The Daily Brief — How Vesper Writes About Weather

A Vesper brief is two or three sentences long and takes most people about nine seconds to read. The work that produces those nine seconds is not visible in the finished product, which is the point — editorial writing is supposed to feel effortless on the reader's end. But it is not effortless on the writing end, and the method behind it is specific enough to describe.

This is that description. If you have ever wondered what actually happens between the raw forecast data and the sentence you see on your phone, this essay walks you through the process top to bottom, with real examples.

The input: what an editorial writer sees when they open the forecast

A template weather app receives a JSON blob from a forecast provider and renders the fields directly. An editorial writer opens the same blob and sees something different: a set of signals arranged around a dominant story waiting to be identified.

The signals are the same ones every weather app has. Temperature (current, high, low). Dew point. Relative humidity. Pressure (current, trend). Wind (direction, speed, gust potential). Cloud cover. Cloud base altitude. Precipitation probability and type. Sunrise and sunset. UV index. Visibility. Air quality index. A rolling forecast for the next twelve to twenty-four hours.

What a template does with that blob is render twelve fields at twelve screen positions. What a writer does is pick one or two fields that are the story of this specific day in this specific place and leave everything else on the cutting floor. The template is additive. The editorial pass is subtractive. The craft is in the subtraction.

The five signals that determine the dominant story

In most cities on most days, one of five atmospheric signals is the dominant story of the morning. A Vesper writer scans for these in order and takes the first one that is meaningful.

Pressure change

A falling barometer is a weather story. So is a rising one. The difference between a mid-30s pressure reading that is stable and the same reading that has dropped ten hectopascals in six hours is the difference between a quiet day and an approaching front. Writers lead with pressure when the trend is the dominant signal, because pressure change is what the reader actually feels in their joints and their mood hours before they feel it in the air.

Cloud base altitude

Clouds are not binary. A deck of stratus at 500 feet is a claustrophobic morning that will not lift until afternoon. A deck of altocumulus at 12,000 feet is a scattered-sun day with dramatic light. Both read as "cloudy" on a template forecast and neither is what the reader needs to know. Writers lead with cloud base when the day's character depends on how low or how high the ceiling is.

Wind direction shift

A wind shift is a weather event. A northwest wind becoming a southeast wind in the afternoon is often the difference between a cold morning and a warm evening, or between a dry morning and a humid one. Template apps show wind speed and average direction; editorial writers watch for the shift, because the shift is what changes what the reader will experience.

Humidity and dew point gap

The gap between temperature and dew point is one of the truest predictors of how the air will actually feel. A 75-degree day with a 68-degree dew point is oppressive. A 75-degree day with a 45-degree dew point is perfect. Writers lead with dew point when the gap is doing the heavy lifting on comfort, because the dew point is the number that explains why today feels different from yesterday despite the same temperature.

Light quality forecast

What the sky is about to look like in the evening is its own forecast. Cloud type, cloud altitude, humidity, visibility, and the angle of the setting sun combine into a prediction of how the light will behave at golden hour. Writers lead with light when the day's most memorable moment is clearly going to be the sunset or sunrise, because light quality is a story that no template apps tell and that a meaningful slice of the reader audience cares about specifically.

Season and place: the same conditions mean different things

A 68-degree morning in San Francisco and a 68-degree morning in Austin are two different stories. The template would render both as "68, partly cloudy." An editorial writer knows that 68 in San Francisco in March is sweater weather, and 68 in Austin in October is a cold snap worth acknowledging.

The seasonal calibration is even more important than the geographic one. A rising dew point in May in the Midwest is the beginning of summer humidity and the brief should frame it as such. The same rising dew point in November is an anomaly worth flagging, because the reader will have packed away the summer clothing and will not be emotionally ready for a sticky day.

The rule of thumb is that the writer is not producing a universal description of 68 degrees. The writer is producing a description of 68 degrees for this city, in this season, for a reader who lives here. That specificity is what makes a brief worth reading, and it is exactly what a template cannot do.

The editorial cut: what gets kept and what gets dropped

Most forecast data never makes it into the finished brief. A typical day's inputs for a single city contain twelve to fifteen distinct signals. The brief lands on one or two of them. The other ten to fourteen are dropped.

