Anthropogenic Warming Is Part of the ‘Hard Core’ of Climate Science

The party line of climate change skeptics these days is that global warming has paused, or even reversed, in the last 15 years. According to the Nongovernmental Panel on Climate Change, “Global temperatures stopped rising 15 years ago despite rising levels of carbon dioxide, the invisible gas the IPCC claims is responsible for causing global warming.” Typical denialist refusal to accept the facts, right? Perhaps not! A recent paper in Geophysical Research Letters explains, “Although the Mount Pinatubo eruption in 1991 caused a short-term reduction in TOA radiation, increasing greenhouse gases should have led to increasing warming. However, sea surface temperature (SST) increases stalled in the 2000s and this is also reflected in upper ocean heat content (OHC) for the top 700 m in several analyses.”1 Despite this apparent anomaly, however, climate scientists have not been jumping ship from their consensus position that anthropogenic global warming is occurring.

Diagram of apparent 'missing heat'. Courtesy Science, via
Diagram of apparent ‘missing heat’. Courtesy Science, via

Skeptics see this lack of response as evidence of liberal bias or even conspiracy, but I think there is a much more compelling explanation: anthropogenic warming is part of the hard core of climate science. That is, the AGW claim is not subject to revision, and so when anomalies occur—when observations fail to meet predictions—other components of climate science must be revised to preserve it.

“Hard core” is a term from philosopher of science Imre Lakatos’s account of scientific change. According to Lakatos, scientific research programs consist of a hard core of propositions that are not subject to revision surrounded by a “protective belt” of propositions that are subject to revision. These revisable propositions are also known as “auxiliary hypotheses”. When observations conflict with the research program’s predictions, it is the auxiliary hypotheses that are altered in response. In Lakatos’s account, research programs are judged “progressive” if alterations to the auxiliary hypotheses generate novel predictions and “degenerative” if they are not. Research programs that are continually degenerative ought to be, he claims, abandoned.

Faced with temperature readings that did not match model predictions, climate scientists did not abandon the AGW proposition, because it is part of climate science’s hard core. Rather, they searched for where the “missing energy” might have gone. Recently, two papers have offered possible solutions. First, the above-quoted paper from Balmaseda, Trenberth, and Kälén argues that this missing energy has been located in the deep ocean, below 700 m. Based on model reanalysis and measurements, they conclude that there is “a long-term ocean warming trend, while heating continues during the recent upper-ocean-warming hiatus, but the heat is absorbed in the deeper ocean. In the last decade, about 30% of the warming has occurred below 700 m, contributing significantly to an acceleration of the warming trend” (1754). Starting around 1998, they tell us, changes in wind patterns have caused stirring of the ocean in a way previously unaccounted for by climate models. It is this part of the climate science research program—propositions concerning deep ocean circulation—that should be revised.

Second, a forthcoming article by climate scientists Kevin Cowtan and Robert G. Way concludes that the missing heat is not missing at all, but is instead underestimated by surface measurement networks. They report, “Microwave sounding data from satellites, reanalysis data from weather models, and observations from isolated weather stations all confirm that the unobserved regions of the planet, and in particular the Arctic, have been warming faster than the globe as a whole since the late 1990s.”2 It is no simple task to measure global temperatures, partly because not all parts of the planet are equally covered by measuring stations. Cowtan and Way argue that there is a correlation between poorly-covered locations, such as the Arctic, and locations where there has been significant surface heating. Therefore, surface temperatures have been rising all along; our measurements have just failed to record this warming.

Because it was not an option for climate scientists to revise their belief that anthropogenic global warming is occurring, they were forced to search for other explanations for the apparent failure of climate model predictions to match observations over the last decade. Such explanations have now been found. Skeptics see these recent reports as ad hoc attempts to preserve an untenable hypothesis, but the work of Lakatos councils us to judge them differently: as corroborated novel predictions resulting from the alteration of auxiliary hypotheses. It is conceivable that, in the future, the current consensus climate science research program will degenerate and need to be abandoned, but there is no sign of that yet. This is just science as usual.


