Brian Eno's Theory of Democracy

Organizing and generating variety in politics

[image by Brian Eno, from “77 Million Paintings.” Eno says in this dialogue that he doesn’t mind people using these images for non-commercial purposes]

This post’s title is a little cheeky. Brian Eno does not have an explicit theory of democracy that I know of, although he is visibly and emphatically committed to its practice. But Eno’s arguments about the arts tell us important things about how democracy ought work, and what kinds of democratic stability and variety we ought aspire to.

The back story to this post doesn’t start with Brian Eno. Back in 1991, the political scientist Adam Przeworski published a book, Democracy and the Market. Most of the book was about the democratic transition in Central and Eastern Europe, and it was very good as best as I can judge these things. But the first chapter was much, much better than very good. It laid out a brief theory of democracy that reshaped the ways in which political scientists think about it.

Przeworski’s theory starts from a simple seeming claim: that “democracy is a system in which parties lose elections.” It then uses a combination of game theory and informal argument to lay out the implications. If we assume (as Przeworski assumes) that parties and political decision makers are self-centered, why would the ruling party ever accept that they had lost and relinquish control of government? Przeworski argues that it must somehow be in their self-interest to so. He argues that they will admit defeat if they see that the alternative is worse, and (this is crucial) because democracy generates sufficient uncertainty about the future that they believe they might win in some future election. They know that they will hurt their interests if they refuse to give in, and they have some (unquantifiable but real) prospect of coming back into power again. Democracy, then, will be stable so long as the expectation of costs and the uncertainty of the future give the losers sufficient incentive to accept that they have lost.

This is a more beautiful idea than I am able to explain in a brief post, and certainly much more beautiful than any argument I will ever come up with myself. It compresses a vast and turbid system of enmeshed ambitions and behaviors into a deceptively simple nine word thesis. That thesis can then be unfolded into a variety of more detailed models of how democracy may sustain itself through the interlocking ambitions of politicians who strive to win office, but grudgingly accept that they must leave it if they lose. If Przeworski’s argument is not well known outside political science, it is likely because it is a chapter in a book that is mostly about different questions from a different time. But within political science, it is rightly regarded as a classic. If you have a Ph.D. in comparative politics, or even a good undergraduate degree, you will likely have been required to read it as a photocopy or a PDF.

Equally, Przeworski’s argument leads you to think about democracy in some ways, and not others. It suggests that the central elements of democracy are a set of self-reinforcing expectations about the alternation of power, which can be mapped with game theory. Many political scientists were inspired by Przeworski’s ideas to come up with more formal models of democracy, ditching Przeworski’s claims about uncertainty and informal arguments in favor of greater determinacy and cleaner math. Personally, I think that Przeworski’s initial, messier formulation better captures some of the dynamics of the system,.

This understanding implies that as soon as democracy has become established, it will continue to work into the indefinite future, because everyone’s expectations converge in ways that reinforce each other. Once a democracy has become sufficiently wealthy and established, parties and leaders should recognize that they have too much to lose from destabilizing the system. It would take a massive external shock (e.g. a major economic crisis), remaking the environment that democracy works in, to shift people out of their ordinary beliefs and behaviors.

These assumptions mostly stem from the way that game theorists think about the world. In their language, democracy is a self enforcing equilibrium, which will just keep on going so long as the parameter values don’t shift too radically. You don’t have to worry too much about destabilizing forces emerging from within democracy.

Unfortunately, it has become clear to many (including Przeworski himself) that this understanding can’t really explain why democracy is degenerating in the U.S. and elsewhere. Look around you: there are visibly circumstances under which democracy can be self-destabilizing rather than self-reinforcing, but standard game theory has little to say about this. As a result, Przeworski has begun to ask different questions about people’s expectations than he used to. People’s preferences and beliefs about democracy can change for endogenous reasons, as they are affected by the back-and-forth within the democratic system. Whether they are stabilizing or not will depend on circumstances that aren’t easy to capture with game theory.

