Trust us – we're smarter than you: climate and Superfreakonomics

Posted on December 4, 2009


sstAug24-09 Back in 2005, self-described “rogue economist” Steven D. Levitt teamed up with journalist Stephen J. Dubner to write Freakonomics, a book that rose to #2 on the NY Times Nonfiction Bestseller List based largely on the controversial topics within its covers. Some of those topics included analyses of cheating by teachers, the economics of being a crack cocaine dealer, and the impact of legalized abortion on the crime rate. Levitt and Dubner (hereafter L&D) have recently published a second book, Superfreakonomics, and even before it was published it had made a huge splash in climate circles over its last chapter (Chapter 5 – “What do Al Gore and Mount Pinatubo have in common?”), the one that attempts to tackle climate disruption.

I’m greatly troubled by the content of Chapter 5, but only partly because of the many factual errors that L&D made. What troubles me the most about Chapter 5 is its arrogant tone, fed partly by a group of smart Seattle-based entrepreneurs, and the major inconsistencies that arrogance engendered. It is one thing to believe that you know more about climate disruption than everyone else when you’ve studied the field for years. It’s something else entirely to believe you know more than everyone else without verifying it first. And not only did L&D fail to verify their own knowledge about climate, they also failed to determine whether the entrepreneurs they interviewed knew more about climate than the real experts. The result was a fatally flawed Chapter 5.

L&D set the tone of Chapter 5 in the first 11 pages. They give a six-page explanation of the economic concept of an “externality” (the effect of an action that benefits the actor, but the costs affects someone else) that could have been explained as well in a couple of paragraphs. They claim that driving isn’t as bad for the global climate as eating locally-ranched beef, allegedly disproving both the “drive less” and “eat locally” green memes at the same time. And L&D label climate disruption as a “religion,” complete with a high priest (James Lovelock), a patron saint (Al Gore), “true believers” and “heretics,” and a quote from the mayor of London saying

the fear of of climate change is like a religion in this vital sense, that it is veiled in mystery; and you can never tell whether your acts of propitiation or atonement have been in any way successful.

And L&D naturally set themselves apart from those suffering from what they label as religious zeal.

L&D go on to toot their own horns when they claim that economists like Levitt

are trained to be cold-blooded enough to sit around and calmly discuss the trade-offs involved in global catastrophe….

They might as well have said “we’re better than you are because you’re ‘excitable’ and prone to ‘believing’ in global warming. We’re better than that.”

But it’s when we’re introduced to the entrepreneurs of Bellevue, WA-based Intellectual Ventures (IV) that the arrogant tone of Chapter 5 gets turned up to 11.

L&D extensively interview IV co-founder Nathan Myhrvold and former astrophysicist (and IV employee) Lowell Wood, building both men up to be larger than life. Myhrvold is described as “so polymathic as to make an everyday polymath tremble with shame” and L&D quote Bill Gates as once saying that he didn’t “know anyone [he] would say is smarter than Nathan.” With such a glowing recommendation, L&D’s statement that “Myhrvold thinks Wood is one of the smartest men in the universe” lends Wood intellectual gravitas. So does the listing off of Wood’s achievements during the development of the Star Wars missile defense system and how Wood trained under renowned physicist Edward Teller. But just because both men are smart doesn’t mean that either has any credible expertise in the field of climate disruption – Myhrvold is the former chief technology officer for Microsoft, not a climate modeling expert. Wood is a former astrophysicist who worked on lasers, not a climatologist. Yet L&D elevate both men to being the equal of an actual climate expert like Stanford climatologist Ken Caldeira.

caldeiraI mention Caldeira because he’s also an IV consultant, and he is presented by L&D as the climate expert that he actually is. But L&D’s purpose in interviewing Caldiera appears to have been to burnish of Myhrvold’s and Wood’s climate credentials. Caldeira’s work is only briefly mentioned , and in a way that either fails to provide the necessary context or, in some cases, even is applied to the wrong context. For example, Caldeira is quoted as saying that trees planted in the wrong place can increase climate disruption because tree leaves may be darker than the ground was. The context of this quote in Chapter 5 is that “many climate disruption memes are wrong,” but L&D didn’t bother to point out that planting trees in other places would reduce climate disruption. Doing so would have reduced the impact of L&D’s preferred, and opposing, frame.