  1. Magdalena A. Balmaseda, Keven E. Trenberth, and Erland Kälén, “Distinctive climate signals in reanalysis of global ocean heat content,” Geophysical Research Letters, Vol. 40 (2013), 1754.
  2. Cowtan, Kevin, and Robert G Way. 2013. “Coverage Bias in the HadCRUT4 Temperature Series and Its Impact on Recent Temperature Trends.” Quarterly Journal of the Royal Meteorological Society (November).


  • Boaz Miller
    Boaz Miller Reply

    A few comments:
    1. Why is Cowtan and Way’s claim a novel prediction rather than an ad hoc explanation?
    2. The fact that a research program is progressive is a reason for scientist to pursue it (and it is also rational to pursue a degenerative research program, as long as you are aware it is degenerative), but it is not necessarily a reason to believe it, or act on it. The move from Lakatos’ epistemology to a policy-relevant conclusion is too quick.
    2. We are no longer regularly teaching Lakatos in introductions to philosophy of science. I don’t teach him in the graduate introduction I teach at Bar Ilan University, and Anjan didn’t teach him as well. I do teach Longino and Solomon. My thought is that if a graduate student wants to produce a publishable paper, it’s better for her to frame her argument within current approaches, and react to more recent materials by living people. I wonder if this is a mistake.

    • Greg Reply

      1. They aren’t novel predictions. But Lakatos doesn’t demand that the auxiliary changes provide novel predictions in the same context which forced the auxiliary revisions (that would be too stringent). Novel predictions in other contexts, using the improved surface observation network, would be sufficient to show progression

      2. I don’t see Mike making the claim that Lakatos’ epistemology (LE) provides a reason to act on or believe in AGW. He does not mention any policy relevant conclusions. Here LE is merely an explanatory device that accounts for the actions of scientists, and Mike’s point could likely be recast in terms of holism or Bayesian induction (or pick the way you think scientists reason). These other frameworks might make it more explicit why something is considered “hard core” in the first place, and would not be as subject to revision as something auxiliary (there is overwhelming evidence for AGW, and our failure to detect the missing energy is better evidence for a fault elsewhere), but LE will do.

      • Mike Thicke
        Mike Thicke Reply

        I think they could be considered novel predictions. In the first case, the ‘novel prediction’ is that the deep ocean has been heated. The scientists then go looking to see whether the deep ocean has heated, and find that it has. In the second case, the ‘novel prediction’ is that heating is concentrated in areas that have previously been poorly measured. An ad hoc modification to climate science could be something like tuning models to predict greater cooling effects from volcanic eruptions than previously thought, without proposing some way to measure the effect, or proposing some modification to data analysis methods to make surface ‘measurements’ come into accord with satellite data and model predictions.

    • Mike Thicke
      Mike Thicke Reply

      (1) I believe Lakatos would judge these additions to climate science based on whether they generate further novel predictions.

      (2) I’m a bit of an anarchist with respect to theories of science. I think Lakatos is a good emendation upon Kuhn, and I think they both capture interesting aspects of how science works. Whether he should be taught in a course that is already teaching Kuhn would depend on how much time you have in the course. I would never write a paper defending Lakatos, but that’s because POS is interested in different things now—it wouldn’t be relevant.

    • Mike Thicke
      Mike Thicke Reply

      For what it’s worth, Philosopher’s Index shows a fair number of Lakatos references post-2000. Most of them are better classified as history of philosophy, but there are still some treating Lakatos as a live option. Eg. Zenker, Frank “Lakatos’s Challenge? Auxiliary Hypotheses and Non-Monotonous Inference” (2006), Journal of General Philosophy of Science

  • Jason Reply

    Been a while, Mike. But I noticed the post on GW through Facebook, so thought I’d throw in my two cents.

