I should shy away from controversy more often, but that doesn’t seem to be in my nature. The following essay/article is just one among many examples where I have failed to follow that sage advice. In what follows, I say what I mean, unfiltered by what is considered popular or acceptable. I’m not an “expert” in this field, but given that the experts treat the topic with religious zeal, I don’t have much faith in them. I only have room for one religion in my life. The rest needs to support it’s own weight.

Read at your own peril.

Why Write This at All?

Among the many things I do at work these days is oversee work done by several student hires (engineers in training). As part of that, I ask them to ask me any questions they have when they submit their weekly progress reports. In that request, I don’t limit the questions to work or engineering related topics, and I quite enjoy the opportunity to delve into topics that aren’t often on the menu in the venerable halls of the Space Dynamics Laboratory. I’ll generally spend a lunch hour responding to the questions that aren’t necessarily work related, but this time I got a question that I don’t believe I can do justice to without taking more time than I can justify during the work day. As a result, I’m posting the answer here where I can put it together at my leisure while sitting in my home office.

The question was pretty simple, and basically asked me about my views on climate change. I had voiced some skepticism about how “settled” the science was, how much we actually know about the “settled science,” or how effective the dogmatic prescriptions for “solving” the crisis could be. I don’t really remember the exact topic, but this student was surprised that someone with my education and background was skeptical in that manner. As a result, he asked to understand better.

Preliminaries

To avoid any simple misconceptions about what I do and don’t believe or understand, I’ll state some fundamental positions that I accept as accurate for current purposes. They fall into two categoreies: things that I find morally correct and don’t expect to require scientific rigor, and things that I accept courtesy of the quality of the theory and experimental evidence behind them. It’s a high bar for non moralistic stances, namely: scientific models that predict a specific describable, observable, and testable behavior that are coupled with robust experimentation that consistently observes the predicted behavior.

• We have a moral responsibility to be wise stewards of the resources that constitute our environment (one of my moral absolutes).
• We have a moral responsibility to minimize human suffering to the extent we are able (another of my moral absolutes).
• Human activity impacts the “environment”
• Human activity is increasing the concentration of CO2 in the atmosphere relative to concentrations extant in the time period leading up to the industrial revolution.
• CO2, courtesy of it’s vibrational and rotational quantum modes, absorbs infrared energy and retains it for a time as heat (kinetic energy of the molecule). This meets my understanding of what it means for something to be a greenhouse gas.
• The “climate,” if defined by global average temperatures, changes.

So Why Don’t I Buy The Alarmism?

The students who I answer questions for will recognize by now that almost every answer has some form of “it depends” built into it. The world is far to complex for almost any honest, nontrivial, and interesting question to be answered by a simple definitive statement. Why I don’t buy into the apocalyptic climate alarmism isn’t exactly an “it depends” answer, but it’s close. It’s more of an “it’s complicated” answer.

It’s Complicated

The first and foremost reason I can’t sing along with the doomsayers ultimately comes down to the fact that the universe, and everything in it is complicated beyond human comprehension. That includes the “climate” we are trying to save.

The deeper you look into the microscopic, molecular, atomic, and sub-atomic worlds, the more unbelievably complex, finely balanced, and poorly understood it gets. The further out we look into the cosmos, the more complex, finely balanced, and poorly understood it gets. At every level, the more you learn, the better you understand how little we actually know.

A simplified version of this phenomenon is my progression in understanding from being a typical “know-everything” teenager to what I am today (a cynic). When I was 18, I “knew” many things (my parents would say I “knew” everything). I didn’t hesitate to speak in absolutes. I was frustrated with anyone who didn’t see things my way. My parent’s “just didn’t get it,” and I thought they should pay more attention to what was happening around them.

By the time I was 25 and in graduate school, I started to realize that some of the things I “knew” were simply assumptions based on an incomplete world view and set of information. But I still “knew” a lot, and was even more confident in the things I thought I knew courtesy of my enhanced education and the accompanying theories, experiments, and confidence of the academic world.

