Lightning in a Bottle/Prelude
"The sciences, each straining in its own direction, have hitherto harmed us little; but some day the piecing together of dissociated knowledge will open up such terrifying vistas of reality, and of our frightful position therein, that we shall either go mad from the revelation or flee from the light into the peace and safety of a new dark age."
0.0 Motivation and Preliminaries
The world is messy, and science is hard. These two facts are, of course, related: science seeks to understand a messy world, and that’s a difficult task. Scientists have a variety of tools at their disposal to cope with this messiness: the creation of idealized models, the scientific division of labor, and the proliferation of increasingly elaborate pieces of technology all serve to help us predict and control a complex world. Not all tasks call for the application of the same tools, though, and so the scientific project takes all kinds: there’s room for a variety of contributions, and we must be willing to change tactics as new problems present themselves. Adaptation, flexibility, and collaboration are at the heart of scientific progress. This dissertation is intended not to be a work in the philosophy of science precisely, but neither is it, strictly speaking, a work of “pure science” (whatever that might mean). Rather, it is a philosophical contribution to science itself: I will attempt to employ the methods and tools of the philosopher to engage with a concrete issue in contemporary science—the problem of global climate change.
In March of 2010, Dr. Jon Butterworth of University College, London’s high energy physics group published a short piece in The Guardian titled “Come on, ‘philosophers of science,’ you must do better than this,” in which he called upon philosophers of science to make a real contribution to the emerging (and increasingly important) climate science debate. Butterworth’s call for philosophers of science to “do better” was inspired by another contribution to The Guardian from a few days earlier, this one written by Nicholas Maxwell, a philosopher at University College, London. Maxwell’s piece, “Scientists should stop deceiving us,” criticizes scientists generally (and climate scientists in particular) for producing what he calls “incomprehensible gobbledygook” that (he suggests) is to blame for the public’s rejection of scientific insights. Going even further, Maxwell suggests that underlying this problem is an even deeper one—an insistence on the part of scientists (especially physicists) that scientific theories be “unified”—capable of applying to all parts of the world in their domain—and that more explanatorily satisfying theories are rejected on the basis of disunity, leading to a thicket of incomprehensible theories that make little contact with the values of contemporary society.
As Butterworth points out, there is surely some truth to Maxwell’s criticism:
Science often falls short of its ideals, and the climate debate has exposed some shortcomings. Science is done by people, who need grants, who have professional rivalries, limited time, and passionately held beliefs. All these things can prevent us from finding out what works. This is why the empiricism and pragmatism of science are vital, and why when scientific results affect us all, and speak against powerful political and financial interests, the openness and rigour of the process become ever more important.
Science (to recapitulate the point from above) is hard, and indeed does often fall short of its goal of predicting what will happen in the world. The reasons for these failures are varied and complicated, but Maxwell is surely right to say that some of them have to do with the attitudes of some scientists themselves. With Butterworth, though, I have a hard time seeing the force of the claim that much of the blame for this problem is to be laid at the feet of specialization: the division of scientific labor is a natural, reasonable, and deeply effective response to a messily complex world. The “gobbledygook” that Maxwell decries is (as Butterworth notes) a kind of sophisticated short-hand meant for communication between experts themselves, not between experts and the public; the problem there, then, is less with the science itself and more with the communication of science. The problem, to put the point another way, is that it is difficult for working scientists themselves to take a high-level view of the project as a whole, and to see the scientific forest for the experimental trees. This, perhaps, is where a philosopher might help.
Butterworth closes his article with a few distinctly philosophical-sounding assertions.
Science is a form of systematised pragmatism: it finds out what works, and in the process we increase our understanding of the universe in which we live. I have no objection to philosophers watching, and trying to understand and improve the processes. It might even work. But they really ought to (and often do) have an understanding of what they are watching. … This is worth discussing, and I sincerely hope philosophers of science can do better than Maxwell in contributing to a debate of huge significance for the future of our species.
I agree whole-heartedly with this sentiment. Philosophers of science do indeed need to do better with regard to climate science—it is a real, pressing issue: perhaps the most pressing contemporary scientific issue facing us. To a very great extent, this means doing something: the degree to which philosophers have engaged with climate science at all is minimal even compared to the general paucity of philosophical contact with applied contemporary social issues. While some people in philosophy departments have begun to take notice of this (more on this later), it is high time that more followed suit, and that this became a topic of wide-spread discussion among philosophers. It is in this spirit that this project is conceived; my hope here is not to solve the climate change problem (that is not my job), nor is it simply to provide the kind of abstract theoretical criticism that Butterworth rightly calls down Maxwell (as a representative of philosophy of science generally) for being obsessed with. Rather, it is to sketch the lay of the land. My hope is that this dissertation will open the door to contributions by my peers (many of whom are, I am sure, far better equipped to deal with these issues than I am) to begin to have a conversation about this pressing social and scientific problem. My hope is that this will be the beginning of philosophers of science at least trying to do better.