    What exactly do you mean when you say AGW might not be subject to revision.
    AGW is just the two-part claim that (a) humans are changing the composition of the atmosphere, and (b) there is a causal link between the composition of the atmosphere and the surface temperature.

    I don’t think anybody – even the most of extreme denialists – challenges (a). CO2 concentrations, along with other so-called “greenhouse gases” (GHGs), are indisputably rising.

    The only debate is over (b), i.e. how increasing CO2 (and other GHG) will affect temperature. But the idea that GHGs like CO2 “trap” heat comes from some very, very fundamental physics and chemistry – physics and chemistry that seems to work exceptionally well in pretty much every other situation. Thus, unless you’re going to revise some very basic science, your only choice is to look for energy sinks/feedbacks/etc. to explain why the trapped heat isn’t causing a rise in temperature. That’s not particularly ‘ad hoc’, that’s just saying our simplified model (even as complex as it already is) still doesn’t incorporate enough of the variables to sufficient detail.

    • Mike Thicke
      Mike Thicke Reply

      Thanks Jason—

      Scientists don’t just arbitrarily decide that something isn’t up for revision. Revising the AGW hypothesis would have, as you say, major implications for atmospheric physics. Climate scientists would also have to puzzle over satellite data suggesting that there is a flux imbalance between energy impacting the Earth and energy reflecting or emitting from the Earth.

      However, I think denialists would claim something along these lines (a) Although CO2 acts as a heat trap in the laboratory, it is uncertain how it behaves in the real atmosphere, in the concentrations it is now in or maybe (b) There are many complex natural mechanisms that we don’t understand and that might affect warming.

      According to the ‘NIPCC’ ( ‘Research suggests, however, that the IPCC’s view of atmospheric CO2 as “the strongest driver of climate change” should be reconsidered if not outright rejected. The studies reviewed here find it is likely rising atmospheric CO2 concentrations will have little impact on future climate.’

  • Iain Reply

    Of course, the most common response to this denialist claim is simply to point out that it is false. The “pause” is only a pause if you take the outlier year 1998 as on trend. But since temperatures fluctuate above and below the mean all the time, this is silly. If you look at any reasonably constructed trend lines, there is no pause.

    What there does seem to be, however, is a reduction in the rate of increase – and this is an anomaly in search of an explanation. The papers you cite seem to me to be aimed at explaining this anomaly, and not the false one claimed by the denialists.

      • Iain Reply

        Mike, as I’m sure you know, there are standard statistical methods for drawing trend lines – though there is some subjectivity in the choice of method, it is not just in the eye of the beholder. The most obvious way of smoothing out annual variability to get at long term trends is to use moving averages. And when you do that, the trend is clearly continuing upward.
        My favorite response to the “no warming since 1998” claim, though, is a single gif:

        • Mike Thicke
          Mike Thicke Reply

          Of course you can use moving averages, and that is what the IPCC reports do (generally 5-year moving averages). But skeptics could (quite fairly I think) reply that a moving average will tend to disguise short- to medium- term changes. There is a theoretical assumption behind a moving average: that there is a single long-term trend that is being obscured by short-term noise. That is the assumption behind your GIF. If the deep ocean heating explanation is correct, and there has been a significant change in the rate of surface heating since 1998, then the “rational” straight line of that GIF is *wrong*, given that the underlying data doesn’t include deep ocean heating. And I would say that the skeptical “pause” trend would be “more right”, in that it sees a change in surface temperature trends starting in 1998 contained within the data.

          Since the dispute in question is over whether there is “a little” warming or “no” warming (in the previous data), subjectivity is going to play a significant role in judging what the true trend is behind rather noisy data. I don’t think you can fairly say that claims that surface warming have paused are simply false (before these recent results). If surface temperature measurements had continued to rise as climate models predicted, it would be much harder for denialists to construct a credible non-warming trend.

