I’ve been thinking a lot lately about the concept of détente. I am fairly certain it is going to play a big role in my next book, but I haven’t figured out the precise details.
A détente is a general easing of tensions within an adversarial relationship before underlying conflicts have been resolved (otherwise you would call it “peace”). I think of détente as a “happily almost ever after” narrative pattern. Unlike a truce though, a détente is a sort of indefinite cessation or slowing down of conflict without specific expectations of alternative approaches towards resolution, or specified time limits. You know a decisive drive towards an outcome will be resumed, but you don’t know when, why, how or where for sure. You just collectively agree that now is not the time or place.
I’ll sketch out in general terms why the concept is interesting, but I am going to wander quite a bit along the way and use this post as an excuse to philosophize about game theory and academic culture, and share an interesting anecdote. You’ve been warned.
Utopia for Realists
In political science, where the concept of détente originated, the adversarial relationship is generally one involving a symmetric balance of power, such as the US-USSR relationship during the height of the Cold War. The easing of tensions in that case had a somewhat literal and quantitative character, since it involved attempting to slow down nuclear and missile proliferation. A détente is usually consciously engineered by the concerned parties.
If you interpret the term more broadly as any sort of temporary peace with an element of conscious intent, you start to see why the idea is so useful. It is the realist’s substitute for utopia.
The narrative pattern is ubiquitous in fiction. The end of The Hobbit is a détente because of the unresolved matter of the One Ring and the Necromancer. It was conscious because Gandalf could have chosen to drive that matter towards resolution immediately after the Smaug adventure, but chose to let that sleeping dog lie for a while. The Necromancer too, prefered to delay an active hunt for the One Ring while recovering his strength (this narrative pattern is faithfully replicated in the Harry Potter series). Each of the Terminator movies ends in a détente rather than a happily-ever-after condition.
Most real-life struggles never really end. All we enjoy are periods of détente with the forces we all constantly struggle against, individually and collectively.
So to engineer a détente is to choose peace and relaxation for a while. It is happily ever after for realists.
The idea of a détente is interesting because it represents a change in the character of an equilibrium situation through some sort of engineered, negotiated process. I believe we’re going to be doing a lot of such engineering and negotiation in this century, but around technology, not politics or war.
A Digression into Game Theory
Before I get to détente, I need a little digression into game theory, the most common framework used in the social sciences for modeling partially adversarial situations (including détente). The game-theoretic approach is so dominant for such questions that if you choose to ignore it, or even reduce it to secondary status (as I plan to), you sort of have to defend your decision to pre-empt certain futile debates.
I need to talk specifically about the well-known Iterated Prisoner’s Dilemma (IPD) model, sometimes called the e. coli of social science research.
If you aren’t familiar with the prisoner’s dilemma (PD) paradox and its fertile, iterated cousin, you should read the Wikipedia entry linked above before proceeding.
I have a love-hate relationship with game theory in general and IPD in particular. I even worked briefly (and due to my inexperience at the time, not very successfully) with Robert Axelrod, the pioneer in the field, before finally cooling on the field permanently. I suppose it helped that my PhD adviser was equally suspicious and wary of the subject. He viewed it as a rabbit hole from which, if you were not careful, you could not escape. I initially disagreed with him, but after a few conversations and exploratory forays into the field, I came to share his views.
In my opinion, game theory in general is a frustrating but fundamental cognitive framework for anyone interested in complex social systems. If that describes you, you should have more than a passing familiarity with it. Even the non mathematically-inclined should make an effort to understand the basics. The basic math is not very hard at all. The subject is challenging because it demands agile shifting of perspectives among various modeled points of view, not because the math is hard. In a way, game theory is like calisthenics for perspective refactoring.
Axelrod’s two books The Evolution of Cooperation and The Complexity of Cooperation give you a sense of how powerful game theory can get. If you are inclined to wander beyond the Prisoner’s Dilemma to models of things like management-labor negotiations or market competition, Thinking Strategically by Dixit and Nalebuff is the standard text for simplified, applied game theory. For those of you who want to get into the more mathematical end of things, there are plenty of standard texts in a variety of flavors ranging from the Von Neumann and Morgenstern Ur text to more modern treatments to suit the subtly differing tastes of mathematicians, sociologists, economists and engineers. Explore the selection and pick a suitable starting point if you need to educate yourself on the subject.