0.1 Outline and General Structure
The somewhat unusual nature of this project, though, means that the structure and methodology of this dissertation will be somewhat different from most works both in philosophy and science. Before beginning the project proper, then, I want to say a bit about why I chose to structure things as I have, and why I have focused on the issues that I chose. My hope is that in flagging some of the unorthodox aspects of this work as intentional, I might short-circuit a few lines of objection to my project that would (I think) serve only to distract from the real work to be done. To get the ball rolling, let me lay out a sketch of how this work will proceed.
First, there are foundational questions. These questions concern the structure of science generally, the relationship between the various branches of science, climate science's continuity (or lack thereof) with the rest of science, and other issues that don’t seem to be investigated directly by any other branch of science. Foundational questions include those that are traditionally thought of as the purview of the philosopher: "how do we know that we can trust science?" is a paradigmatic foundational question (and a surprisingly difficult one to answer, at that). Chapters One, Two, and Three of this work will focus on foundational questions. Specifically, Chapter One outlines a novel approach to philosophy of science based on recent advances in information theory, and lays the groundwork for applying that approach to the problem of climate science. Chapters Two and Three review some contemporary work being done in complexity theory, with a particular focus on attempts to define and quantify the notion of “complexity” itself, then sketch an account of complexity that builds on the work done in Chapter One.
Second, there are methodological questions. These questions are more specifically concerned with the structure and operation of a particular branch of science; the methodological questions that will concern (say) a fundamental physicist will be different from the methodological questions that will concern a climate scientist. Questions about how climate science makes its predictions, on what basis we ought to trust those predictions, how we might use the tools of climate science to make better predictions, how to interpret the climate data on record, and how to best make use of our limited computing resources are all methodological questions. "How should we decide which factors to include in our climate model?" is a paradigmatic methodological question. Chapters Four, Five, and Six will focus on methodological questions. Chapter Four consists in a general introduction to the project of climate modeling, with a focus on the limitations of simple climate models that are solvable in the absence of pure computer simulations. In Chapter Five, I examine the challenges of building more complex climate models, with special attention to the problems posed by non-linearity and chaos in the climate system. In Chapter Six, I examine the role that computational simulation plays in working with climate models, and attempt to reconcile the novel problems posed by “science by simulation” with the results of climate science.
The answers to questions in each of the categories will (of course) be informed by answers to questions in the other categories; how we ought to react to a rapidly changing climate (an evaluative question) will clearly depend in part on how much we trust the predictions we've generated about the future (a foundational question), and that trust will depend in part on how we design and implement our climate models (a methodological question). My purpose in delineating these categories, then, is not to suggest that this division corresponds to essentially different spheres of inquiry—rather, this way of carving up the complicated and multi-faceted problems in the philosophy of climate science is just a pragmatic maneuver. Indeed, it is one of the principal theses of my project that none of these groups of questions can be effectively dealt with in isolation: they need to be tackled as a package, and a careful examination of that package is precisely what I am concerned with here. With this general structural outline in mind, then, let me say a bit more about what I intend to do in each chapter.
Chapter One is the most traditionally philosophical, and deals with general questions in the philosophy of science. In particular, I focus on the question of how philosophy can make a contribution to the scientific project. I offer an apocryphal quotation attributed to Richard Feynman, viz., " Philosophy of scientists is about as useful to scientists as ornithology is to birds," as my primary target, and attempt to see how a philosopher of science might respond to Feynman's charge. I argue that none of the accounts of science on offer in the literature can easily meet this challenge, in large part because they're often concerned with questions that are of little real consequence to practicing scientists. Drawing on concepts in information theory, I construct a novel account of structure of the scientific project that (I hope) skirts some of the stickier (but, I argue, less important) issues in which 20th century philosophy of science often became mired. With that account of science in hand, I argue that philosophy has a real contribution to make to the scientific project as a whole—I argue, that is, that there are issues with which the scientific project ought to be concerned that are not precisely scientific issues, and that philosophers are in a good position to tackle those issues.
In offering this account of the structure of science, I also give a novel way of understanding what it means to say that the scientific project is "unified." This is not merely an abstract point, but has real consequence for what will and will not count as a legitimate scientific theory: as we saw, one of the criticisms Maxwell offers is that scientists reject what he considers perfectly good theories on the basis of disunity. Is this true? In what sense is the unity of science an important guide to scientific theory, and how should we evaluate the relative unity of different theories? Does the unity of science conflict with the obvious methodological division of labor across the different branches of science? In addressing these questions, I hope to set the stage for a more fruitful examination of climate science's place in the scientific project overall.
