A hot day is not an argument

I read Sadiq Khan on climate this morning, and started writing this post with a list.

Kew Gardens recorded no measurable rainfall in July — the first calendar month without any since the station opened in 1871. England had 6.5mm across the whole month, its driest July in nearly two centuries of observations. Southern England had its driest calendar month since the regional series began in 1836. Western Europe has just had its hottest June on record, 3.05°C above the 1991–2020 average.

Then I looked at the list again.

Every item on it is true, sourced and recent. As an argument, it is close to the weakest version available.

The symmetry problem

A hot day is not proof of climate change.

A cold January is not evidence that it has stopped.

Those two sentences have to travel together, and the moment you separate them the argument breaks. If the case rests on look how hot it is today, then anyone can wait for a cold week and run the identical logic in reverse. You have handed them the method. They only need to change the reading.

This is not a small rhetorical point. It is most of why the climate argument keeps restarting from zero. Both sides are arguing from the same kind of evidence — a recent, vivid, local observation — and that kind of evidence genuinely cannot settle the question. Neither can win, so the argument is permanent.

The evidence was never in the heat. It is in the pattern: repeated observations, historical baselines, changes in frequency and intensity, attribution work connecting particular events to a warming system. June 2026 matters far less as a hot month than as a record that broke the record set in June 2025. A collection of individual weather stories becomes something else entirely once the signals are connected — but the connecting is a separate act, and it is the act nearly everyone skips.

The deficit model is wrong

There is a standing assumption underneath a lot of talk about misinformation: that people are short of facts, and if they could just see the evidence they would understand.

People are not short of facts. They are surrounded by them, in greater volume and at lower cost than at any previous point. What is scarce is everything that has to happen to a fact before it becomes understanding — deciding which facts matter, how they connect, what the uncertainty means, whose interpretation to trust, and what larger account those facts appear to support.

That is a sense-making problem, and I recognise it because it is the same problem organisations have. It is the Evidence Engine argument at national scale. An organisation can hold thousands of transcripts, interviews and research documents without becoming any more intelligent. Capture is not understanding. Someone still has to determine what matters, what connects, what contradicts what we thought we knew, and what should change as a result.

The chain runs the same way in public as it does inside a company:

Signal        a reading, a hot day, one dry month
    ↓
Connection    placed against a baseline and against other readings
    ↓
Evidence      a trend that survives comparison and attribution
    ↓
Learning      a shared account of what is happening
    ↓
Action        policy, investment, adaptation

The difficult part is not collecting the evidence. It is moving through the rest of that chain without distorting it — and every step is somewhere the chain can be broken, honestly or otherwise.

Deniers, delayers and doers

Khan’s own distinction is more useful than the one I reached for first. He divides the field three ways — deniers, delayers and doers — and argues that the threat from deniers has largely been replaced by the threat from delayers: people who say the right things and defer the action at every opportunity.

That distinction matters, because delay is not a dispute about facts. It is a dispute about what the facts mean and how urgent they make things. Someone who accepts every measurement and disputes the significance is doing narrative work, not evidence work — and you cannot answer narrative work with another dataset.

Which is why climate denial was never simply a contest between facts and no facts. It is a contest over how evidence gets organised into meaning.

One person sees a heatwave.

Another sees weather.

A climate scientist sees an observation inside a much longer dataset.

A politician sees evidence for policy. A campaigner sees urgency. Someone already suspicious of climate policy sees exaggeration.

The underlying event has not changed at all. What differs is the structure each person is fitting it into — and that structure is doing more work than the measurement is.

The discipline this requires

If editorial work is the thing that moves evidence along that chain, it comes with obligations, and they cut in both directions.

You should not claim more than the evidence supports. You should not flatten uncertainty because it complicates the story. You should not treat one dramatic anecdote as proof of a systemic trend — which is exactly the failure I opened this post with, and I had written four sentences of it before I noticed.

But there is a fourth, and it is the one that gets left out. A demand for more evidence is not automatically a neutral act. It can function as delay regardless of what anyone intends by it, simply because the request can always be repeated and there is always another study. When substantial evidence has already accumulated, “we need to know more” stops being a research position and starts being a scheduling decision.

Both failures are failures of the same discipline. Overclaiming breaks the chain by skipping steps. Endless deferral breaks it by refusing to take the last one.

There comes a point where the job is not finding another fact.

It is helping people see the pattern.

Evidence is necessary. Sense-making is what makes it usable.