Building in a Swamp

According to Karl Popper, science is a process of conjecture and refutation. A hypothesis is formulated, and using deductive reasoning, empirical tests are proposed; these tests are performed and the hypothesis is either corroborated or falsified. For a new hypothesis to falsify an existing one, it must have greater explanatory power – it must explain all that was explained by the existing hypothesis, and either resolve anomalies (for example, quantum mechanics explained the anomalies that Planck observed in measurements of black body radiation that were not predicted by classical physics) or make new predictions (for example, Dirac was able to use quantum theory to predict the existence of a new sub-atomic particle, the neutrino).

Popper calls the measurements that we make to test our hypotheses basic statements. At first sight, these feel a little like what we call “facts” in everyday language. However, Popper was anxious not to turn the “empirical basis” into a collection of “facts”. Philosophers have argued long and hard about how our perceptions (which are ideas) are connected with our sensory experiences of the world we live in. For example, if I say that my shirt is blue, what does that mean? How do we connect the underlying physical phenomena – the absorption and reflection of photons by the fabric of my shirt, the impact of some of those photons on my retina and the transmission of a signal by my optical nerve to my brain – with my belief that my shirt is blue? Popper argues that attempts to establish “facts” – empirical statements that we trust as “true” or “reliable” – leave us trying to test objective statements about the world (scientific hypotheses) by comparison with statements that are founded on psychological states (our belief that they are “facts”).

Instead, he argues, science is an objective process based on deductive reasoning. For example, if we hypothesise that electrons have a property called “spin” that has one of two values, +¹/₂ or –¹/₂, then we must design an appropriate experiment to test this hypothesis. Such an experiment was devised by Stern and Gerlach, who passed a beam of silver atoms through a magnetic field towards a screen and observed that two spots were formed on the screen. Silver possesses an unpaired electron, and these spots corresponded to atoms in which the unpaired electron had a spin state of either +¹/₂ or –¹/₂, which were deflected in opposite directions by the magnetic field. However, while anybody could observe two spots on a screen, it is not at all obvious that the observation of those spots corroborates the hypothesis that an electron can have a spin of either +¹/₂ or –¹/₂. For the spots to have that significance requires us to lay out a deductive argument that leads from the hypothesis to the observation, and that includes both a description of the measurement apparatus, and also theoretical understanding (e.g. that the magnetic moments of silver atoms will be different if the spin of the unpaired electron is different).

The concept of electron spin has explanatory power, because it gives us a means to account for the fact that two spots are formed on the screen using the Stern-Gerlach apparatus. However, the basic statements that enabled us to corroborate the hypothesis were not self-evident truths. Thus, the process of conjecture and refutation is not one in which we establish “objective truth”, but one in which our knowledge evolves towards greater verisimilitude (truth-likeness). Theories compete for survival, and those that have greater explanatory power have a competitive advantage. Popper summarises the argument as follows:

“The empirical basis of objective science has thus nothing ‘absolute’ about it. Science does not rest on solid bedrock. The bold structure of its theories arises, as it were, above a swamp. It is like a building erected on piles. The piles are driven down from above in to the swamp, but not down to any natural or ‘given’ base; if we stop driving the piles deeper, it is not because we have reached firm ground. We simply stop when we are satisfied that the piles are firm enough to carry the structure, at least for the time being. [1]

My Fact is Bigger than Yours

Popper developed a philosophical framework that was an extension of his model for the scientific method. He called this “critical rationalism”, and characterised it as “mutual control through rational criticism”. Popper was also a realist: he believed in the existence of an objective external reality. Most experimental scientists are also realists: when we make measurements, we receive nature’s verdict on our hypotheses; the outcomes of experiments have force because of the concrete, objective nature of the external world that we encounter through them.