That is why I think that Brian Eno’s ideas are politically as well as artistically valuable. To be clear, this is a notion that I’ve come to slowly. I’m a third generation product of the loose school that Przeworski and others founded in the 1980s.* But like Przeworski himself, I’ve begun to think over the last several years that the standard understanding misses a lot of what is happening. We need a different way of thinking about democracy. In a 2019 Boston Review article with Bruce Schneier, I waved towards this (other parts of the article are mostly Bruce; as best as I remember, this single sentence was mostly me):

What we really need is what complex systems theorists would call a “dynamical” model of democracy, which would capture how democratic systems can remain stable in the face of deep disagreements and the changing needs of a complex environment.

A little over a year ago, I came across Eno’s shortened essay,** “Organizing and Generating Variety in the Arts.” As the title makes clear, it is not an essay about politics, and it doesn’t mention the word ‘democracy’ once. It is still the best brief description of such a “dynamical model of democracy” that I’ve ever read.

Eno writes about music, drawing out the distinction between classical composition, and the forms of composition that interest him. As I interpret him,*** the differences have to do with how to interpret the score. In classical music, the score flattens out variety. Musicians perform the score as it is written - virtuosity consists of interpreting a score that is fixed in advance in compelling ways within a prescribed and narrow band of variations. The kinds of music that Eno wants to talk about treat the score in very different ways: as a set of simple instructions for generating variety rather than constraining it.

Cardew's score is very simple. It is written for any group of performers (it does not require trained singers). There is a piece of text (from Confucius) which is divided into 24 separate short phrases, each of one to three words in length. Beside each phrase is a number, which specifies the number of repetitions for that line, and then another number telling you how many times that line should be sung loudly. The singing is mostly soft.

All singers use exactly the same set of instructions. They are asked to sing each line of the text the given number of times, each time for the length of a breath, and on one note. The singers start together at a signal, and each singer chooses a note for the first line randomly, staying on it until the completion of the repetitions of the line.

The singer then moves on to the next line, choosing a new note. The choice of this note is the important thing. The score says: 'Choose a note that you can hear being sung by a colleague. If there is no note, or only the note you have just been singing, or only notes that you are unable to sing, choose your note for the next line freely. Do not sing the same note on two consecutive lines. Each singer progresses through the text at his own speed.'

A remarkable amount of order emerges from this seemingly unconstrained process.

A cursory examination of the score will probably create the impression that the piece would differ radically from one performance to another, because the score appears to supply very few precise (that is, quantifiable) constraints on the nature of each performer's behaviour, and because the performers themselves (being of variable ability) are not 'reliable' in the sense that a group of trained musicians might be. The fact that this does not happen is of considerable interest, because it suggests that somehow a set of controls that are not stipulated in the score arise in performance and that these 'automatic' controls are the real determinants of the nature of the piece.

There’s a cybernetic logic behind this argument.

Perhaps the most concise description of this kind of composition, which characterizes much experimental music, is offered in a statement made by the cybernetician Stafford Beer. He writes: ‘Instead of trying to specify it in full detail, you specify it only somewhat. You then ride on the dynamics of the system in the direction you want to go.' In the case of the Cardew piece, the 'dynamics of the system' is its interaction with the environmental, physiological and cultural climate surrounding its performance. The English composer Michael Parsons provides another view on this kind of composition: … This is an example of making use of 'hidden resources' in the sense of natural individual differences (rather than talents or abilities), which is completely neglected in classical concert music, though not in folk music.

Again, these are arguments about music. Still, as Eno acknowledges, they could just as easily be arguments about any form of social organization.

What I want to argue is that these ideas hold particular force for democracy, which more intimately resembles the collective creation and performance of music than most observers of either recognize.*** Both combine the production of variety with large scale coordination. Doing this well requires continuous adaptation both to the environment and to the endogenous and spontaneous variations that emerge from within the coordinating process. This is quite different from the usual static ways in which we think about organizations.

At one extreme, then, is this type of organization: a rigidly ranked, skill-oriented structure moving sequentially through an environment assumed to be passive (static) toward a resolution already outlined and specified. This type of organization regards the environment (and its variety) as a set of emergencies and seeks to neutralize or disregard this variety. … Proposing an organizational structure opposite to the one described above is valueless because we would probably not accord it the name organization: whatever the term does connote, it must include some idea of constraint and some idea of identity. So what I shall now describe is the type of organization that typifies certain organic systems and whose most important characteristics hinge on this fact: that changing environments require adaptive organisms. … An organism operating in this way must have something more than a centralized control structure. It must have a responsive network of subsystems capable of autonomous behaviour, and it must regard the irregularities of the environment as a set of opportunities around which it will shape and adjust its own identity.

