On the Non-Triviality of Going from Individual to Collective Rationality

It’s been a mainstay of the radical left for a long time to blame the lack of radical activity by whatever particular collective subject they believe to have potential on some sort of capitalist subterfuge. The various arguments for what exactly happened range considerably, but they tend to assume that a subset of the population who would otherwise revolt against the system have been brought off and/or propagandized into submission.

Obviously propaganda has an effect and cooptation of movements is a constant throughout history. But I think what explanations for our present state of affairs tend to ignore is the basic point that collective action is hard.

To give an obvious example (that appears in most introductory texts on game theory), consider a simple model of a revolution. If enough people commit, then the oppressive government is overthrown, some revolutionaries die and all who survive benefit. If insufficient numbers commit, the revolutionaries are killed/imprisoned and those who don’t take part suffer no losses (again this is a simple model). Finally, the more people take part, the less likely each individual is to incur loss.

It may seem like an obvious decision on the part of those who hate the regime. All they need is to simply determine if they reach the threshold for a successful revolution and then go do a revolution. But introduce basic uncertainty into this model and suddenly it becomes much more difficult. If people have only a limited awareness of how many others are committed to radical change or have doubts about the private commitments of others, what can happen is that large numbers of people can hate the regime and are personally willing to risk to bring it down, but nobody is willing to take the necessary steps because they don’t believe others are committed. Even serious altruists are driven to inaction in this model because they can’t continue to do good for others if they’re dead / in prison.

Now this is a toy model of social change. But it nevertheless captures a basic point about collective action problems: just because something is a net positive from the perspective of an aggregate does not mean that it will necessarily be a net benefit to the individuals making up the aggregate who will do the work of bringing it about.

One can add all sorts of nuances to models of collective action problems to further illustrate the disjunction between what is rational for the individual and what is rational for the collective. Let’s use a simple model of radical social change derived from Marxist assumptions. Assume that the majority of individuals are motivated by self-interest and their conditions are getting increasingly worse thanks to increasing exploitation (although they exist on a spectrum of badness). Once a particular threshold of badness is reached they will revolt (and win because they have numerical superiority), which will bring about a more egalitarian / productive society where every person will see material benefit.

Of course this has never actually happened. The actual history of actual working class movements is considerably more complicated than this simple model (which itself is a simplification of Marxism). One thing it cannot account for are the various turns towards reformism by large ostensibly radical movements.

Again the actual history of the turn to reformism is complicated. But one of the factors was simply the interests of actual working class people. And when we introduce the option to fight for reforms into our simple model we can see how this can happen. If we assume that reforms require less of the population than revolution to be successful, do not result in as high returns, but incur considerably less risk on individuals taking part the rationale for why reforms is obvious.

Since the level of risk and numbers necessary are both lower, you would expect self-interested individuals to choose reform. And hence what you get is an equilibrium point wherein individuals fight for reforms up until the point where the costs expended no longer makes reforms worth it. Because workers are merely responding to immediate deprivation, many will stop fighting once they reach a particular level of comfort. Capitalists don’t need to peel off every worker to reach equilibrium, just enough such that those motivated cannot reach the threshold required for successful collective action.

Even if you think something like the falling rate of profit is in effect and will eventually make it impossible to buy workers off, that in no way necessitates action on the part of workers. Given that no Marxist has shown us how to tell how slowly the rate of profit declines and the empirical failure of predictions about the end of capitalism, it’s pretty rational for individuals to ignore it given that there’s no way to tell when it’s going to occur.

This is because the rate at which profit falls matters considerably when it comes to action. To see why, consider a simple decay function like e^-t*C. You can play around with the rate of decay (C) and increase the time taken for it to reach zero by orders of magnitude. If the profit rate only falls to zero over the course of centuries or millennia it is a non-issue for workers in the present who face only mild dispossession because of capitalism.

Again these simple models in no way capture the complexities of Marxism (let alone actual reality). But it does illustrate a basic point that mere oppression is not enough to motivate radical action that will effectively change society.

Now all this might seem like an argument for vanguardism, for a party of professional, disciplined revolutionaries who can lead the masses around these sorts of incentive traps. But any sort of centralized structure to manage the masses creates its own set of collective action problems.

