Human impact and climate change

Two smiling men pose in front of a plaque fixed to a white wall. — The researchers Charles David Keeling and John Chin in 1997, in front of the observatory building that had just been named after Keeling; John Chin measured carbon dioxide (CO2) in the air there from 1964 to 2000. John Miller (NOAA), Public domain, source

On climate, there is not one disagreement but two, and they are not of the same kind.

This distinction matters more than any figure on this page.

The mechanism, in four steps

Before the disagreements, the mechanism itself. It is short, and no one disputes it.

One. The Sun sends out light. Part of it is reflected back into space by clouds, snow and ice; the rest is absorbed by the ground and the oceans, which warm up.

Two. Every warm body radiates. The warmed Earth therefore sends energy back upwards, but in another form: invisible infrared, of longer wavelength than the light received.

Three. Certain gases in the atmosphere let sunlight through but absorb this infrared, then re-emit it in all directions — including downwards. They are called greenhouse gases: water vapour, carbon dioxide (CO₂), methane. They act like a blanket: they do not add heat; they slow its escape.

Four. Without them, the average surface temperature would be about −18 °C. It is about +14 to +15 °C depending on which organisation is cited — a difference of some thirty degrees. The greenhouse effect is therefore not a problem: it is what makes the Earth habitable. The problem is its intensification.¹

What humans have changed

By burning coal, oil and gas, we are putting back into the atmosphere carbon that had been removed from it and buried over hundreds of millions of years. Deforestation and agriculture add their share.

The blanket thickens, heat escapes more slowly, the temperature rises. The concentration of CO₂ in the atmosphere rose from about 280 parts per million before the industrial era to nearly 424 in 2024 — half as much again — and higher than at any time in the last eight hundred thousand years — which we know from the air bubbles trapped in ancient ice.²

Nothing in this chain is a recent hypothesis. That CO₂ absorbs infrared was measured in the laboratory by John Tyndall in 1859. What is debated comes after that, and that is the subject of this page.³

The two disagreements

The first concerns what is. Is the Earth warming? By how much? Why?

This kind of disagreement is settled by measurement. One looks, compares, and checks again. On these questions, it has been very largely settled — the rest of this page explains how.

The second concerns what should be done. Which decisions? At what cost? Who pays? In what order?

That one is not settled by measurement, because it is not a question of fact. It is a political question, in the true sense of the word: it concerns collective choices, and people do not have the same interests or the same values.

This site has a rule for live subjects: describe the disagreement without settling it. It applies to the second. It does not apply to the first — applying it there would amount to presenting as undecidable what has been decided.

This distinction is worth keeping in mind: it is useful on many other subjects.

How we know it is not the Sun

Here is one of the finest pieces of reasoning on this whole site, and it can be followed in full.

Two explanations compete: the Sun is heating more, or the gases we emit are holding in the heat.

At ground level, they predict exactly the same thing: it gets warmer. There is no way to tell them apart there.

So one has to look for where they do not predict the same thing.

If it were the Sun: the whole atmosphere would warm, from the ground to the top. This is what climate scientists point out: if the warming of the lower atmosphere came from the Sun, the upper stratosphere would also warm instead of cooling.⁴

If it is the greenhouse gases: the heat is held in down below. So the lower atmosphere warms — and the upper atmosphere, above it, cools.

Two opposite predictions, in the same place. All that remains is to go and look.

What is observed is the second case: the troposphere is warming, the stratosphere is cooling.⁵

This reasoning has a name on this site: the counterfactual test. If the other explanation were the right one, what would be different? This is how it is established that the tides do not come from the Earth’s rotation, that astronauts do not float for lack of gravity, and how Newton showed that colours are in the light and not in the glass. The same tool, more than three centuries apart.

The measurement is even more reliable than one might think: the upper atmosphere is much less turbulent than the lower one, where the ocean creates year-to-year variations, so a forced trend is easier to detect there. A 2023 study shows that by adding measurements from the upper stratosphere, the ratio between the human signal and natural noise is multiplied by five. Its authors see in this a human fingerprint on a global scale, of which stratospheric cooling is one of the main features.⁶

Two caveats, as always. The cooling of the upper atmosphere has several causes — greenhouse gases, but also ozone depletion and major volcanic eruptions: it cannot be attributed to greenhouse gases alone. Series of measurements at altitude are difficult to splice together across instruments, which has fuelled divergent interpretations.

The “97%”, and what it really measures

This figure is everywhere: 97% of scientists are said to agree on human-caused warming.

It comes from a study published in 2013, and it is worth stating exactly what that study did — because it is instructive both ways.

The calculation

The team examined thousands of scientific papers and sorted them into three groups:

66.4% did not take a position on the cause.
32.6% attributed it to human activity.
0.7% rejected it.⁷

The 97% is the ratio of those that endorse it to all those that take a position. It does not cover all the papers, and it does not concern scientists.

There is no cheating here: that is the question the study asked. But it is not what the phrase “97% of scientists” implies.

One question, three results

This is where the case becomes fascinating.

