Nimble skepticism (Updated)
The Karl et al. study highlights something I've been thinking about lately. I don't know if the Karl paper is important, good, or bad. It claims to debunk the slowdown in surface warming. Other papers will claim the opposite. This won't be the end of it, but imagine that it was – imagine that we saw a decisive breakthrough in climate science, or a series of them, that debunked the slowdown, and another body of work that settled on 3.1 °C for ECS.
If you're a climate skeptic, or better yet, a person who is currently skeptical of burdensome future human-caused warming, you should be ready not to be.
You're presumably skeptical because of issues you see with the evidence, levels of certainty or uncertainty, perhaps features of climate science and its methods or predictive track record.
All of those issues can in theory be resolved by new evidence, or new types of evidence and methods.
If you're a skeptic or a lukewarmer I wouldn't assume that the evidence is going to roll your way. (ECS seems to have had a bit of a downward run over the last few years, but who knows.)
You should be ready for anything, evidence-wise. Don't get too comfortable. Life is full of surprises, and so is nature. Earth's climate answers to no one. It will do whatever it does. It is completely uncoupled from our desires, agendas, elections, ideologies, beliefs, arguments, pride, etc.
I don't think people should hitch their ideological wagons to the behavior of a planetary climate system. That's odd. This applies to everyone of course.
1. Measuring surface temperatures sure is complicated. In fact, as Gavin Schmidt said, global mean temperature isn't measured per se. It's estimated. Scientists can come along in 2015 and redo the temperature estimates for the past several decades. That's strange. Most sciences don't work that way, don't have this constant process of re-estimation of past measured variables. If scientists can redo temperature estimates in 2015, they can presumably redo them in 2016, and 2017, and perhaps in 2023. I think we need to understand this better. Maybe they're closing in on maximum feasible bias reduction and we won't see much adjustment in the future, but this should be explained.
2. Knowing about or believing in human-caused climate change is nothing like knowing about gravity or that the earth is not flat. This is not like looking at something and seeing that it's there, or figuring out the horizon, or dropping a ball. It's so much more complicated, driven by inferential estimates and wicked statistics. Climate activists should be much less mean to skeptics, and stop trying to treat this issue as though people are obligated to march to the claims of a young, complex, and revisionist science. I don't think people are obligated to believe in things they cannot observe or confirm directly except in special circumstances. Believing in everything the media folds under "science" is probably unwise, and it's unclear how a rational knower is supposed to navigate our media/science culture. I don't have any kind of prescription.
The eternal caveat applies: The science is just the science. It doesn't have to matter to you, not politically, not philosophically or personally. People get to choose their political philosophies and ethical systems, and you don't need to catastrophize any arbitrary level of future adversity if you don't want to. You don't have to care about the science of obesity, or the science of testicular cancer, or the science of sadness, or an increase in storm count. There are lots of things a person could choose to care about or not care about, and it's unclear why anyone has to care about any particular science or diffuse future risks.
There's a mindset in modern politics that wants to "Do Something!" about everything. I think we'll find that some of it is driven by affluence – that people worry about more things, smaller things, the more affluent a society becomes. In any case a person's quality of life is powerfully shaped by their perspective and framing – we know how profound that can be, the glass half full vs. half empty mindset. It's strange that we never seem to apply that wisdom to environmental issues. You could put me on the gulf coast and jack up the hurricane count by a third, and I wouldn't care if I had someone to love and books to read. There are so many other things going on in a human life than weather and sea levels, so much more beyond material and economic concerns. Some people (not me) would move to Mars if they had a chance, even though the climate would be so hostile that they'd be confined to quarters.
That's not just about affluent American space geeks – most people in the world don't care about climate change, even when forced to choose six "priorities" in a biased UN survey. The UN wouldn't let me participate in the survey because I couldn't find six things on their list that were priorities to me. The list is framed from a top-down, government-centric bias that enjoins people to express vague wishes for "better" roads, health care, food, and so forth. They don't offer priorities like "end the drug war", "deregulate immigration", "cut taxes", "eliminate income taxes", "free market healthcare", or "get the UN out of my life." It was designed for the UN to be able to say that adults around the world want governments to deliver things like "affordable and nutritious food" and "action on climate change". The items and forced choices will systematically discriminate against non-leftist participants, as well as people who don't think there are lots of problems they need authority figures to solve – such people won't even be allowed to submit their answers. The stated goal of the survey is "that global leaders can be informed as they begin the process of defining the new development agenda for the world", what economist William Easterly would call the "Tyranny of Experts". And still, even with the rigged design, people don't choose climate change.
Relatedly, Bjorn Lomborg was correct to say that Pacific Islanders don't care about it. (Choose Oceania in the dropdown.) Pretty much no one does. It doesn't make the cut on any continent or region that they list. As for affluence, start with the Low HDI countries option and work your way up – the poorest countries care the least. As per my hypothesis above, more people care as you work up HDI, yet it never makes the cut even in the richest. I didn't know that until today. I thought environmentalism was more popular than this, but I now realize that I probably just know a lot of environmentalists.