To understand why, let’s consider another simple model of collective action that asks “when will it be rational for individuals to take action for collective interests?”

Individuals obviously differ in terms of what motivates them to take part. Those who are only self-interested will only act if they believe the returns from engaging in collective action will outweigh the costs.

Those who are more altruistic are obviously willing to sacrifice on behalf of others. But just because altruistic individuals exist does not mean they will be able to undertake effective collective action. After all, a rational altruist who wants to effectively help others will not just sacrifice everything they have at the drop of a hat to the first person who comes along (indeed contra popular associations between altruism and naivety, one might expect any serious altruist to be more rational / calculating in their actions if only because the potential upside is so much higher because of basic diminishing returns to individual consumption).

So is the solution for a vanguard of rational altruists? No.

Even ignoring the problem of determining whether someone really is a rational altruist or the psychological effects of how power shapes people, rational altruists coming together by no means ensures rational group action. There are many pressing problems in the world and no obvious ways to determine what should be focused on or how it should be addressed. Given that any course of action must then involve individuals coming to agree over how to proceed, this creates significant costs in terms of hashing out what’s important and what should be done about it.

So while on paper large voluntary associations may have the ability to leverage much larger resources until they reach agreement, they have a harder time deploying them (and of course agreement is not a one and done thing but something that needs to constantly be reevaluated as conditions change over time). Conversely, smaller groups are, counterintuitively, more effective in certain domains than larger groups at achieving their particular interests because it’s easier to get agreement on what to do.

But the only reason members of small groups are able to efficiently come to agreement is that they can ignore or minimize the interests of those outside the group. The sheer number of possible ways to configure the world and the specific interests of individuals are overwhelming. This is not to say that it is impossible to take the considerations of others into account, merely that once you start to do so, the benefits that small groups have when engaging in collective action diminish.

Hence any vanguard that actually tries to represent the interest of a “class” is caught in an incentive trap far more dangerous than the one our dispossessed proletariat found itself in. To be able to effectively act, they must minimize the concerns of the people they claim to represent. This will happen regardless of the motivations of the vanguard, whether they are cynically using the masses for their own gain or motivated by genuine selflessness. Regardless of motivation there are strong reasons to want to simplify the problem so that it can be made tractable.

One simple practical example: say you’re an upstart revolutionary who overthrows the government by promising land reform. Well that’s a relatively simple change (as seen plenty of times throughout history peasants are pretty good at taking land from their landlords). But having just created a class of people who now no longer have dependence relations and have capacity to produce for themselves, you now have a set of people who can resist your edicts. If you don’t suppress your liberated peasants new actions made by you will have to take them into account because they have more capacity to resist your edicts.

And this means that such power over others is definitively not a universal currency. When you have it is simply easier to act selfishly than not. Combine this with the insight that a small number of people have an easier time coming to agreement and you have a simple model to explain why class differences emerge in society where there are positions that give people significant influence over others, even if property relations have been dramatically reshaped. The problem of weighing up concerns just makes it hard for any individual or institution to effectively direct others on behalf of a “class interest”, whether that be leading it in the fight against an oppressor or organizing things in an egalitarian manner after the oppressor has been vanquished.

All this might sound defeatist. And yeah, if you’re from the traditions of the left that think that it’s trivial to go from individual rationality to collective rationality, this all kinda sucks for your grand “scientific” theories (although if you come from such a tradition you probably stopped reading after I started talking about individual incentives and you accused me of falling for bourgeoisie ideology or whatever). But just because elites have an easier time of achieving coordinated action does not mean they are efficacious in what they try to do. The mechanisms of control they have are hampered by the limits to information flow / processing that hamper larger groups. The considerations required to effectively enact change over the masses would require the sort of deliberation that hamstrings collective action in large groups. When they try to affect society at scale, they are necessarily limited to blunt tools.