Chapters Two and Three taken together are primarily a contribution to the foundations of complex-systems theory. Building on the account of science from Chapter One, I argue that the traditional bifurcation of science into physical and social sciences is, at least sometimes, misleading. I suggest that we should also see some scientific problems in terms of a distinction that cuts across the physical/social science division: the distinction between complex-systems sciences and simple-systems sciences. After reviewing some of the attempts to define "complexity" in the (relatively nascent) field of complex-systems theory (and arguing that none of the attempts fully succeeds in capturing the relevant notion), I use the machinery assembled in Chapter One to construct a novel account of complexity that, I argue, unifies a few of the most plausible definitions in the literature. This concept, which I will call dynamical complexity gives us a theoretical tool to help us think about the difference between systems that seem intuitively "simple" (e.g. a free photon in a vacuum) and systems that seem intuitively "complex" (e.g. the global climate) more clearly, and to begin to get a grasp on important differences between the methods of sciences that study systems with high dynamical complexity and those of sciences that study systems with low dynamical complexity. I then argue that, based on this definition, climate science is a paradigmatic complex-systems science, and that recognition of this fact is essential if we're to bring all our resources to bear on solving the problems posed by climate change.
In Chapter Four, we turn from explicitly foundational issues in the philosophy of science and complexity theory to more concrete methodological questions. I introduce the basics of climate science, and construct a very simple climate model from first principles. This chapter closes with a consideration of the limitations of the methods behind this basic model, and of the general principles that inform it. This paves the way for the discussion of deeper challenges in Chapter Five.
Chapter Five describes some of the specific problems faced by scientists seeking to create detailed models of complex systems. After a general introduction to the language of dynamical systems theory, I focus on two challenges in particular: non-linearity and chaotic dynamics. I discuss how these challenges arise in the context of climatology.
We'll then focus on a more concrete examination of a particular methodological innovation that is characteristic of complex-systems sciences: computer-aided model-building. Because of the nature of complexity (as described in Chapter Three) and the various special difficulties enumerated in Chapter Five, many of the techniques that simple-systems sciences rely on to make progress are unavailable to climate scientists. Like economists and evolutionary biologists, climatologists' most potent weapon is the creation of complex mathematical models that underlie a host of computer simulations. In Chapter Six, I examine some of the widespread criticisms of this "science by simulation," and argue that they are either misinformed or not fatal to the project of climate science. Drawing further on the resources of complex-systems theory, I argue that the function of computational models is not exactly to predict, but rather to act as “tools for deciding,” helping us coordinate and organize our more detailed investigation of the global climate.
0.2 Methods and Problems
The relative paucity of philosophical literature dealing with issues in the foundations of climate science puts me in the somewhat unusual position of having to cover an enormous amount of territory in order to mark out the lay of the land. In order to do what I want to do, then, I need to sacrifice a certain amount of depth in the name of achieving a certain amount of breadth. This is a deliberate move, but it does not come without consequences. Before beginning the actual project, I want to take a few pages to review some of these issues, flag them as problems that I have considered, and offer a few justifications for why I have chosen the approach that I have.
There is some risk that in trying to speak to everyone with this dissertation, I will end up satisfying no one at all. I suspect that individual readers will find my discussions of their particular areas of specialization somewhat unsatisfying: philosophers of science operating in the tradition of the profession—those who have inherited their methods and problems down from Hempel, Kuhn, Popper, van Fraassen, and so on—will likely find my discussion of the structure of the scientific project in Chapter One unsatisfying in virtue of the fact that it makes very little contact with the classic literature in the field. Mathematicians and physicists working in dynamical systems theory will likely find my discussion of dynamical complexity unsatisfying in virtue of its relatively informal and non-mathematical presentation. Practicing climatologists will likely find my discussion of Mann's work in particular (and the methods of climate science in general) unsatisfying in virtue of the fact that I am not myself a climatologist, and thus lack the kind of sensitivity and feel for the scientific vernacular that comes from years of graduate school spent simmering in the relevant scientific literature. Ethicists and political philosophers will likely find my discussion of the moral and social issues surrounding climate science's predictions unsatisfying in virtue of the fact that I (quite admittedly) know very little about the state of the ethics literature today, and thus will be presenting largely what I see as common-sense approaches to solving these problems that are as devoid of ethical theory as possible.
