Would an “emotional momentum” signal actually interrupt doomscrolling?

I’m working on a browser extension idea that would show a local “emotional momentum” signal when a feed keeps moving toward more alarming, anger-provoking, or anxiety-inducing material while I continue scrolling. Not a diagnosis—more like: “Your recent feed has become more emotionally intense, and you’ve continued.”

I noticed during my own testing that I sometimes kept scrolling through increasingly agitating posts even after I’d stopped finding them useful. A small interruption might have helped, but it could just become another dashboard metric to monitor. Some people reasonably prefer urgency and relevance over emotional neutrality, especially with news.

The practical question is whether the calculation can stay local, using only coarse patterns like rapid scrolling, session length, and repeated exposure to high-arousal content. That would avoid sending this interaction history to the platform, though the extension would still need sensitive permissions. Would you trust, understand, or actually use a signal like this—or would it become part of the distraction?

I’d be careful not to imply the signal has been shown to work. The research supports emotion being used in recommendation systems, and one preprint reports anger reinforcement in a YouTube audit, but the user-facing intervention remains untested. I’d treat this as an experiment, not a wellbeing feature.

“Emotional momentum” sounds measurable until you try to define it. Rapid scrolling can mean boredom, accessibility behavior, or simply a long thread. I’d expose the inputs and confidence, or skip the score entirely. False precision is the failure mode here.

The wording matters more than the color or number. “The last ten minutes got more intense” invites a decision; “momentum: 78” invites optimization. I’d make it optional, quiet, and dismissible, with a pause control nearby rather than a second feed of self-monitoring.

Local processing helps, but it isn’t automatically private. An extension reading page content, clicks, and scrolling has a substantial permission surface. I’d want a narrow allowlist, transparent code, and no telemetry by default. Otherwise the cure creates a new observer.

Users may understand the signal perfectly and still ignore it. Relevance usually wins at the moment of choice. A useful test would measure whether people pause or change settings, not whether they say the feature is insightful in a survey.

I like the distinction between feed intensity and user emotion. The former is observable enough to report cautiously; the latter is an inference. That boundary should be prominent, because “you are anxious” would be an unjustified claim.

A local rolling window seems like the right architecture: coarse events, short retention, no raw post archive. I’d avoid trying to classify every item perfectly. Even a rough “more intense than your recent baseline” indicator could be useful if it admits uncertainty.

My concern is notification creep. First it says the feed is intense, then it asks whether I’m sure I want to continue, then it keeps score. Make it a browser button I can check, not a nagging conscience stapled to every page.

There’s a legitimate countercase: emotional intensity is sometimes the point. During a breaking event, a neutralized feed could hide urgency or make important reporting look suspicious merely because it is upsetting. The user should choose the threshold and whether the feature runs.

I’d ship the least ambitious version first: a local trend line over the last few minutes, no content labels, no universal score, and a one-click pause. If that changes nobody’s behavior, we learn something. If it works, add detail later.