Correction / counting
In one of my posts on the Cook fraud, I listed 19 social science, survey, marketing, and opinion papers that were coded as "climate papers" endorsing of anthropogenic warming. I made a mistake listing this one:
Vandenplas, P. E. (1998). Reflections on the past and future of fusion and plasma physics research. Plasma physics and controlled fusion, 40(8A), A77.
It was coded as mitigation but not as endorsement. My apologies.
Since we have a 1:1 replacement policy at JoseDuarte.com, I'll give you another paper that was indeed counted as endorsement:
Larsson, M. J. K., Lundgren, M. J., Asbjörnsson, M. E., & Andersson, M. H. (2009). Extensive introduction of ultra high strength steels sets new standards for welding in the body shop. Welding in the World, 53(5-6), 4-14.
In fact, this one was counted as a "mitigation" paper and as "explicit endorsement."
As I said in that post, there are lots more where that came from, lots more that I haven't listed. I don't provide them all for two reasons: 1) I want to get people thinking about validity and discourage something I saw after I posted that list of 19 papers – counting and recomputation of meaningless data. 2) I'm saving some for the journal article.
The Cook study was invalid because of the search procedure, the resulting arbitrary dataset that gave some scientists far more votes than others, the subjective raters having a conflict of interest (which is unheard of), the basic design, the breaking of rater blindness and independence, etc.
That means there is nothing to count, no percent, not a 97%, not any percent. You could recompute the percentage as 70% or 90% or 30% if you went through the data, but it would still be meaningless. There is no method known to science by which we could take their set of papers and compute a climate science consensus. I need to do a better job of explaining what it means to say that a study is invalid. People have this instinct to still play with data, any data, because it's there, like Mt. Everest. It's an unfortunate artifact of human nature and first-mover advantage, especially in cases where lax journals don't act swiftly to retract.
As always, I want to stress that the study was a fraud, and that this is a completely separate issue from validity. I always remind people of this because I think it would be irresponsible to pretend that is wasn't. They lied about their method. They claimed their ratings were blind to author and independent – they routinely broke blindness and independence in a forum with other raters. Lying about your methods is fraud. That alone makes the study go away. There's no counting to be done. The above welding paper is an example of the third act of fraud – saying that these were climate papers. There are all kinds of these absurd papers in their 97%. This was never real. There was never a 97%.
I heard someone say something like: "If you don't like a study, run your own to debunk it." That's a common outlook in science, and it's good advice in most cases. In this case, it's inappropriate, bad scientific epistemology. Invalid studies, certainly fraudulent ones, never impose a burden on others go out and collect data to debunk it. Invalid research should simply be retracted and everyone should carry on with their lives as if the research never happened, because in a sense, it didn't. Fraud of course should be retracted with prejudice. In such cases, there's nothing to refute or debunk with new data – you'd be swinging at air. If we want to know the consensus, we already know it from quality surveys of climate scientists (it's 78 - 84%), so running around doing another study to "refute" the Cook study would be silly (especially a study based on subjective ratings of abstracts.) Sure, some people believe the 97% figure right now – Cook and company were good at the media angle. That will be corrected in time, this study will be dealt with, and I expect the people involved will face appropriate consequences.
A few days ago on Twitter I said that climate skeptics needed to distance themselves from the conspiracy theorists and harassers. This upset a dozen or so climate skeptics and led to a flurry of rebukes that I had trouble following due to their sheer volume and the constraints of a 140-character medium. I can't coherently respond or clarify my thoughts on Twitter itself, so I'll do so here.
Since joining Twitter, I've posted variously on science, social science, climate science, statistical methods, fraud cases, interesting articles in the Atlantic or New Yorker, and the catch that was definitely a catch.
Most of my tweets (oh God) on climate issues over this period would have been more satisfying for climate skeptics than environmentalists. The Cook 97% fraud case has been a popular issue with skeptics, and understandably so. The climate science consensus has been systematically overestimated by untrained researchers, fraud, invalid methods, failure of peer review, and an ancient enemy of science that I will elaborate upon in upcoming peer-reviewed literature.
I've occasionally reminded people that I'm not a climate skeptic, and I try to cleanly separate climate science from estimates of the consensus from policy issues and so on. These are all very different things to me. Because I've called out the Cook fraud, people seem to have sorted me into the skeptic camp, including skeptics themselves. That's a mistake.
My perceived scolding of skeptics was prompted by this:
I know of no body of evidence that would lead a reasonable person to conclude that climate science is the most massive scientific fraud in history.
This whole conspiracy issue has come up a lot recently as a potential personality disposition. I think it's worth lingering on what would have to be true for the science of human-caused climate change to be a massive fraud.
First, a lot of people would have to be in on it, virtually all climate scientists that work on the issue and the boundaries of the issue. I think for that to be true would in turn require both an unrealistic model of human psychology and an unrealistic model of how science works.