The other way Caldeira’s reputation is used to burnish the qualifications of Myhrvold and Wood is by lending credibility to their climate engineering ideas. L&D point out that several years ago, Wood proposed pumping sulfur dioxide (SO2)into the stratosphere as a way to cool the Earth, and after running simulations, Caldeira admitted that it might work and that additional research was warranted. L&D use this apparently unimpeachable recommendation as the jumping-off point for proclaiming throughout most of the rest of Chapter 5 that climate engineering is the wave of the future. What L&D neglect to mention is that Caldeira has actually come out against geoengineering using SO2 at this time for reasons entirely unrelated to global temperatures – it only treats one of the symptoms (rising global temperatures), does nothing for ocean acidification and other carbon dioxide (CO2) related problems, and is therefore too risky. And it doesn’t help that L&D baldly misrepresented Caldeira’s actual statements and beliefs regarding climate engineering, as the prior links illustrates.

But based on Caldeira’s presence at IV, L&D remake Myhrvold and Wood into experts on geoengineering and climate. And so Wood’s statements on climate modeling are given the same weight as Caldeira’s using guilt by association. Or rather in this case, expertise by association.

Wood is quoted by L&D as saying that climate models are

crude in space and they’re crude in time. So there’s an enormous amount of natural phenomena they can’t model. They can’t do even giant storms like hurricanes.

Wood also claims that the models are all tuned in a similar fashion in order to keep the research money flowing and that current models don’t know how to handle water vapor. Wood’s claims have several problems. They’re presented in a way that’s misleading, they’re not given any actual scientific context, or they’re outright wrong. And because Wood, the former astrophysicist, is a “climate expert by association,” we’re led to believe that these claims are reasonable.

First, while L&D point out that computing power limitations drive climate modelers to make simplifications including lower spatial and temporal resolution than the modelers would prefer, those limitations don’t make the models “crude.” Neither does the fact that the scientists doing the modeling admit that they don’t yet agree on the best ways to model clouds and aerosols (two areas that scientists have made significant advances in since the publication of the IPCC AR4 in 2007). But neither of these issues mean that the latest models are crude. In fact, a quick analysis of the models illustrates that, contrary to Wood’s claims, the existing state-of-the-art general circulation models (GCMs) are quite sophisticated given the computational limitations.

The spatial resolution of an average IPCC AR4 climate model is about 55,350 square kilometers, or a square 235 km on a side (the highest resolution models resolve areas as small as 100 km on a side). For comparison, this is about one fifth of the state of Colorado, which has at least two different climatic zones. Sampling theory (Shannon’s sampling theorem) says that, in order to perfectly reproduce a signal, you need to sample it at twice its frequency. Applied to an average GCM from the AR4, the smallest spatial pattern that could be accurately resolved without help would be 470 km or larger. While there are regions around the globe where you can move through multiple climatic zones over the course of 470 km, they’re not exactly common. And while the average GCM wouldn’t accurately model these areas, climatologists have developed what are called “parameterizations” to represent small scale effects like cloud formation, aerosol effects, and convection that vary locally within the model spatial cells. And generally speaking, it’s not the spatial resolution that presently limits the models’ accuracy (or increases the models’ uncertainty), but rather how the GCMs handle these intra-cell parameters.

The GCMs don’t yet have enough spatial resolution to handle multiple major climatic transitions in small areas, but the models are presently accurate enough to draw conclusions for large climatic regions like “the desert Southwest,” or “the Great Plains,” “the Indian subcontinent,” or “the African savanna.” And while it’s unlikely that any model will ever be able to predict microclimate variations that can change from one kilometer to the next, there is no reason to believe that such high spatial resolution would ever be required for a climate model. So from my perspective, the word “crude” applied to the models used in earlier IPCC Third assessment reports rather than to the latest generation of GCMs. The figure below illustrates how the latest models compare to earlier generations.


Furthermore, sampling theory suggests that Wood is also wrong with respect to the temporal resolution of climate models. Climate models attempt to project the climate, which is itself a long-term average taken over years or even decades. Sampling theory says that we only need to model at twice the frequency of the fastest changing event we care about. One of the fastest changing climate events is El Niño/La Niña, but even at its fastest, it takes weeks to change significantly. This suggests that that sampling once a week or so would be enough, but due to noise and other limitations on sampling, it is usually best to sample more often than the minimum rate. Most climate models sample every 12 hours. If El Niño/La Niña is one of the fastest major climate changes that climate models need to model, then sampling twice a day represents 28x oversampling (assuming that El Niño/La Niña doesn’t pop up any faster than over the course of two weeks). It’s unlikely that the temporal resolution of climate models will need to improve much beyond where it is today. As a result, it’s clear that GCMs are pretty refined with respect to their temporal resolution.