But once you know the basics, my advice is to severely limit your use of game theory as an intellectual tool except for very special situations (but since the subject is intellectual cocaine, there is a strong chance you will ignore my advice and become a serious addict anyway. But I can’t in good faith tell you to “just say no,” since being an addict is better than being ignorant of the field).
But basically, game theory should never be your main tool. Only a secondary tool.
Game Theory as a WMID
Because game theory is the finest discussion-killer known to humanity. In the wrong hands, it is a weapon of mass intellectual destruction. It takes extraordinary sensitivity and taste to use it in a way that opens rather than closes discussions. A level of sensitivity and taste that very few people reach. In fact, I consider myself as lacking the necessary level of taste to safely use game theory, and I am not being self-deprecating.
This is why Axelrod’s work is deservedly famous. He used game theory as a fertile and exploratory story-telling tool rather than as a destructive weapon. At the risk of annoying a lot of people, I think most classically trained economists use game theory in the WMID way.
Let me give you an example of the dangers from my younger, greener days.
Many years ago, I’d built some computer models I was quite proud of. The models pretty neatly captured many features of combat in a modern battlefield with lots of moving parts, such as drones, humans, networked C2 systems, AI systems and so forth. I felt that the models were at exactly the right level of abstraction to get at some interesting questions without getting bogged down in the details of hardware. Fun things and pretty graphics happened with my models when you hit “Go.”
I got an opportunity to talk about my work at Wright-Patterson AFB. After spending an hour explaining what I’d done, with nobody falling asleep or leaving, we were in the Q&A portion and having an interesting discussion of various subtleties. Then one guy in the audience (a visiting professor) offered up a comment with a rather supercilious air: “this is all very complex and interesting, but unnecessary. Let me offer you an even more abstract model than yours: at the highest level you are just doing some simple game theory.” (emphasis his).
In case you haven’t been exposed to much academic-speak, ”more abstract” and “higher level” are generally code for “I am smarter than you” status assertions.
I protested: “Yeah, sure, you could probably model this at a more abstract level using game theory, but that was not my point. I was trying to explore the phenomenology of decision-making in the presence of…”
But it was too late. The damage had been done. The discussion had been derailed and killed.
In hindsight, this was one of those formative events that led me to abandon academic research (which is full of such posturing and Q&A skirmishes that go nowhere), but that’s another story. I learned from that incident to spot and pre-empt use of many WMIDs that expert academics use to kill interesting debates that they are threatened by (game theory is one of the biggest status-asserting WMIDs, but there are others, such as global optimization, as in “that’s just a global optimization problem.”)
But to get back on track.
Game theory has this effect because it is more than a field of applied mathematics. To many, it is an incredibly satisfying appreciative world view for Life, The Universe and Everything. The entire universe starts to make a seductive and nihilistic sort of sense once you adopt game theory as a religion rather than a tool. The analogy to religion is not shallow. There are people who say “non-zero sum” the way religious people say “and then there was light.” You see evangelical fervor in their eyes.
But the apparent appreciative power of game theory, for me is ultimately unsatisfying. Yes, such and such complex situation is a Nash equilibrium. Or such and such evolutionary process looks like an Evolutionary Stable Strategy (ESS) in IPD.
Real-world systems are generally too complex to allow direct application of game theory as a practical tool. But even its appreciative use is toxic, because unlike most appreciative views that tend to open up fertile avenues for further exploration, game theory mostly makes you dismiss further exploration as unimportant. It makes you stop thinking rather than start.
A game theorist would most likely approach the idea of a détente as a shift from a regime of costly mutual defection (D-D) to one of less expensive mutual cooperation (C-C). Axelrod showed how this could happen even without centralized coordination.
There are many examples of varying complexity. Two countries at war declaring a ceasefire is one. Companies competing fiercely agreeing to form a cartel is another. A circle of people, each pointing a gun at his neighbor to the right, mutually agreeing to lay down the guns (or exchanging the guns for knives or fists) would be a third.
There are also plenty of examples in evolutionary biology, involving species such as bees and ants. You can read about them in Dawkins old classic, The Selfish Gene, which is a work that is very game-theoretic in both spirit and specifics, as is the rest of Dawkins work.
This general applicability is actually revealing. If the same model applies to ants, trench warfare in World War I, and brands competing in the marketplace, it is highly unlikely that you’ll get at the interesting differences among the situations using that model. The fact that ant or bee brains can “run” behavioral algorithms that result in game theoretic dynamics tells you how simple they are.