For Popper, critical rationalism means that one simply cannot believe whatever one wants to: one’s beliefs can be questioned (criticised) and subjected to rational analysis. In debates around climate change, scientific reasoning and political argument converge: questions about energy policy, for example, which have massive sociological impact, are informed by empirical evidence. Science can never tell us how we should live, but decisions about energy policy (with consequences for taxation, the warmth of our homes in winter) can be informed by scientific evidence.

In 2024 the UK government announced plans to grant licences to drill in the North Sea, claiming that such a policy would lead to reduced greenhouse gas emissions than importing oil and gas from overseas and that it would help the UK’s energy security. The BBC fact-checked the government’s claims. However, the BBC’s approach was not merely a presentation of facts; it was really an application of the Popperian critical rationalism in which BBC journalists reasoned deductively from the Government’s claims to formulate hypotheses that could be tested by comparison with data. For example, the BBC considered the claim that granting new licenses to drill in the North Sea would improve energy security for the UK. It considered the amount of oil and gas in the North Sea, and the fraction that is sold internationally, at no benefit to the UK and questioned whether the extent of the benefit of new drilling to the UK would be as great as was claimed. Thus such “holding-to-account” is based on rational criticism.

However, in the culture wars, “facts” are used very differently as gotchas designed to be thrown at opponents. Numerous examples can be found in videos of the US right-wing political activist Charlie Kirk. Although Kirk was feted after his murder as a debater, who spread reasoned dialogue across US campuses, my impression of many videos uploaded to the Turning Point You Tube channel is that his approach was more properly characterised as rhetoric. He was incredibly impressive, but his performances tended to rely on sharp put-downs and rhetorical trickery rather than rational argument.

For example, Turning Point UK (the UK arm of the far-right activist organisation he founded) posted a video on You Tube in which, Turning Point claims, “a British leftist student tells Charlie Kirk facts are unfair”. The student begins by challenging Kirk about the sources of his “facts”, which he says Kirk doesn’t give. Kirk tries to draw him into an argument about abortion but the student asks Kirk to expand instead on claims he made about the correlation between poverty in US cities and the political allegiances of their mayors. The student says that Kirk had claimed that the poorest 10 cities in the US were run by Democrats (Kirk agreed that this was a claim he had made), and said that when he looked into the matter, he found that the 10 wealthiest cities were also run by democrats and in fact that most cities were run by Democrats – so where was the evidence of a correlation between liberal politics and poverty?

Far from arguing that “facts are unfair”, the student was arguing that Kirk’s analysis of the data was questionable. There is certainly an important question to be explored here. Let us suppose for the sake of argument that in a list of American cities the 10 poorest were actually run by Democrats and the 10 wealthiest by Republicans. In this counterfactual America, an important question would still remain: in the relationship between liberal politics and wealth, what is the cause and what is the effect? Do Democrats make people poorer, or do poor people vote for Democrats? And do Republicans make people richer, or do rich people vote for Republicans?

Rather than taking up an opportunity for debate, Kirk instead adjusted his claims. He invoked what he called the “laboratory of democracy theory, which we all believe in” and proceeded to claim, without sources, that states that were run by democrats were crime-ridden and poor, whereas states that were run by republicans were wealthy. In other words, far from debating the student’s argument, he evaded it. The student began by asking about sources. However, Kirk had provided none, and instead, when in a corner, invoked an imaginary entity to shore up his position.

The conversation moved to healthcare in Cuba. Kirk had previously claimed that life expectancy in Cuba was 15 years less than in the US, but the student had done research and had found that life expectancy was 1 year longer in Cuba than in the US. Again, without giving any sources, Kirk said he rejected the Cuban data because they were produced by the Cuban government and that other sources had found them to be false. He did not say what these other sources were, and continued “we can have a 30 minute conversation about the Cuban health statistics…but show me another statistic of anything I’ve talked about…everything I do is rooted in years of research and backing and data”.