Second, climate models don’t need to model any single storm, not even a hurricane, in order to be broadly accurate. Any individual storm has such a short term effect, and the effects are so localized, that they’re essentially weather, not climate. Four years after Hurricane Katrina hit New Orleans, the city is recovering and the parts of Louisiana and Mississippi that were damaged are physically different, but not so much that it qualifies as a local change of climate. The aggregate energy transfer from all hurricanes is important to climate disruption, but no single storm is going to transfer so much energy from the tropics to the subtropics to be critical. Remember – climate is an average over decades, not years or months or weeks. So the fact that no climate model presently models even the largest hurricanes is a distraction, not a real criticism.

Finally, Wood’s claim that models are tuned alike for research funding reasons runs counter to a social dynamic in the scientific community that values original research and intellectual competitiveness over the “same old, same old.” Most scientists live for discovering something new and having the opportunity to research it. Most scientists also enjoy finding errors in another scientist’s work and correcting the error. This is especially true of scientists at the top of their fields. As a result of these dynamics, most scientists would be more likely to tune their models to be different in order to garner attention from their scientific peers – if there was a physically legitimate way to do it. Furthermore, given that climate disruption will very likely become a nightmare for billions of people, discovering that climatologists had got it all wrong would greatly boost the scientist’s, the host organization’s, and the host nation’s prestige. So climate modelers have an incentive to tweak their models as much as scientifically justified in order to generate something new, publishable, and in opposition to the prevailing scientific understanding of climate disruption. The fact that all the models function similarly and reach similar conclusions suggests that, even with this social dynamic at work, scientists and labs haven’t been able to find anything major wrong with the science embedded in each other’s models.

There’s an irony here, though. Myhrvold and Wood, “experts by association” with Ken Caldeira, don’t think that climate modeling is accurate enough. Yet Caldeira’s climate expertise is based significantly on his experience modeling the effects of CO2 and other greenhouse gases on plants, the ocean, and global climate in general. L&D only implicitly acknowledge this fact:

Caldeira ran a climate model to test Wood’s claims [about geoengineering].

Caldeira’s study showed that doubling the amount of carbon dioxide while holding steady all other inputs – water, nutrients, and so forth – yields a 70 percent increase in plant growth, and obvious boon to agriculture.

It is, after all, Caldeira’s models of the effects of pumping tons of SO2 into the stratosphere to cool the globe that forms the basis of L&D’s main focus in Chapter 5 – geoengineering as a cheaper alternative to slashing CO2 emissions.


In case you’re unfamiliar with the term, geoengineering is essentially using a technological, engineering-based solution to solve climate problems that were created by human civilization. L&D believe that geoengineering is the cheapest and most effective method to addressing climate disruption, and it’s this belief that drove them to talk to Myhrvold and Wood at IV. The two IV men believe that, for less than a billion dollars, they can built a big straw that would deliver tons of SO2 directly into the stratosphere. Once there, the SO2 would chemically react to form aerosols that scatter and reflect solar energy back into space before it could be absorbed by the Earth. The basic science underlying this idea is pretty solid – large volcanic eruptions blast SO2 into the stratosphere where the sulfur aerosols naturally cool the globe for a few years. So it’s reasonable to believe that it would work.

Except as Myhrvold and Wood point out, there are a few problems. Minor problems, if you believe L&D. For one, it’s not possible to know at this time what effect all that extra sulfur in the stratosphere would do. Myhrvold and Wood also worry that it would make people complacent about continuing to emit GHGs instead of using the sulfur straw as a short-term solution. And they are worried that the technological solution is so cheap that it would be a political problem – could a country with a straw hold the world hostage by threatening to turn it off, or could it get turned up higher in order to cool and dry the climate of another, rival nation? These are potentially serious issues, but L&D largely ignore the broader political implications, giving the problems only three paragraphs after spending ten pages building the case for geoengineering. And L&D don’t even mention a problem that a sulfur straw would do nothing to fix, and what Caldeira is perhaps best known for – ocean acidification and the associated damage to marine ecosystems.

The bulk of Chapter 5 is spent talking up geoengineering – how it’ll work, how it’s “harmless” because the climate would just revert to its prior state if the straw were turned off, how it’s cheap. But there’s another irony that L&D don’t appear to recognize, and this one cuts to the bone of the geoengineering argument. Myhrvold and Wood, and by extension L&D, are proposing that we pumps megatons of SO2 into the stratosphere in order to cool the Earth, but the only way to make this plan politically viable would be to rely on projections from climate models – the very same climate models that have been dismissed by the authors and the IV crowd.