So a preference for game theoretic models is actually a preference for simplicity and generality, which I lack. I like complexity and narrative specificity.
To a certain extent, preferring simplicity and generality is a matter of taste and interest. But it is also a pattern of denial that rationalizes a distaste for actually diving into the details of things that have nominally been “explained.”
I view this stance as the essence of classicism: the intellectual aesthetic that draws the most narrow boundaries around what ought to be considered “interesting” and seeks to elevate things within the boundary as the only legitimate subjects of inquiry.
Dawkins is a very interesting scholar primarily because he is a sort of revolutionary classicist, an odd combination. I read The Selfish Gene (a sort of revolutionary perspective-refactoring of Darwinism into neo-Darwinism) and his most original work, The Extended Phenotype, but then felt no desire to read the rest of his books or follow him into militant atheism. Mainly because I felt he was repeating himself endlessly, adding more color and detail to his original ideas from the late seventies.
Dawkins well-known feud with his evil twin, Stephen Jay Gould is also revealing. Gould is probably best known (at least to lay people like me and, I assume, most of you) for his work on punctuated equlibria. This is, roughly speaking, the idea that speciation in biological evolution happens in bursts, with periods of equilibrium in between (a sort of life-versus-environment détente model).
Dawkins had specific technical and factual objections to Gould’s models, but his primary objection was an aesthetic one. He simply didn’t think it was a very important idea, deserving of the publicity it has received. He is quoted as saying: about it:
“It is a “minor gloss,” an “interesting but minor wrinkle on the surface of neo-Darwinian theory,” and “lies firmly within the neo-Darwinian synthesis”.
This critique reveals something about Gould as well: classicists are most likely to disparage the views of romantics: those who draw the most relaxed boundaries around “interesting” (if they draw boundaries at all).
Romantics are not satisfied with just the premise and outcome of a story. They want to know the story itself; narrate the whole thing in all its path-dependent glory, and worry even about the local specifics that do not usefully contribute to the general theory. They are particularists rather than grand unifiers. They want to tell satisfying stories.
Of course, they have their own denials (often a weaker intellectual capacity for elegant generalization and unification, coupled with a sort of “stamp collector” addiction in the sense of the Rutherford quote), but romantics generally appreciate the value of classicist views more than classicists appreciate romantic views.
This is a high-level example, involving far more famous and important people, of the same sort of romantic-classicist skirmish that killed the discussion at my talk.
It is my belief that this (primarily aesthetic) stance adopted by Dawkins’ is due to his use of game theory as a primary tool. In fact I consider it something of a great tragedy that Dawkins himself never explored the ideas in The Extended Phenotype (which are not primarily game theoretic) much further. I suspect his game-theoretic roots made him abandon the most fertile directions suggested by EP. The directions required a romantic, rather than classicist, imagination to explore. And since Dawkins has mainly attracted other classicists to himself, these territories (the idea of “memetics” among them) remain both inadequately and inappropriately explored (by unifiers seeking to generalize rather than particularists seeking to narrate), and sort of permanently fenced off from all but the bravest romantics.
Now I don’t know enough about evolutionary biology to discuss the relative merits of the contributions of the late Gould and still-living Dawkins, but I have to say, aesthetically I am firmly on Gould’s side. Punctuated equilibria, whatever their origin, are fascinating as a narrative pattern. To me, they are the main plot in the story of evolution. I want the blow-by-blow account of the Cambrian explosion. To dismiss the idea and focus exclusively on just the classicist framework is like saying about Shakespeare “all the plays are either comedies or tragedies, enough said.”
One of these days I’ll write more about romantics versus classicists (I recognized myself as a romantic after my PhD adviser sensitized me to the difference: he would remark about work that he approved of: “that is romantic science.”)
But back to our subject.
Modeling Détente: Take Two
The idea that détente is a shift from mutual defection to mutual cooperation, end of story, is where I was until a few months ago (shows you how much my own thinking has been compromised by game theory).
At the risk of going too meta, let me note that game theory itself is a mechanism by which détente can emerge within an intellectual culture, since it can shut off fertile areas of exploration and declare a temporary intellectual peace (I call it a WMiD because in thinking, war is preferable to peace, so anything that kills a war is prima facie a bad thing unless proven otherwise). In this case the idea of a détente hasn’t been seriously studied since Schelling and other Cold War era thinkers (at least to my knowledge, correct me if you know of interesting new developments), despite the obvious broader utility of the concept even (and perhaps especially) in our chaotic times.