By the time I finished my PhD, I started to understand that most of the things I “knew” were ultimately founded on assumptions, simplifications, approximations, and assertions that happened to be consistent with (many, but not all) observations, but that weren’t actually valid or provable at the end of the day. In fact, I ultimately found out that many of the underlying assumptions that under-girded my prior confidence were demonstrably false when examined closely. They just happened to work well enough for the applications at hand to make them worthwhile.

Now, well into middle-age, I have few if any absolutes left that aren’t moral in nature. In terms of science, I have accepted that we can “know” nothing beyond a statement that observation is consistent with predictive models. And even then, we can only claim that knowledge in the relatively rare cases where we can make a definite prediction, then design experiments that generate observations that match the prediction.

The world according to an 18 year old me was pretty black and white. It contained few shades of gray, and didn’t admit much room for uncertainty. The scientific world as I know it now (middle age) is nothing but grays and uncertainty, but contains models that happen to be useful under correct conditions and within a limited scope. At the end of the day, though, everything is an approximation of a complex system. And no model we can conceive is capable of correctly identifying and/or determining the significance of all the potential inputs for real (complex) things.

One of the harshest corrections to my youthful confidence was the realization that outside of carefully contrived and controlled experiments that could we watched only imprecisely (currently believed to be lower-bounded by the Heisenberg uncertainty principle), everything around us is influenced by more factors than can be accounted for. We must approximate and simplify everything in order to describe it. Even then, to be honest with ourselves, we have to describe it statistically. Every interaction between two things at any level, can only be dealt with through approximation and statistical description.

For simple systems with only a few inputs (like experiments in a physicist’s lab), we can describe the statistical model reasonably well in a tractable form. For simple systems on a scale where quantum effects don’t matter so much, we can approximate away the inherent randomness and write deterministic equations. However, as the dimensionality of the problem increases (we identify more factors that influence the outcome), our ability to describe even the statistics falls apart and we are forced into making progressively grosser (and more inaccurate) approximations and simplifications to keep the math doable. The further we are from the low-level physics-based descriptions of how things interact, the less confidence we can have in the results of the approximation.

Climate science is just one discipline where I apply this kind of skeptical approach. Because it is not exempt from the universal, climate science is built upon a stack-up of approximations, guesses, generalizations, and assumptions that, in the end, make the results doubtful at best and comprehensively stupid at the worst. I’ll provide a rationale for claiming this complexity and associated approximation and inaccuracy later, but for now I state it as something as close to fact as my prior statements allow.

In the case of a system as intricate, expansive, and infinitely complex as the “climate” or “environment,” we can’t even begin to comprehend, much less model, all of the linkages, dependencies, correlations, interactions, constituents, feedback mechanisms, and other factors that contribute to the behavior of the thing as a whole. So how can we honestly make predictions, design countermeasures, and forcefully extract resources from the unwashed masses to implement economically catastrophic measures that are supposed to prevent outcomes in an ultra-complex system we don’t actually understand? How can we have any confidence in the results of the prescribed measures given that they are based on models that have never accurately “predicted” anything other than historical data? How can we be so ignorant to believe that we actually understand any of it to a level that justifies the kind of apocalyptic hysteria that abounds? The answers to these questions are unsatisfying. Yet, we are told in loud voices with religious zeal that we must destroy the world as we know it to prevent a cataclysm, and do so now!

Every climate model I’ve encountered necessarily makes gross simplifications to accommodate computational and comprehensional limitations. In so doing, they disconnect feedback mechanisms, smooth out significant perturbations, work with very sparse data, uncritically inherit limitations of finer-scale models, and generally wipe away the complexity.

