be studied as a socially organized practice, with its own costs and dependencies.

A second preprint, "Seeing the Politics of Decentralized Social Media Protocols," compares ActivityPub, AT Protocol, Nostr, and Farcaster. It argues that decentralization distributes control differently across identity, curation, and infrastructure. 6

Read it beside Törnberg rather than after it. Törnberg asks how existing platform environments reward expression; Oshinowo, Hwang, Zhang, and Monroy-Hernández ask where power is placed when platforms are rebuilt around protocols. Together they make a useful test for the decentralization debate: not "is it centralized or decentralized?" but "who can decide identity, ranking, moderation, portability, and infrastructure failure?"

The common claim across these pieces is blunt: platform power is no longer located in one visible place. It is in attention incentives, topic selection, training scripts, copyright disputes, refusal practices, and protocol design.

That makes the next question harder. If power has moved into the conditions that make action feel sensible, where should critique intervene first: in ranking systems, worker intermediaries, cultural policy, user refusal, or protocol architecture?