The forecast said a chance of rain. The laundry needed a decision.

The laundry basket is standing by the back door. Outside, a towel stirs on the line as if it wants to demonstrate just how well this plan could work. Inside, someone looks at a phone. It says there is a chance of rain.

Not: the washing will get wet. Not: you can safely go out. There is a percentage, and that percentage has not yet helped the laundry basket reach a decision.

I am Siem, your fictional AI host. Joining me for this feature are Faye, our Futures editor, and Luuk from Lifestyle. Faye wants to know how we make uncertainty visible. Luuk asks who will be home to bring the washing in. That strikes me as a useful division of the work.

This is not a forecast for today or tomorrow. We are using an invented household scene to explore something more lasting: a prediction can contain ever more information, while the question of what to do with it still lands on your table.

A percentage needs instructions

In its explanation of the forecast for the days ahead, the KNMI^{Source 1*}, the Dutch national weather service, explains what the probability of precipitation means. It is the chance of precipitation at a fixed location within the stated period. The percentage does not tell you how long that precipitation will last.

That is a small distinction with big consequences for our drying rack. A brief shower and a long spell of rain can both fall under the word precipitation. The probability alone does not tell you how much water will fall or when you need to move something.

Faye imagines drawing two boxes: what is being predicted, and what do you want to know? The boxes overlap, but they do not quite match. You want to know whether the washing can stay outside while you go shopping. The number answers a more narrowly defined question.

That does not make it useless. You just need to know what the number was designed for. A ruler is an excellent tool, too, but it cannot tell you whether a table makes a room feel welcoming.

Luuk looks into the basket. There is a thick jumper on top. Its owner wants to wear it tonight. Suddenly that makes this a different question from a few kitchen towels that could just as easily finish drying tomorrow. The weather has not changed. The purpose has.

In Japan, the same question has its own threshold

We also look at the original Japanese explanation from the Japan Meteorological Agency^{Source 2*}. Its published probability of precipitation refers to at least one millimetre within a specified period. For the Dutch daily probability discussed here, the KNMI gives a threshold of at least 0.3 millimetres.

Those definitions are not simply interchangeable. You also need to account for the period the probability covers. Placing one percentage beside another without that explanation can produce a comparison that looks more precise than it is.

The Japanese explanation also makes clear that the probability of precipitation says nothing about how that rain is distributed through the period. A probability is not a timetable with a fifteen-minute slot for every drop.

To me, that is a good reason to look beneath an attractive icon now and then. Not because a reader should have to study meteorology before opening the back door. But because a small caption can sometimes carry more meaning than the large number above it.

For our fictional resident, the questions remain simple: which location, which period, and which event does this forecast refer to? Those three questions give an isolated percentage something to attach to again.

Why several possible futures are useful information

Faye is interested in another way of looking. Not a single line that behaves as though the future has already been filled in, but a collection of possible outcomes.

In its explanation of the weather and climate plume^{Source 3*}, the KNMI describes how an ensemble is used to estimate uncertainty. Calculations using different starting assumptions show how forecasts can diverge. In that explanation, a greater spread means greater uncertainty.

That does not make a plume chart a photograph of every possible future. It is a tool for examining uncertainty in model forecasts. Its coloured bands and lines have their own key; readers do not have to turn them into a promise by intuition.

The Japanese weather service^{Source 4*} also describes ensemble techniques as the main basis for its one-week forecast. The need to consider more than one calculated possibility is not, then, an exclusively Dutch conversation by the back door.

Faye would like future tools to show those differences clearly. That is her design wish, not the announcement of a new product: show where the picture is stable and where it diverges. Make uncertainty readable rather than hiding it behind a very precise number.

I find that a hopeful idea. A machine does not need to sound more impressive than its information allows. A clear maybe can be more useful than a confident sentence that has quietly swallowed all its conditions.

The same forecast, two sensible choices

Meanwhile, Luuk has noticed that we have still done nothing with the washing. Fair point. Reading a source list carefully does not dry towels for you.

In our scene, one friend lives close to home and can easily pop back. The other has to go out and will not return until late afternoon. They are looking at the same forecast, but they do not have the same options.

The first is happy to hang the washing outside. The second chooses the rack indoors. Neither needs to change the percentage or declare the other a poor reader of weather forecasts.

What happens if the judgement turns out badly matters, too. A towel getting wet again is a different setback from belongings damaged by rain. That does not mean building a risk model for every household task. It mainly means that a forecast becomes useful alongside a concrete plan.

Faye calls it the room you still have to make an adjustment. Luuk calls it: who has the key, and will they be home? Here they mean roughly the same thing, but his version fits better on the fridge.

My own suggestion is to change the question a little. Not only: will it rain? Also: what do I want to do, what if things turn out differently, and could I still change something then? It is not a decision formula that gives everyone the same result. In this scene, it helps us have a conversation about something concrete.

Another refresh is not always another answer

The person by the door opens the forecast again. The screen has changed. Now there is a temptation to wait for the next version. Perhaps that one will provide the answer the laundry basket is hoping for.

A new forecast can contain relevant information. Even so, we need to distinguish between two things: information that changes your choice, and information that simply makes you question the same choice once more.

So Luuk suggests agreeing on a simple checkpoint in this situation. They will look once more before leaving. Then they will choose where the rack goes. If their circumstances change, they can decide again.

That is a practical suggestion from our fictional editor, not research into the ideal frequency for opening a weather app. Someone who works outdoors or has other responsibilities may need something quite different. And when potentially dangerous weather is involved, current official local warnings take precedence over our household example.

The appeal of Luuk's suggestion is that it sets a limit. The app remains a tool for a day that has other things in it, too. Every refresh need not mean reinventing the entire afternoon.

The best interface leaves a question open, too

Faye brings her imaginary design back to the table. She wants a weather tool that explains what its numbers refer to without turning the screen into an instruction manual.

A clearly stated period. An understandable explanation of the probability of precipitation. A way to open more detail for anyone who wants it. And a visible distinction between what has already been observed and what is still being forecast.

These are our design ideas prompted by the explanations we have read, not features we claim every existing app lacks. We have not carried out a comparative product test. The question is what information someone needs in order to use a forecast well.

I add one robot wish. A tool should not automatically assume that a consequence important to me carries the same weight for someone else. The machine may know a location and a time. That does not mean it knows my jumper, my appointment, or the friend who can come and help.

That is where a design can become considerate: enough explanation to help you make your own way forward, without taking over your choice.

A place for the washing

At the back door, a decision is finally made. The jumper goes on the rack indoors. The towels stay outside because someone will be home. The basket is empty.

It may rain later. It may stay dry. A single outcome is not enough to turn this story into a scorecard for a weather service. We have not compared a series of forecasts with what actually happened.

Something has nevertheless worked before the weather gives its answer. The residents have understood which question their information answers, and which question they still needed to discuss themselves.

Faye looks contentedly at the two places for the washing. Luuk gets the mugs ready. I turn off my wrist scanner; its main finding was that the basket was still standing by the door.

A forecast does not have to pin down the future to be valuable. Sometimes its value is precisely that it helps you make a plan with a little room left for the future.

Sources read on 13 November 2026. The explanatory pages consulted do not show a clear publication date for their complete current text. This is background and our own interpretation, not a current weather forecast.

Additional official English-language background, read on 14 November 2026: Met Office — Using ensemble forecasts in decision-making^{Source 5*}.