  • Iain Reply

    The gif uses simple regression to get straight lines, not moving averages. The point is to show that you get absurd results if you cherry pick the start and end years of your data series. It’s not that the lines all have a different slope, its that they are radically discontinuous when you try to fit them together. Any analysis that supports a no-warming claim is based on a data series that starts at 1998 and simply ignores what came before, and this is a lousy methodology.

    A moving average gives a curve, not a straight line, and will show changes in the trend fairly quickly – if there really has been no warming since 1998, then a 5-year average would have flatlined years ago. In reality, there has been a decrease in the slope (which, as I said, may call for an explanation – or may just be a statistical fluctuation), but no flatlining.

    Even if there were a way of interpreting the data that supported the no-warming position, it is clear that this would involve a methodology that departed from that of the research programme. So, within that research programme, the “pause” is not a pause, and thus not an anomaly that needs to be explained. Nothing within the programme requires any beliefs to be revised.

    We are agreed, of course, that there is no crisis, and that all of this is just science as usual.

    • Mike Thicke
      Mike Thicke Reply

      The last image in the GIF is a single straight line, that is what I was referring to.

      The whole point of those articles is that there *is* an anomaly to be explained! (Edit: And so arguments that the “pause” claimed by skeptics is merely the result of statistical cherry-picking aren’t tenable, because climate scientists agree that there was an anomaly. If the “pause” didn’t exist, there would have been no anomaly to explain in the first place.)

    • Mike Thicke
      Mike Thicke Reply

      Of course if you look at skeptical arguments as a whole, they are extremely ad hoc, selective, and deceptive. It just happens that in this case they happened to be somewhat close to the truth.

  • Boaz Miller
    Boaz Miller Reply

    The problem may be with “novel predictions” as a concept, but these don;t sound like novel predictions. This has the structure of an ad hoc codification: “we should expect X. We don’t observe X. Why? Maybe Y. Yes, Y”.
    If indeed the anthropogenic global warming hypothesis is part of the hard core, the skeptics have a point, because no observation can refute it. It may be “science as usual”, but I don’t see how this helps the case against the skeptics.

    • Mike Thicke
      Mike Thicke Reply

      Thinking about your previous comment I came across this paper today: Hunt, J Christopher. 2012. “On Ad Hoc Hypotheses*.” Philosophy of Science 79 (1) (January): 1–14. doi:10.1086/663238 (

      In this article I review attempts to define the term “ad hoc hypothesis,” focusing on the efforts of, among others, Karl Popper, Jarrett Leplin, and Gerald Holton. I conclude that the term is unhelpful; what is “ad hoc” seems to be a judgment made by particular scientists not on the basis of any well-established definition but rather on their indi- vidual aesthetic senses. Further, a hypothesis considered ad hoc can apparently be retroactively declared non–ad hoc on the basis of subsequent data, rendering the term meaningless.

      I haven’t had a chance to read beyond the abstract yet. My rough understanding of ‘ad hoc’ matches yours. However, a good scientist will adjust beliefs when confronted with unexpected evidence. So proposing Y when confronted with ~X doesn’t automatically count as an ad hoc modification. Because these modifications have testable empirical consequences, I don’t think they should count as ad hoc. But as the above paper suggests, this is all going to get pretty murky pretty fast.

      Regarding the case against the skeptics, Lakatos doesn’t judge research programs on the basis of just one modification. Rather, they are judged over a series of modifications, and we are supposed to decide whether those modifications tend to be progressive or degenerative. So, as Greg suggested above, for climate science to be judged progressive, we should expect to see further novel predictions from these modifications over time. If instead climate models are continually failing to predict future temperature measurements and climate scientists are continually looking for new ways to explain predictive failure, then we might think that climate science is on the wrong track. I don’t expect that to be the case though.

      And on the topic of whether Lakatos is relevant to our understanding of science today, I think this is a reasonable procedure for evaluating its strength as a research program. As demarcation criteria go, Lakatos’s seems to work well in this case.

Leave a Reply

Your email address will not be published. Required fields are marked *