If it were a fraud, one ethical and capable climate scientist could blow it open. Even a capable outsider like Steve McIntyre could blow it open. What do I mean? If CO2 doesn't cause warming, or if the true value of ECS was 0.1 °C or something, and the field was hiding this fact, this would be discoverable. Someone could report their analyses revealing the truth.
Assume that they couldn't get into journals because the field is a fraud and they censor any off-message findings. They could post it online. Most people wouldn't believe it at first, would do the whole source/peer-review harrumphing. But it would certainly circulate among climate skeptics and eventually climate scientists would have to confront it. Republicans in Congress would start asking them about it and so on.
Has that happened? Not to my knowledge. Where is there a finding or analysis that debunks the AGW hypothesis?
Instead, all we see are lower estimates of human forcing, ECS, model quality, etc. When Judith Curry argues against climate science orthodoxy and overconfidence, she's not saying it's all false. Her scientific work offers lower estimates of ECS, for example (Lewis and Curry, 2014). Steve McIntyre has never argued that the hypothesis is false, or offered any debunking of it. His work has been focused on specific issues with specific papers and methods. If it were all a hoax, I think McIntyre would have been both inclined and able to expose it some years ago.
If this is about disagreeing with the magnitude or confidence of prevailing climate science estimates, for example where they project a 3 or 4 °C warming and you think it's 1 or 2 °C, then that's not a fraud issue. That's disagreement. You can argue bias, tribalism, or political ideology as factors driving their estimates, but that's a long way from fraud.
Fraud is willful deception. My vehemence against the fraud cases I discovered should not be confused for a loose definition of fraud. When I say "the Cook fraud", I mean concrete acts of fraud, like claiming this:
When they did this:
They lied. The ratings were not independent. They asked each other for feedback on their ratings, undermining both the validity of the ratings and the ability to compute rater reliability.
Authors and journal identities were not hidden, which was essential for a subjective rating study – they were freely disclosed. They routinely broke researcher blindness, and routinely outed scientists who were friendly or hostile to their cause, sometimes even by "smell".
In fact, the entire papers were freely distributed, shattering any notions of blindness. The first author himself broke protocol and blindness, identifying the authors of papers. And they all worked at home, so they could pull up the whole paper and break blindness anytime they wanted.
We would panic if researchers broke blindness in a biomedical study – e.g. the person handing out placebo or treatment pills to patients. If the researchers in a double-blind study knew what condition the patient was in, what kind of pill or other treatment they were giving them, it could affect the study. They might interact differently with patients if they knew their condition, and those differences could have downstream consequences, like for example the patient inferring what group they were in from the researcher's confidence or other cues. Or something about the researcher's administration of more complex treatments could change were blindness broken. In any case, if a biomedical study claimed that the researchers were blind, and it turned out they weren't, that study would be retracted in a heartbeat. Everyone would get it.
This is much worse. This is a subjective rating study where the researchers read textual material and made complex inferences from it. Given the nature of the topic, blindness to author and journal was critical, as would be the procurement of qualified neutral raters situated in a controlled environment on campus, not activists working from home. Breaking blindness in this kind of study is much more serious than for a pill handler. And lying about researcher blindness is the crux of the fraud issue in both cases.
That's what fraud is. They knew they had this forum where they routinely broke independence and blindness, yet claimed their ratings were independent and blind.
Show me that kind of fraud in climate science, and then we'll talk. Bias isn't fraud. Disagreement isn't fraud. Lying is fraud, but you can't just assume that people who disagree with you are lying.
People on Twitter complained that I had called out skeptics unfairly, that climate activists or scientists behave worse. I don't like that kind of blame shifting.
First, we have to keep in mind that human communication operates under lots of constraints. It's sequential and somewhat discrete. We can usually only say one thing at a time. This is especially true on Twitter. It's not reasonable to expect that when I criticize skeptics I'm also going to put it in context by criticizing environmentalists. Grown-ups should be moderately tough and resilient. Ethically, we should be open to criticism of our groups, our friends, ourselves. We know, a priori, that some of that criticism will be true (over the course of a human life.)
There are features of human language that shape the ways we communicate on these and other issues. The content of present communication is usually much more salient than past communication, past experiences, knowledge of a person's character, etc. And that's just one class of constraints. For the last two years, I've thought about how we could create new languages that would enable humans to communicate classes of concepts that are difficult or impossible with current languages, as well as overcome some of the temporal and salience issues. Linguistics, constructed languages, and human concept-formation and reasoning are deeply interesting topics. Eventually I think we'll have better ways of communicating, including constructed languages, but that's a very long-arc projection (50 - 200 years). It's useful to step back and consider the constraints in how we communicate with each other. That's a tangent, but some of these constraints impact our increased political polarization. Another tangent: It's fun to think about less ambitious ways to improve communication, like evolutions of Wolfram Alpha's Computable Document Format.
José L. Duarte
Social Psychology, Scientific Validity, and Research Methods.