At this point, scientists don’t know what pumping hundreds of kilotons of SO2 into the stratosphere continually for decades or even centuries would do to the atmosphere. With less energy falling on the Earth in general, the water cycle would change and as a result, rain and snowfall patterns would change and would create water winners and losers. Projecting how and who would fall to climate models.

Stratospheric chemistry isn’t a straightforward science – it relies on atmospheric concentrations of ozone, oxygen, hydroxyl radicals (which are also necessary to convert methane to CO2, among other things), CO2, nitrogen, the availability of solar radiation in various wavelengths, humidity, prevailing winds, and so on. Projecting how the atmosphere will react to all that SO2 being dumped continuously into the stratosphere will fall to climate models.

A separate chemical effect that needs to be considered is the effect of SO2 on ozone depletion around the world. Lower ozone levels in the stratosphere may result in more skin cancers, cataracts, and may result in higher incidence of certain diseases. In addition, more ultraviolet radiation reaching the surface will have an effect on ecosystems and species as well as agriculture around the world. Climate models would be necessary to determine how these effects will vary seasonally as well as regionally.

Even the claim that the sulfur straw would be “harmless” because it could be turned off requires verification – and the only way to verify this claim before turning the straw on is with climate models.

If the models truly are as bad as Myhrvold and Wood claim, then we can not rely on them to project the effects of geoengineering. And if the models are good enough to accurately model the effects of geoengineering, then perhaps we should trust what they tell us about addressing global climate disruption, namely that cutting greenhouse gas emissions is the best way to curtail overall climate disruption.

Finally, Myhrvold claims to be advocating for a go-slow approach to creating his sulfur straw, opting for more research and development. He claims that

it’s like having fire sprinklers in a building. On the one hand, you should make every effort not to have a fire. But you also need something to fall back on in case the fire occurs anyway.

The problem is that Myhrvold dismisses the GHG-emission cutting goals of climate disruption activists so vehemently that it’s hard to take seriously his claim that he just wants more research and development. He says that

[global warming activists] are seriously proposing a set of things that could have enormous impact – and we think probably negative impact – on human life. They want to divert a huge amount of economic value toward immediate and precipitous anti-carbon initiatives, without thinking things through. This will have a huge drag on the world economy. There are billions of poor people who will be greatly delayed, if not entirely precluded, from attaining a First World standard of living.

Myhrvold apparently hasn’t heard how seriously climate activists are pushing for technology transfers and direct financial assistance to poor countries so that they don’t have to use high-carbon energy sources to improve their standards of living. Nor has he apparently read about the many studies that say the costs of addressing climate disruption will not affect the global economy as much as he seems to believe. Myhrvold instead seems to be more motivated by money than by research. The fact that IV has been accused of being “patent trolls” doesn’t help his case – IV is Myhrvold’s company, after all, and L&D mention the “patent troll” issue themselves.

It’s clear that Levitt and Dubner are both smart men and that they found other smart men in Myhrvold and Wood. The problem is that all of them are so convinced that they’re the smartest people around that they can’t seem to realize that they’re mired in myopia. Levitt and Dubner are certain that Myhrvold and Wood have come up with “The SolutionTM” in the SO2 geoengineering project that they don’t notice, or don’t care, that the arguments in favor aren’t logically consistent and don’t stand up to basic scrutiny. Myhrvold and Wood are so sure that they’re right about climate models and geoengineering that they seem to blow off the serious problems and ignore the experts who actually know what the hell they’re talking about – experts like Ken Caldeira and his fellow practicing climatologists. And none of the men – not Levitt, not Dubner, not Myhrvold, and not Wood – seem to realize that they’re making statements that are counter to the best available science and, in many cases, could have been detected as wrong with some simple and basic fact checking.

Climate disruption is a huge issue, and geoengineering can’t be ignored as a potential insurance policy. But it’s presented as the first line of defense against climate disruption instead of a fall-back plan. And as much as Levitt and Dubner try, their arrogance wrote checks that their intellects couldn’t cash.

“Trust us – we’re smarter than you” doesn’t work when you make it so abundantly clear that you’re not actually smarter after all.


There are legions of critiques of the climate disruption chapter of Superfreakonomics on the Web. Here’s a short list of some of the better ones:

Image Credits:

Posted in: Uncategorized