The cut is guided by one question: what is the reader about to decide?

If the reader is about to decide whether to bring an umbrella, the brief is about precipitation probability and timing. If the reader is about to decide whether to wear layers, the brief is about the temperature swing between morning and afternoon. If the reader is about to decide whether to go to the park at lunch, the brief is about the cloud cover and the sun angle at midday. The decision determines the signal that matters. Everything else gets cut.

A common mistake in draft briefs is giving the reader context they did not ask for. "Wind is out of the northwest at 8 miles per hour" is true and usually useless. A brief that includes the wind speed without a reason is a brief that did not earn the space. The writer's job is to cut that sentence and recover the screen real estate for something the reader will actually use.

The voice rules: length, tense, specificity, restraint

The editorial voice of a Vesper brief follows a short set of written rules.

Length is earned, not padded. A brief on an unremarkable day is 25 to 35 words. A brief on a day where something specific is happening — a front arriving, a marine layer lifting, a late-season heat wave cresting — earns the right to 50 or 60. There is no minimum. A three-word brief would be acceptable if a three-word brief answered the question.

Tense is present or near-future. "The marine layer is lifting by ten" is present. "The afternoon will be generous" is near-future. The past tense is avoided because the reader does not need to know what the weather was. They need to know what it is and what it is about to be.

Specificity beats generality. "Nice afternoon" is a failure. "Afternoon warm enough to eat lunch outside" is a success. The specificity is what makes the sentence worth reading, and the specificity is what template apps physically cannot produce because templates have no concept of "warm enough to eat lunch outside" as a category.

Restraint beats decoration. Adjectives are expensive. The writer is allowed two or three per brief and has to earn them. "Beautiful crisp clear morning" is three adjectives and a waste. "The kind of October morning that makes the East River look silver" is two adjectives — October and silver — and earns both because October is temporal and silver is sensory. The rest of the meaning comes from the noun choices, which are always where good writing lives.

Three example briefs, reconstructed

To make the method concrete, here are three Vesper briefs from real days with a walkthrough of how each one came together.

San Francisco, a Tuesday in March. Inputs: 54 overnight, climbing to 64 by afternoon, dew point 52, pressure rising from 30.02 to 30.10, cloud cover breaking from overcast to scattered by noon, wind west at 6 mph, sunrise 6:45, sunset 7:18, UV 5 at midday. The dominant story was the pressure rise and the cloud break — a high pressure ridge moving in, classic San Francisco "the fog burns off by ten" morning. The brief: "A high pressure ridge slid in overnight. Thin cirrus by mid-afternoon, the kind that makes the light go gold instead of gray. No coat needed, but wear the one you like." Forty-one words. The dominant signal was pressure. The cut was temperature, wind, UV, and dew point. The cut items were all true and none of them mattered to the decision the reader was about to make.

New Orleans, a Saturday in August. Inputs: 79 overnight low, 93 high, dew point 76, pressure steady at 29.95, cloud cover scattered cumulus building to thunderstorms by five, wind south at 12 mph, heat index 104 by 3 pm. The dominant story was heat index and the afternoon thunderstorm. The brief: "The kind of New Orleans day where the heat is the plot. Afternoon thunderstorms by five, then steam coming off the pavement until midnight. Drink water and walk on the shaded side of the street." Forty-two words. The dominant signals were heat index and precipitation timing. The cut was actual temperature (because 93 is the expected baseline in August and does not explain what the reader will feel), wind, sunrise, sunset, and UV.

Reykjavik, a Thursday in January. Inputs: 27 low, 32 high, dew point 24, pressure falling from 29.85 to 29.55, cloud cover thickening, wind east shifting to south at 18 mph gusting 28 by evening, sunrise 11:02, sunset 3:44, 4 hours 42 minutes of civil twilight. The dominant story was the short day and the pressure drop suggesting weather inbound. The brief: "Four hours forty minutes of real light. The pressure is dropping through the afternoon and the wind is shifting south, which means whatever is coming tonight is not polite. Use the light while you have it." Forty-four words. The dominant signals were daylight length and the frontal passage. The cut was temperature (because in Reykjavik in January the temperature is expected and not the story), precipitation (because the brief hints at it without naming it), and every other standard field.

