{ "blyg": "0.3", "id": "0q0wk1ecfwh0r3xmf2v8r620wc", "kind": "fragment", "origin": "https://www.msweet.net/notes/", "page": "129-bronze-eels", "author": { "name": "Matthew McDowell-Sweet", "url": "https://www.msweet.net/" }, "created": "2026-10-19T15:40:47Z", "updated": "2026-10-19T15:40:47Z", "version": 1, "content_md": "Shortform video has a genre I’ll call *reaction to X*. It consists of a viewer, *Z*, watching a creator, *Y*, react to something, *X*.\n\nHere’s how it works. The creator, *Y*, documents themselves reacting to *X*. *X* is usually a flash cultural moment, commonly sports or music-related. Sometimes only *Y* is visible, sometimes *X* plays synchronously in view while *Y* reacts, sometimes *X* is played then paused for *Y* to react, and sometimes there are compilations of many *Ys* reacting to a single *X*. *Z*’s engagement, measured or explicit, signals enjoyment. This [boosts the instance of *reaction to X* algorithmically](https://www.eugenewei.com/blog/2020/8/3/tiktok-and-the-sorting-hat), which means more people learn that *X* caused *Y*’s reaction, and that becoming a *Y* reacting to *Xs* is an attention-generating mechanism they can leverage.\n\nThus we have a [positive feedback loop](https://www.eugenewei.com/blog/2020/9/18/seeing-like-an-algorithm). New *Ys* copy what has already worked, so they converge on proven *Xs*, and attention concentrates on a smaller and smaller set of them. Meanwhile the *Xs* themselves are barely aware of the *Zs* at all.\n\nThis existed as a cultural phenomenon long before shortform video. Twitter’s [quote tweets](https://www.eugenewei.com/blog/2021/2/15/american-idle) are an example. This is what tabloids did, as well. But the loop has a final turn: once enough *Ys* exist, some *Xs* start being made *for* them, engineered to be reacted to. Which is why it’s funny to see the pattern re-emerging whenever one of the frontier labs drops a thud-piece attacking some apparently critical civilisational element, like [employing juniors](https://www.anthropic.com/research/expectations-for-the-future). Within days the *Ys* arrive, like [this reading of Anthropic’s take on the changing nature of human work](https://surfingcomplexity.blog/2026/10/18/anthropic-and-the-changing-nature-of-work/).\n\nThis can be both an anti-pattern for the average quality of information in the ecosystem, and a perverse incentive for producers to design for reaction: a subclass of engagement baiting, solving for heat over light. Yet its recurrence across mediums suggests it’s a feature rather than a bug, waiting for the right evolutionary niche.\n\nIt’ll be interesting to see how [Blygger](https://blygger.org/) resolves it. Questions around quoting, [mentions](https://blyg.blygger.org/t/7eeh3bfzfcnhh90d5pkvkc2r1n/), [comments](https://blyg.blygger.org/t/2mzqh4jt3ye14xm6sr5gf32cah/) and notifications are emerging precisely because it makes *reaction to X* a first-class primitive, rather than a second-class effect. But it may already have answered the hardest part by design: no public feed, no algorithm, no counts of any kind, and only verified citations notify anyone. Strip out the *Z* metrics and the loop loses its engine.\n\nWhat emerges in its place? I guess we’ll find out.", "content_html": "

Shortform video has a genre I’ll call reaction to X. It consists of a viewer, Z, watching a creator, Y, react to something, X.

\n

Here’s how it works. The creator, Y, documents themselves reacting to X. X is usually a flash cultural moment, commonly sports or music-related. Sometimes only Y is visible, sometimes X plays synchronously in view while Y reacts, sometimes X is played then paused for Y to react, and sometimes there are compilations of many Ys reacting to a single X. Z’s engagement, measured or explicit, signals enjoyment. This boosts the instance of reaction to X algorithmically, which means more people learn that X caused Y’s reaction, and that becoming a Y reacting to Xs is an attention-generating mechanism they can leverage.

\n

Thus we have a positive feedback loop. New Ys copy what has already worked, so they converge on proven Xs, and attention concentrates on a smaller and smaller set of them. Meanwhile the Xs themselves are barely aware of the Zs at all.

\n

This existed as a cultural phenomenon long before shortform video. Twitter’s quote tweets are an example. This is what tabloids did, as well. But the loop has a final turn: once enough Ys exist, some Xs start being made for them, engineered to be reacted to. Which is why it’s funny to see the pattern re-emerging whenever one of the frontier labs drops a thud-piece attacking some apparently critical civilisational element, like employing juniors. Within days the Ys arrive, like this reading of Anthropic’s take on the changing nature of human work.

\n

This can be both an anti-pattern for the average quality of information in the ecosystem, and a perverse incentive for producers to design for reaction: a subclass of engagement baiting, solving for heat over light. Yet its recurrence across mediums suggests it’s a feature rather than a bug, waiting for the right evolutionary niche.

\n

It’ll be interesting to see how Blygger resolves it. Questions around quoting, mentions, comments and notifications are emerging precisely because it makes reaction to X a first-class primitive, rather than a second-class effect. But it may already have answered the hardest part by design: no public feed, no algorithm, no counts of any kind, and only verified citations notify anyone. Strip out the Z metrics and the loop loses its engine.

\n

What emerges in its place? I guess we’ll find out.

", "content_hash": "sha256:1f81b704f4a64babce5e38fbe639438d4bcc6eed81fb50e1d94e8afec86363c2", "media": [], "changelog": [ { "version": 1, "at": "2026-10-19T15:40:47Z", "note": null } ] }