Also on stage with Kirk was a group of supporters who joined in the argument. They claimed that because Kirk could cite sources for his statements they were “facts”; the student countered that Kirk had selected a small sub-set of data from his sources and had used these selectively to support an argument that the data, when considered as a whole, did not support. Kirk and his supporters argued that because Kirk was a debater, he wasn’t obliged to use facts that didn’t support his point of view. This is an extraordinary argument: rational discussion surely involves the consideration of competing explanations for observations. The student challenged Kirk’s hypothesis that when cities are run by democrats, they are poorer. He argued that data showed that on the whole most cities are run by Democrats, and that such cities span the range of wealth from rich to poor. These data falsify the hypothesis that there is a correlation between Democratic politics and the wealth of a city; selecting just data on the poorest cities in this set and claiming this as “factual” evidence to support Kirk’s claim that Democratic cities are poorer is neither rational nor honest.

Climate Misinformation

Here we see “facts” being used to batter opponents, not as part of a rational dialogue but rather in the manner of a stand-up comedian wisecracking at the expense of audience members. This approach – selecting from a dataset the entries that suit our purposes and ignoring those that do not – runs contrary to scientific method. When students ask me whether it is OK to remove an anomalous data point from a graph, my response is “no, unless you have an objective reason for doing so (e.g. the spectrometer crashed halfway through collecting a spectrum”). Sometimes, anomalies turn out to be incredibly significant and in all experiments, there is uncertainty in data points, which we quantify in the form of error bars. Most important of all, the process of selectively deleting outliers is usually based on a belief about their rightness; it is as though the fact that these points are not in conformity with our expectations (our beliefs) makes them dubious. Thus the rejection of outliers usually brings subjectivity into a process that should be objective.

The selective elevation of specific data to a “gotcha” status is central to the misinformation work of the Heartland Institute. For example, on its “climate realism” web page, Anthony Watts recently complained about BBC Radio 4’s outstanding programme More or Less, which featured a special issue on climate change in which leading climatologist Friederike Otto from Imperial College London discussed the effect of climate change on rainfall in the UK. The Met Office published data recently showing that in January 2026, the UK recorded 117% of its average rainfall.

However, Watts argues that this isn’t showing the whole picture, because Scotland recorded below-average rainfall. Thus, he is arguing that the Met Office claim that the UK as a whole is experiencing increased rainfall is dubious because one region of the UK has less rainfall. This claim is nonsensical, of course. It would be expected that the rainfall data for the whole of the UK would contain within them some regional variations; the hypothesis that mean UK rainfall is increasing s not contradicted by the observation that rainfall in Scotland is reducing. In a section of is report headed “Rainfall: strong regional contrasts”, the Met Office reported that Northern Ireland saw 170% of its average rainfall in January 2026, for example. In the same way, “global warming” will cause the mean global temperature to increase over time, but if the Atlantic Meridional Overturning Circulation (AMOC) shuts down, the UK will get a lot colder despite the rise in global mean temperatures.

First the Hypothesis

Most climate misinformation revolves around unscientific abuse of data. The selective identification of small groups of data that are outliers to broader trends is a common tactic. These approaches are subjective: they involve searching for ways to select members of a data set in such a way as to support a particular point of view, rather than to test a hypothesis. This is not Popperian science. When we reason scientifically, we begin with a hypothesis; when we test the hypothesis we also consider all of the uncertainties in experimental data. Climate data are subject to large scatter, some of them random errors and some of them (e.g. El Niño effects) intrinsic to the way the Earth’s weather systems operate. I have seen many graphs on the web pages of climate misinformation specialists in which trendlines have been added to graphs that do not fit data, or where a straight line has been plotted through a small group of points that lie on a curve. The weaponisation of outliers as “gotchas” is not scientific thinking, but anti-scientific propaganda. Science is an objective process, whereas “facts” come laden with beliefs. To test our hypotheses we need to reason deductively from a hypothesis to testable predictions, and to identify the conditions under which a hypothesis will be considered to be falsified. Such considerations are entirely lacking in the work of misinformation factories.

Reference

1. K. R. Popper, The Logic of Scientific Discovery, Hutchinson, London, 1980, p. 111.