aislop.day
SATURDAY, 20 MAY 2023

Field Notes on Public Transport Scroll

The commute became the feed's most reliable uninterrupted window, and the feed scheduled itself accordingly.

3 MIN READshort video behavior

Location: Bus, metro, train, or any vehicle where the subject is not required to be alert Observation Period: Journey start to journey end, extended

The journey is twenty-three minutes. The subject opens the app at the second stop, after the initial boarding activity has resolved and a seat has been secured. The first clip is thirty seconds. The second is forty-two seconds. These numbers are not tracked by the subject.

By the fourth stop, the subject has watched nine clips. This is established without intention. The feed structures itself around the gap between where attention would otherwise go and where it can be redirected within a single swipe.

Each clip is easy to leave while the feed removes any natural stopping point from the available options. This is not a design flaw. It is the design. The absence of a stopping point is the feature that converts a commute into a session.

Under the governing rule, public transport scroll extends the break by one clip before stopping is permitted. The clip plays. The next clip begins before the decision about whether to stop has been reached. The decision is deferred again, at the same cost, with the same justification: it is only one more.

The subject misses the station landmark they intended to watch for. This is the observation that prompted the field note. Not the missing of the landmark specifically, but the absence of any internal event marking the moment when watching became ongoing rather than chosen.

The break becomes a session requiring a separate act of will to exit, which is a different kind of break.

A second observation: the feed's calibration improves over the commute window. Clips surfaced after the third minute are better matched to retention than clips surfaced at the start. The system has learned the duration. The subject's departures feel, in retrospect, more like interruptions than completions.

The commute ends. The subject pockets the phone. The total elapsed time on the app is not available in the interface. An estimate would be possible. The subject does not produce one.

Filed for: the friend whose feed reliably produces this, and who would recognize the landmark problem immediately.

TAGSshort video behavior
Share this