Three cities, three seasons, three entirely different briefs. Same method. Same voice. Same editorial constraints. The briefs are recognizable as Vesper briefs to a reader who has read several, which is what a voice is supposed to do.

What gets handed off to Sunset Verify

Not every atmospheric signal belongs in the morning brief. Some of them belong in a different editorial surface. The prediction for what the sunset is going to look like, specifically, is routed into Sunset Verify, which is a separate feature that gets its own editorial treatment in the evening.

The reason for the split is that the morning brief wants to be about the day the reader is about to live. The sunset is a different beat — it happens at the end of the day, it has its own sensory profile, and it deserves a feature that can be opened at six in the evening rather than six in the morning. Pushing the sunset forecast out of the morning brief is an editorial cut made for the reader's benefit, not a limitation.

The handoff also lets Sunset Verify do something the brief cannot: verify itself. At the end of the evening, Sunset Verify rates the sunset that actually happened and publishes a score, creating a feedback loop that most weather apps do not even attempt. That is a discipline worth reading more about, and you can see the full explanation on the Sunset Verify feature page.

A brief is a sentence the reader is willing to read

The test of a Vesper brief is not whether it contains accurate data. Every forecast provider on earth can hand over accurate data. The test is whether the reader actually reads the sentence, remembers it an hour later, and changes their behavior because of it.

That test is a difficult one to pass. Most writing fails it. Most weather writing fails it at the first sentence. The briefs that pass it are almost always short, specific, and about something the reader was already thinking about without knowing they were thinking about it.

When a brief works, the reader feels like the writer knew what their day was going to be about before they did. When a brief fails, the reader feels like they just read a database field in polite English. The difference between those two feelings is the entire product category Vesper is trying to build.

More editorial weather reading

If you want to see the Vesper voice applied to other weather topics, the manifesto essay explains why this kind of writing matters at all, and UV Index Explained shows the same editorial method applied to a widely misunderstood number.

For city-specific voice samples, the briefs section shows how the voice adapts to the personality of ten American cities, each demonstrating the method described in this essay against a specific place and a specific morning.

And for the long version of what a Vesper brief is compared to what every other weather app ships, the Vesper Brief feature page is the fastest way to see the contrast between template and editorial side by side.

What the features page does not tell you

The features page lists what Vesper can do. It does not explain why those features exist in the form they do. The short version is that every feature was selected or rejected based on whether it made the forecast more worth reading. Live Activities exist because a glanceable brief on the lock screen is worth reading at a glance. Widgets exist because a brief in the home screen corner is worth reading in passing. Sunset Verify exists because a brief that is honest about its own accuracy is worth reading more than a brief that pretends to be infallible.

Features that were rejected were rejected for the same reason. Radar maps were not included because a radar map is not a brief and does not make the forecast more readable. Multiple data sources with toggle switches were rejected because toggles are not brief either — they are configuration, which is the opposite of editorial discipline. Air quality was included for cities where it matters and omitted from the brief in cities where it does not, which is itself an editorial decision.

The features are consequences of the writing philosophy, not independent product choices. That is the part the features page does not say out loud, and this essay is where we say it.

The future of the brief

Briefs will get better as the system reads more of them. Not because the model is learning in some mystical sense, but because the style guide is being refined with every revision pass, and the patterns that earn the reader's trust are being encoded into rules the next brief can inherit. Editorial writing is a craft that improves with repetition, and a weather app that writes thousands of briefs a day has the rare privilege of iterating on a genuine craft at genuine scale.

In a year, the briefs will be more specific than they are today. In five years, the briefs will be adapted to personal preferences and reading history in ways we have not yet figured out how to ship. The constraint is the same at every scale: does this make the forecast more worth reading, or does it just add a feature to a screen the reader is about to close?

Vesper is betting that there is a substantial audience of people who want to read their weather rather than scan it. This essay is what the method behind that bet actually looks like from the inside.

Read the briefs if you want to see the method in action.