I don’t want to press too hard on the resemblance between Eno’s analogies and my own ideas, but I think it’s fair to say that game theoretic accounts of democracy as a self-enforcing equilibrium are more like a rigid structure moving through an environment that it ignores except in crisis, than an adaptive organism with responsive subsystems. Again (and apologies for the technical language), game theorists treat the environment as a set of parameters in which (if the game is indefinitely iterated), an enormous number of equilibria will be possible, so that there is no good way of predicting whether one equilibrium or another will come about. The primary channel of information from the environment to the organization conveys just one single bit: is a particular equilibrium possible or not given the parameters? That feedback only changes when some external shock shifts the environmental parameters so far that the existing equilibrium can’t be sustained any more. And even then, all you can plausibly say with game theory is that some new equilibrium (likely one of many possible equilibria) will arise. Hence, the action involves self-reinforcing expectations which keep on going, and going, and going until suddenly they can’t.

Even game theoretic models that play with the math to provide a somewhat more complex account of the relationship between the environment and the democratic equilibrium tend to assume that expectations rapidly converge on stable outcomes. It’s no surprise that such models don’t do well at explaining change: they’re designed to simplify away most of the processes through which change might occur.

That’s why I think that Eno’s approach to thinking about the world is better suited to understand what has gone wrong, and how to set it right. That does not mean that we want to throw away Przeworski’s central insights about stability. But it suggests that we need to think about stability in different ways. Brian Eno’s account of organization is explicitly not a recipe for an unconstrained system - democracy needs, in some real sense, to remain democratic. But ‘stable’ and ‘static’ mean different things.

just as it is evident that an organism will (by its material nature) and must (for its survival) generate variety, it is also true that this variety must not be unlimited. That is to say, we require for successful evolution the transmission of identity as well as the transmission of mutation. Or conversely, in a transmission of evolutionary information, what is important is not only that you get it right but also that you get it slightly wrong, and that the deviations or mutations that are useful can be encouraged and reinforced.

In an electoral democracy, for example, it is absolutely true that parties that lose elections need to recognize that they have lost, and slink away to lick their wounds and prepare for the next electoral battle. But this will require a perpetual process of adjustment. Parties will - given their incentives - keep trying to push the boundaries of what is possible and allowed, looking for enduring advantages and means to rig the game in their favor. Responding to those behaviors will equally require dynamic responses, many of which will leverage the desires of their opponents to remain competitive.

In other words, we should treat Przeworski’s requirements as the score for a dynamic rather than a equilibrium based account of electoral democracy. Ensuring that parties leave office when they are defeated - and that they do not gimmick the rules to ensure their rapid return to power - is really important, but it will require continual adjustment as ruling parties discover new tricks. Doing that well will involve figuring out what the “set of controls that are not stipulated in the score” might be, how to make sure that they “arise in performance” and how to ensure that they are as effective as possible. The score sets out broad objectives, and some loose machinery for how to achieve them, but it is in the performance that most of the real action happens.

And not just that. On Eno’s arguments, organizations need to be interested in generating variety as well as stability. A successful social organization is not a machine for keeping things the same, or for that matter, for generating randomness, but a means of generating useful variety that can help to match the variety and changes in the environment that the organization lives in. This has obvious implications for democracy. It is through harvesting the variety of perspectives - the “natural individual differences” if you like - among its citizens that democracy can discover solutions to complex problems, and it is through coordination that it can address them. But doing both well involves figuring out difficult tradeoffs with stability. As Jenna Bednar argues:

Diversity—of interests, of issues—is a source of new ideas in the system. If a system of safeguards is too rigid, there is no exploration of ideas and practices that might provide a vision forward when a system needs to adapt. Additionally, if it lacks interest dispersion, there is no potential for cross-cutting cleavages to develop. The success of the architecture is measured by its capacity to manage the perturbations, avoiding maladaptation or system collapse.