So let us ask again, what is a détente?
And this time, let’s take note of the elements of a potential romantic answer, rather than a classicist answer. I can’t provide a complete account obviously, since I am just beginning to explore the idea, but here are the points I’ve noted.
- In the simplest case, it is the engineered emergence some sort of temporary peace emerging from some sort of war, but not through decisive win-lose or win-win (see how seductive game-theoretic language is?) fulfillment/denial of objectives. Rather, it is a shift that emerges because everybody recognizes that decisive outcomes are not possible at that time and that the fighting might as well be abandoned for the time being. It is sort of a consensus that a state of diminishing returns has been reached.
- In a slightly more complex case, it is the deliberate closing of one battlefront in order to concentrate action on another that is preferred by all parties (or a single dominant party), generally because a more decisive outcome seems likely on that front. In this case, engineering a détente on one front is a movement of proceedings to a different local optimum.
- Unlike evolved cooperation (via natural selection), it is consciously engineered by intelligent agents.
- In my ongoing examination of various examples, I’ve concluded that pure one-front détente engineering of the first sort is very rare. The second kind, the “move the action to a front where decisive outcomes are possible” kind, is far more common.
- Sometimes, the front moves to another qualitatively similar front. So a détente engineered on the oceans might move a war to land. In other cases, the action moves to a different kind of front, hence “war is merely a continuation of politics by other means.”
- It is possible to view politics, war, technology and culture as four cyclically connected détente zones. You can generalize the Clausewitz quote in the previous point to the obvious set of cyclic propositions (for example, “technology is a continuation of war by other means”). It is quite rare (perhaps impossible) for there to be quiescence on all four fronts at once.
- Why would anyone want to move action elsewhere? Usually because a decisive outcome in a conflict is a kind of information discovery event about relative power and status. When you win or lose in a conflict, there is new information in play. Conflict without hope for a decisive outcome is in nobody’s interest.
- One of my tentative conjectures is that conflict always moves to the locus where the most asymmetric new information will be generated. Except in cases where all parties are in collective denial because the information that might be generated is of a negative sort, there is usually somebody who benefits disproportionately from moving the action somewhere else, and somebody who loses. So engineering a détente is itself a meta-game, with all parties trying to move the action to fronts that favor them.
It is very tempting to model this meta-game using game theory as well. These models-within-models tend to get very confusing though. I am probably going to take a stab at it as an exploration on the side.
My main interest though, is in narrative patterns that lead into, and out of, détente situations. Who are the actors? How do they frame the situation? What moves are used in engineering the peace? How are new fronts selected? When do actors choose to disrupt a prevailing peace? What defines the realist utopia?
And in particular, I am interested in asking these questions about the story of technological evolution. I’d like a satisfying punctuated equilibrium model of technological change. I’d like to predict and characterize the next technological détente if there is going to be one, or satisfy myself that we’ll end up in long-term future nausea.
Sleeping Dogs and Useless Wars
The idea of a détente is useful in two ways.
First, for events and situations that are playing out on a larger canvas than you personally control, the idea helps you predict which conflicts might go quiescent soon, and which ones might flare up again. Since there is no permanent peace, all you can do is track the action as it moves. If you think in terms of détente, you will recognize that conflicts don’t end and start because decisive outcomes have been attained. They end and start because it is possible, and in everybody’s interests to either declare a temporary new peace or start a new war. This mental model helps you make timing predictions better.
The heuristic here is to follow the information. Conflict is useful only insofar as it drives decisively towards some sort of closure. If a conflict is raging around what appears to be a de facto balance of power, with no hope of one side either prevailing or outlasting the other, you can expect the conflict to die down.
Conversely, if there is a period of apparent peace in some domain, conflict might flare up again if one party of the arranged balance of power has been quietly gaining or losing power, making open conflict worthwhile again (so the US-China relationship is one such, since the US is quietly losing power, while China is quietly gaining).
This is the sleeping dogs and useless wars heuristic.
Second, within domains that you can significantly influence, the heuristic can help you do more than predict the course of events. It can help you pick battles wisely and manage war and peace within your world. You don’t start fights until you know you have a realistic chance of winning. You sue for peace when you know you sense a looming violent stalemate. That’s a subject for another day.