Depending on how the “no position” box is treated, one obtains very different figures:

about a third if they are counted as disagreements,
97% if they are set aside,
99.99% in another study, on other papers, which counts the “no position” papers as agreements.⁸

Three figures, one and the same question of method. Most of the gap comes down to how a single box is treated.

An argument, and the debate over it

How can one know which treatment is the right one?

The author of the 99.99%, James Powell, proposes a test. Apply the method to another field — for example plate tectonics.

Most geology papers take no position on the existence of plates: they assume it, because it is accepted. In his view, this silence does not signal doubt: he therefore counts the “no position” papers as agreements.

Another author replies that this test settles nothing: applied to geology, the method of the 2013 study also yields complete agreement. Both accept the human origin of the warming; they differ on how to count.⁹

Other measurements, other methods

Other teams have repeated the work using different methods.

A 2021 analysis covering about 3,000 papers finds only four in disagreement.¹⁰

Another, the same year, questions researchers directly: from 91% to 100% agreement depending on the degree of expertise, and 100% among the climate scientists who have published most on the subject.¹¹

Several different methods, compatible results. That is the real strength of the case — far more than the figure of 97%.

Why this page insists

Because neither “97% of scientists” nor “97% of studies” describes what the study measured.

Because a badly stated figure, even in the service of a correct conclusion, gives a foothold to those who want to dispute it. They do not need to attack the substance: they need only show that the sentence is inaccurate.

The scientific consensus on the human origin of the warming is real, measured by several independent methods. That is precisely why it deserves to be stated correctly.

What a company knew, and when

There is a second case on this page, and it is not about the climate itself. It is about what was said publicly.

In January 2023, a major scientific journal published the first systematic assessment of an oil company’s climate projections.¹²

Three researchers gathered all the warming projections produced by the company’s scientists between 1977 and 2003 — in internal memos and in published papers. In all, sixteen temperature projections, presented in twelve graphs and one table.¹³

Then they did something specific: they did not judge intentions. They tested these projections against the temperatures observed since, using the statistical techniques employed by international experts.

The result

Depending on the measure used, 63 to 83% of the projections are statistically consistent with what actually happened.¹⁴

The company’s scientists predicted a warming of 0.20 °C per decade, to within 0.04. That is what happened.¹⁵

They had also ruled out the hypothesis of a coming cooling, predicted that warming would become detectable around the year 2000, to within five years, and reasonably estimated the amount of carbon compatible with a two-degree limit.¹⁶

The figure that closes the case

One might object: perhaps these projections were right by luck, or so vague as to commit to nothing.

The authors therefore compared the quality of these models with that of the academic and government models published at the same time.

Skill score: 67 to 75%. Uncertainty margin: about 21%.¹⁷

These values are of the same order as those of the independent models. Their projections were, the authors write, at least as good as those of the academic world.¹⁸

The significance of this result must be weighed. The core of the case for doubt was the claim that climate models are not reliable. Yet the company had models that were as reliable as those it was calling into question.

The authors sum it up in one sentence: on each of these points, the company’s public statements about climate science contradicted its own scientific data.¹⁹

The company’s response

Questioned by the press in January 2023, it replied that those accusing it are wrong in their conclusions, and that its understanding of climate science has developed along with that of the broader scientific community.²⁰

We report it because that is the rule: when a page challenges someone, their response is included.

Readers can compare the two positions and read the documents cited on both sides for themselves.

And two warnings, as always

One should not conclude from this that no research can be believed. These particular projections were good. What was lacking was not the science: it was what was said about it publicly.

One should not turn this into the story of three heroes. These researchers drew on journalistic investigations carried out in 2015 and on their own earlier work.²¹

One last detail

How did this study begin?

It began with a graph reproducing the company’s projections, published in an earlier article. It was superimposed on the temperatures actually observed.

The overlap was striking. That was the starting point of the study.

Two curves laid one on top of the other — exactly the gesture that, in a room at the Lamont observatory, had revealed the mid-ocean ridge.

Current measurements and projections

To round off this overview, two sets of figures give the current orders of magnitude.

Period / Scenario	Observed or projected warming	Source / Reference
From 1850–1900 to 2016–2025	+1.26 °C on average over the decade (range 1.13 to 1.36 °C); human share: 1.37 °C in 2025, rising by about 0.27 °C per decade	Forster et al., 2026 (IPCC 1850–1900 baseline)
2081–2100 (Low scenario / SSP1-2.6)	+1.3 °C to +2.4 °C	Pathway with strong emission reductions
2081–2100 (Intermediate scenario / SSP2-4.5)	+2.1 °C to +3.5 °C	Current-policies pathway
2081–2100 (High scenario / SSP5-8.5)	+3.3 °C to +5.7 °C	Very-high-emissions pathway

Sources for the table²²

The 1850–1900 reference period is a statistical convention adopted to represent the pre-industrial climate before massive carbon emissions. The scenario ranges are those the IPCC judges “very likely”.²³

Note 1

If there were no greenhouse effect, the Earth’s average surface temperature would be a very chilly -18°C (0°F) instead of the comfortable 15°C (59°F) that it is today.

NASA, “Global Warming”, Earth Science Communications