Bednar advocates “distributed, simple safeguards rather than a single, complex, catchall structure” in a federal system that ‘modularizes’ politics. The idea here is to create a democratic system in which a variety of different ideas and solutions can be developed and tested out, and where if they fail, their failures will be limited. You want, on her account, loosely connected federal politics: sufficient for basic rights to be protected, and for safeguards to work, but not so strong as to impose a single vision on constituent parts that may experiment in diverse directions. Or, to repeat Eno’s description:

An organism operating in this way must have something more than a centralized control structure. It must have a responsive network of subsystems capable of autonomous behaviour, and it must regard the irregularities of the environment as a set of opportunities around which it will shape and adjust its own identity.

Bednar is more admiring of the wisdom of America’s founders than I am - there is a lot that can go wrong with their conception of politics and has done so. Others, like Jake Grumbach, are much more skeptical about American federalism and variety. But Bednar provides an excellent sense of what a democratic system that prioritizes discovery and adaptation might look like, and of why political polarization threatens these aspects of American democracy. As she explains:

A third consequence of affective polarization is particularly threatening: the compression of the information space, including the suppression of diverse interests and issues. Polarization does more than divide a population; it simplifies it. When people conform to one another, they let go of their differences … not only are groups becoming more sharply divided, but they do so by becoming more like those in their in-group. At the system level, information is lost. When agents conform, the dimensionality of the issue or identity space is reduced. …

Democracy is like a social information processor; at best, it takes the breadth of human interests and ideas and turns them into useful policy to promote social welfare. Democracy relies on the free flow of diverse information to be innovative and productive. Diversity is an asset, and polarization with conformity burns that asset. Additionally, … it also undermines the supports girding our democracy against a slide toward autocracy.

Bringing Bednar and Grumbach together, the problem with actually existing federalism is that has been polarized into two opposing networks of political action with little internal variety. Remaking that will not only require breaking the existing gridlock, but a lot of practical experimentation. My co-authors and I lay out one agenda for doing that here, which focuses on the level of small group decision making. That is certainly insufficient on its own to reverse the larger positive feedback loop of polarization that Bednar and others have identified.

Both Eno’s essay and Przeworski’s chapter are beautiful, but in different ways. Eno’s emphasis on organizing and generating variety is less likely by its nature to generate precise claims than useful heuristics. Plausibly though, we need heuristics more than we need science right now: useful general guideposts to action rather than law-like statements about the causes of democratic stability.******

Indeed, this also fits better with Przeworski’s recent writing, which implicitly stresses the need for social scientists to show humility: we got an awful lot wrong. We’re in a moment of rapid ferment and change, where the fountains of accepted wisdom have mostly run dry. What social scientists may be able to provide is our own combination of uncertainty and variety. We do not have a stronger claim than anyone else to absolute knowledge of what is happening, but we do have a variety of more or less imperfect models, which can clarify thinking and channel it in more useful directions. Eno’s account of variety is not only a valuable model in its own right, but a justification for drawing on the multitude of differences that we need to draw on, turning hidden resources into visible and usable ones.

* It is always when the third generation comes along that things start going badly wrong.

** The ellipses in the text suggest that there is a longer version somewhere, which I haven’t read, but would love to.

*** I could be mistaken! Since the shortened essay is less than five pages long, I recommend you read it for yourself.

**** One person who did draw this kind of connection was the late critic Mark Fisher. Riffing on Fisher, you could tell a story about perceived changes in the possibilities of collective action by tracing the three decade change from the ecstatic communion of the Weatherall mix of ‘Come Together’ to the noctilucent fractures of Burial’s ‘Dark Gethsemane.’

***** This is the “folk theorem” of game theory, summarized by Dan Davies as follows: “The Folk Theorem in game theory states that any outcome of a repeated game can be sustained as an equilibrium if the minimax condition for both players is satisfied. In plain language, it can be summarised as stating that ‘if we take strategic considerations into account, there is a game-theoretic rationale for practically anything.’ This formulation leads on to my contribution, the Davies-Folk Theorem, which states that ‘if we take strategic considerations into account, there is a game-theoretic rationale for practically fucking anything’ (it’s a fairly simple corollary; proof available from author on request).”

****** Though as already noted, law-like generalizations can provide the broad outlines of a score.