As a rough example, if you were to model the atmosphere over a 10 sqKm village to a height of 10km (about 30,000ft) in a 1cm grid spacing, you would have a 10^15 cells of atmosphere. If we assume that each cell only interacts with the ones immediately touching it, that results in a sparse tensor with 6 x10^15 elements. If each element could be accurately described by a single 16-bit integer, we’d need 2 petabytes of memory just store the state of the model, 12 petabytes to store the tensor (interactions), and a whole lot of time to iterate over the entire thing (about 24petaflops/time step if the interactions were simple and linear — spoiler, they aren’t).

To make it worse, the time-step for models can only be on a scale much smaller than the time scale for change within one of it’s cells. That means that the model must be recomputed multiple times a second on that 1cm^3 grid. Timelines on the order of decades would take centuries to compute even on modern supercomputers.

If that weren’t enough of a problem, there are more. Fluid and energy dynamics operate on scales much smaller than 1cm^3 requiring much finer grid spacing on the order of millimeters (a 1000x increase in data) or even smaller. The atmosphere is much thicker than 10km. In fact, it’s likely that effects out to the ionosphere (50-1000km) are relevant which adds a further 5-20x increase). Finally, my little 10km^2 postage stamp of a village is smaller than the typical grid spacing for atmospheric models. It would take about 1.5e^23 of these hypothetical villages to cover the globe. All told, that would result in something like 6×10^42 bytes of memory just to store the model. There isn’t enough storage in the universe to even hold that model. And you would wait your entire lifetime for the worlds biggest supercomputer to make a single time step. All climate models are necessarily so simplified as to be effectively divorced from the physics they claim to be based upon.

Being complicated isn’t enough though…

The complexity inherent in something doesn’t necessarily mean that we can’t get useful predictions out of seriously simplified models. The predictions of Maxwell’s equations were validated by Heinrich Hertz and many others, and ultimately led to the vast world of radio communications and other life-altering technologies. Maxwell’s equations are built on approximations and simplifications that fit observation. They aren’t how things actually “are,” but they are good at predicting what we see.

Classical physics is demonstrably wrong when applied at a small enough scale, but it is a very useful approximation when dealing with objects on a macro-scale.

All of quantum mechanics is built on similarly shaky simplifications and approximations, but it has been useful in describing and then predicting observable effects that have ultimately contributed to development of further theories and many useful technologies. The utility in these approximations comes down to their ability to predict things that otherwise would have been unexpected, inexplicable, or counter-intuitive, but are observable none-the-less.

The challenge with climate models at the heart of the the climate death cult have a terrible record of predicting observations. They do a great job of fitting historical data (they wouldn’t be published otherwise), but I have yet to see one that makes observable predictions that fit unfiltered/unselected/unedited data. Climate models when I was young predicted a coming ice age. Climate models since have predicted such lovely things as glaciers melting, sea levels rising, and many other such observable phenomena without success.

For every prediction that has come true, there are similar numbers that did not. Those odds are about as good as playing slots in Vegas, and I don’t base my lifestyle and choices on those odds. It’s about like predicting the weather in Hawaii based on how gassy my infant granddaughter is here in Utah.

I can’t tell you how many times I saw predictions for increased hurricane activity while I lived in Florida, only for the season to be average or below. When I lived in Ohio, the models predicted more severe tornadoes that didn’t arrive. When I lived in Alaska, it predicted extinction of polar bears, only for the polar bear population to increase a few years afterwards. And these were mostly macro-scale models that make less use of gross approximation than what the IPCC bases it’s recommendations on.

In order for a highly simplified model to be taken seriously, the predictive outcomes need to be observable and statistically robust. Climate models simply are not. Unfortunately, we don’t spend a lot of energy talking about the predictions that failed. We systematically pre-filter data before publication to bias towards the positive finding (true in all disciplines, not just climate science). We do lots of things to convince ourselves, but it’s just not that robust. My sense is that the outcomes that keep people up at night are predictions based on models that have been less accurate than a blind and deaf fortune teller predicting the end of the world based on lumps on the head of a child born under a full moon.