Should Linux have a local attention budget for apps?

I keep wondering whether an operating system should let me set an attention budget per application: perhaps five interruption requests, badges, or notifications a day, enforced locally. Not a wellness score, streak, or cloud profile—just an access-control rule I choose, like a firewall for my attention.

Linux already has notification urgency levels, but the standard does not cover everything that grabs focus, such as modal prompts or window decorations. A real quota would need to be explicit about what counts, show why an app is asking, and let me inspect which requests consumed the allowance. Calls, alarms, accessibility functions, and trusted contacts probably need emergency exceptions. Shared family devices make per-user profiles or handoff modes important too.

This might remove some of the triggers behind doomscrolling and always-online pressure, but it cannot stop me opening an app deliberately. Would you trust an OS-level attention quota, or would determined users just disable it and return to the same habits?

The accounting problem is harder than it looks. A notification daemon can count protocol events, but badges and focus stealing live elsewhere. You would need a broader desktop contract, otherwise apps simply move the interruption outside the quota.

I would describe the benefit cautiously. There is evidence that notifications can disrupt attention, but that is not evidence that a per-app quota specifically reduces doomscrolling. It seems plausible, not established.

The audit screen matters more to me than the number. “Browser used 4 units” is vague; “two sound alerts, one lock-screen banner, one urgent request” lets the user understand what the system is actually protecting them from.

Emergency exceptions are an attack surface. If apps can label themselves critical, the quota becomes decorative. The request should name the reason, and the user or a local policy should decide which senders are trusted.

I would use it for social apps and then spend the saved attention reading the quota logs. Still probably an improvement.

There is a product trap here: if setup requires understanding six interruption classes, most people will accept defaults. The privacy-friendly design may need a simple mode first, with the detailed accounting available for people who want it.

Keep the ledger local and exportable, not synced by default. A tiny append-only record of app, requested urgency, delivery decision, and timestamp would make the policy inspectable without turning behavior into a platform service.

Shared devices are not a footnote. A family tablet with one quota could suppress a child’s school reminder because an adult exhausted the allowance on shopping alerts. Visible user switching seems less annoying than silently merging everyone’s needs.

A determined user can disable a firewall too, but that does not make firewalls pointless. The useful audience may be people who want friction at the moment of temptation, not people seeking an unbreakable cage.

I would count delivery, not requests. Otherwise an app can burn the budget by asking repeatedly in the background, while bundling ten legitimate messages into one digest gets punished. The ledger should show both numbers.

Please make it boring and desktop-native. Per-app daily counters, a clear override, and no account required would get my vote. Add a “never interrupt” mode for everything except calls and alarms, and I would probably leave it on.