Imagine a commuter whose bus route may be cut. At bedtime she checks the same local-news story again. The proposal has not changed; neither has the date of the vote. She is trying to work out whether there is something she must do tomorrow.

Three visits can look better than one. They may also mean that the first two failed to answer her question.

Now imagine an article that makes the situation legible: what has been proposed, what remains undecided, when the decision comes, and what she can do before then. She reads it once and puts the phone down.

If we judge the article only by whether she comes back, we risk calling the better explanation a failure.

Avoiding the feed is not rejecting knowledge#

In his 2024 Reuters Institute chapter on audience needs , Richard Fletcher reports that knowledge and understanding remain important even among people who avoid news. Inspiration and diversion matter relatively more to these audiences, but that does not erase the demand to understand.

This is a survey of stated needs. It does not show that adding explanation makes people return, or that a good explanation lets them safely stop. It does undercut a lazy assumption: wanting less news means wanting less understanding.

My narrower suspicion is that, on a bounded issue, explanation can sometimes reduce repeated checking without reducing understanding. Not because the world has stopped moving. Because the reader knows where the uncertainty is and what would make it worth looking again.

That would be a useful result even if it cost a visit.

Less is not automatically better#

There is an obvious trap here: replacing the worship of engagement with the worship of disengagement.

In The Welfare Effects of Social Media , Hunt Allcott and colleagues report a randomized experiment in which deactivating Facebook for four weeks before the 2018 US midterm election increased subjective well-being while reducing factual news knowledge.

A quieter life and a better-informed life were not the same outcome. That intervention is not an explanatory article, and an election is not our commuter’s bounded question. But it is enough to stop me treating fewer checks as their own certificate of success.

Confidence is another bad substitute. The abstract of Fisher, Goddu and Keil’s study of online explanation searches reports increased self-assessed knowledge across nine experiments. Feeling that we understand is not a measurement of what we can explain.

These studies concern different interventions and outcomes. They do not form a chain in which explanation produces confidence, confidence stops checking, and stopping produces ignorance. That would be a tidy story made by joining evidence that does not meet.

The distinction we need is simpler: frequency of use, confidence and understanding must not be allowed to impersonate one another.

An ending without false closure#

For our imagined commuter, a useful explanation need not promise that the route is safe. It can say the opposite. The proposal remains live; the vote is still ahead; this is the part she can act on now.

The question is whether she can carry that understanding away from the page. Can she distinguish a proposal from a decision? Can she recognize a later change that matters? Can she choose when to look again rather than repeatedly checking an unresolved threat?

These are test questions, not demonstrated benefits. Randomly assign readers to an explanatory article or a factual report on the same bounded issue. Match reading load, deadlines, action options and access to updates: a missing vote date must not win the comparison. Count voluntary rechecking separately from visits prompted by the test or a real update. Set acceptable limits beforehand for lost understanding and missed changes. Test understanding after a delay and recognition of a consequential update, not reassurance alone. An inconclusive difference is not proof that nothing was lost.

Fewer visits would not count as success if they left readers worse informed by those standards. Nor can the comparison be won by hiding updates or making access harder. An exit is useful only if the reader remains able to return.

Ask why readers stopped: they may have understood, given up or found an answer elsewhere. A visit count cannot tell. Their answers do not replace a comprehension test.

This is not a rule for emergencies or every developing story. Sometimes frequent checking is necessary. Sometimes it is freely chosen interest or pleasure. The point is to stop treating another visit as the only desirable ending.

Who pays for the useful ending?#

There is a material objection. A publisher paid for visits cannot settle it by changing the label on a chart. If an explanation serves the reader while reducing traffic, the benefit and the cost may land on different sides.

I have not supplied a funding model that resolves that conflict. An editorial criterion is not evidence that a commercial institution will adopt it. But the conflict should remain visible, rather than being smuggled into the claim that what keeps a reader returning must be what helps them.

Journalism needs readers. That does not mean a reader owes it permanent attendance.

The bus-route story may remain unfinished. Our commuter may still face a fight. A useful article would not solve either problem for her. It might let her understand the next step well enough to spend the rest of the evening doing something else.

That possibility is worth testing. Helping someone understand should not require keeping them as a customer of their own uncertainty.