In short, no matter who you are, you're probably going to be deeply suspicious of what I have to say, particularly about the topic in which you specialize. Why, then, have I chosen to approach this project in the way that I have? Instead of leaving everyone upset, why not try to please a small number of people and make a deep contribution to just one of the issues I discuss here? There are a few answers to this that are, I think, related. Perhaps primarily, I'm concerned with philosophy's treatment of climate science generally, and a highly general approach is (I think) the best way to express this concern. As I've said, while there has been a not-insignificant amount of value theory done on the topic of environmental ethics, there's been very little philosophical contribution to the actual science of climate change. In effect, then, one of the principal goals of this dissertation is to jump up and down, wave my arms, and shout "over here!" As I inevitably get some (many) of the details wrong in my discussion, I hope others will be inspired to step in and correct things, point out what I've done incorrectly, and do better than I am capable of doing. If I can inspire enough controversy to get the philosophical community involved in the climate change debate, then I will count this as a success, irrespective of whether or not my own views are accepted or rejected.
Relatedly, part of my intention here is to stake out a large amount of territory all at once to suggest how those with expertise in specific problems might make deeper contributions than I make here. In discussing philosophy of science, complexity theory, model-building, and value theory all in a single work, I hope to sketch the general shape that a fully-fledged "philosophy of climate science" literature might take, and to open the door for more systematic contributions to that literature by those who are best equipped to make them. In order to make this goal achievable in only a few hundred pages of writing, I'm forced to make a number of simplifying assumptions in some places, and to ignore significant problems entirely in other places. Whenever possible, I will offer a footnote flagging the fact that I'm doing this deliberately, and suggesting what a more careful elaboration of the topic might look like. If I were to give each topic here the full attention it deserves, this work would be thousands of pages in length (not to mention beyond my ability). I far prefer to leave the project of elaborating and expanding most of what I'm trying to start here to my betters. To facilitate this, I will close each chapter with a series of questions for further exploration, or a brief discussion of the shape that future research might take. I intend to take up at least some of this research myself in the future (particularly work in the foundations of complexity theory and information theory as they relate to climate science and the scientific project as a whole), but I am equipped with neither the time nor the ability to take all of it up; climate change is a pressing issue that demands our immediate attention, and we'll need to work together if we're to solve this problem. If nothing else, this dissertation is a sustained argument for precisely this point.
Finally, it is worth highlighting that this dissertation is motivated by an explicitly pragmatic approach to philosophy and science. I think that Butterworth is precisely correct when he says that "science is a form of systematized pragmatism," and I suspect that most scientists (insofar as they think about these things at all) would, given the chance, assent to that statement. The largest consequence of this is that I wish, whenever possible, to remain totally neutral as to how what I'm saying makes contact with more traditionally philosophical questions—particularly those in mainstream metaphysics. Chapter One places a great deal of weight on facts about patternhood, and there is a temptation to attempt to read what I'm saying as making a claim about the metaphysical status of patterns—a claim relating to the emerging metaphysical position that some have termed "ontic structural realism." I will say a bit more about this in Chapter One when the issue comes up directly, but this is worth mentioning here by way of one last methodological preliminary: while I do indeed have a position on these issues, I think the point I am making here is independent of that position. I'm inclined to agree with something like the structural realist position the James Ladyman and Don Ross have pioneered—that is, I'm inclined to agree that, if we're to take science seriously as a metaphysical guide, we ought to take something like patterns (in a robust, information-theoretic sense) as the primary objects in our ontology—but this is a highly controversial claim in need of defense on its own terms. This is neither the time nor the place for me to enter into that debate. When I couch my discussion in terminology drawn from the structural realist literature—when I speak, for instance, of "real patterns,"—it is merely for the sake of convenience. Nothing in my project turns on taking this language as anything but a convenience, though—if you prefer to take the Humean view, and think of patterns as the sort of things that supervene on purely local facts about spatio-temporal particulars, that will do no violence to the story I want to tell in this dissertation.
Conversely, if you wish to read parts of this (particularly the first three chapters) as the preliminaries of a contribution to the metaphysics of patterns, or as a sketch of how such a metaphysics might be tied to issues in the foundations of complex systems theory, this also will not impact the larger point I want to make. Indeed, I will suggest at the close of Chapter Three that such an exploration might be one of the future research programs suggested by this project. I take it as one of the strengths of this approach that it is neutral between these two interpretations—whether or not you are sympathetic to the Dennett/Ladyman account of patterns as primary metaphysical objects or not, my discussion of patternhood turns exclusively on patterns understood in the (relatively) uncontroversial information-theoretic sense. That's the sense in which I want to maintain metaphysical neutrality here—some of my discussion adopts conventions from the structural realist camp, but this is strictly a matter of convenience and clarity (they have developed this vocabulary more than any other area of philosophy). I'm confident that the points I make could be translated into more obviously neutral terms without any significant problems.
With these preliminaries out of the way, then, let's begin.
- Butterworth (2010)
- Maxwell (2010)
- Butterworth (ibid.)
- See, canonically, Dennett (1991) and Ladyman et. al., (2007)
- I do intend to develop the kind of framework I deploy in Chapter One into a robust metaphysical theory at some point. That is simply not the project with which I am concerned here.