This tension between the ruling class of any society having the means to enact change, but being restricted to clumsy tools is a far more fundamental social “contradiction” than any claims about where the exchange value of commodities come from or whatever. Limits to control can be directly derived from physical limits to information processing and flow. This is not some contingent fact particular to any arbitrary social arrangement, but is something either fundamental to the universe or very close to it. It has shaped power relations in every society we’ve ever had and will continue to do so into the future (artificial minds obviously have limits as defined by things like information theory).

These dynamics are of immense importance but are beyond the scope of this short essay. So instead I’m going to retain my coarse grained perspective so as to highlight some general strategic insights that come with more awareness of collective action problems.

The first is that the mechanisms we create to solve collective action problems are non-obvious and come with tradeoffs. This might not matter for immediate projects where goals are clear and everyone has a pretty good idea where everyone stands. But when making decisions about the long run, the mechanisms you set up have considerable opportunity costs in how you allocate resources and also achieve some degree of lock-in when it comes to mechanisms because particular ways of solving collective action problems are directly in conflict with others.

To give a stark example, contrast a hierarchically managed mass party versus a distributed network of individuals mediated by polycentric institutions, decentralized technical infrastructure, and social norms that facilitate more fluid ways of organizing. These systems are not just diametrically opposed in how they operate, but also contain within them strong individual incentives to prevent the other from forming. When a mass party doesn’t exist and people are already doing it for themselves it’s difficult to justify it for the purpose of coordination / communication. Conversely, the sort of benefits that come with constructing networked ways of relating take time and effort to build and those with access to some centralized structure may see it as unnecessary when they can just plead their case before it.

These trade-offs are further exacerbated by the fact there is no universally efficacious approach to overcoming all collective action problems – centralist and decentralist approaches both have their strengths and weaknesses.

Understanding the weaknesses or vulnerabilities of decentralized approaches is vital. That I admit centralization and hierarchy have a comparative advantage in some domains in no way implies that I think they are long term solutions. Centralization might be optimal for a specific task, but the problem of reorganizing individuals once the task is complete is a serious problem. Social change is fundamentally an open, iterative process. Increasing the freedom of individuals necessarily means creating new dynamics and that means new collective action problems to solve and that might require significant restructuring to manage. It cannot be reduced to an engineering problem.

Yes there are many unnecessary processes in our present society that exist to siphon resources to elites or suppress the autonomy of individuals that if removed would simplify things. But in the final analysis, new possibilities means new opportunities and problems. After all, a freer world is one with more overall options, where individuals gain more capacity to reconfigure themselves, the world around them, and their relationships. This is obviously good but we should never pretend it will never result in novel problems.

That radical change brings unanticipated consequences is a foundational conservative argument. But there’s different ways to change the world. Bottom-up approaches that see individuals negotiate with each other instead of having changes thrust upon them by some outside entity are radically different precisely because bottom-up approaches allow for a more fine grained approach where individuals negotiate between themselves instead of relying on sweeping edicts from above (such bottom-up negotiation can still get ugly but it does avoid the sort of horrible outcomes that top-down approaches are prone to). Furthermore, bottom-up experimentation also allows the mapping of the possible ahead of time, allowing individuals to identify both failure modes and unanticipated benefits ahead of time.

Hence making our ends concordant with our means isn’t just virtuous, it also has instrumental value in letting us probe, evaluate, and construct various mechanisms to resolve collective action problems in more dynamic and open ways.

Bottom-up ways of doing things also lets us minimize reactionary blowback or opportunism. One clear advantage that hierarchical forms of relating have is that they can come into existence much faster. Dominance relations obviously backed by the clear incentives of force can be imposed on people easily. More organic ways of relating require more complex solutions and those just take time to build and popularize. When shit is breaking down and people aren’t sure if they’ll see tomorrow, it is perfectly rational to go with hierarchical solutions to a collective action problem just to survive, even if it reduces their options in the future.

All of this is anarchist common-sense to varying degrees. But there’s value in not just stating the obvious, but also showing how formal theoretical frameworks can support it. Being able to formalize one’s intuitions lets you explain yourself better to those who don’t share such assumptions, as well as being able to extend your insights beyond the immediate, identifying domains where they might break down.

Because however common-sensical the limits of collective action might be, they are frustratingly underemphasized when it comes to attempts at formalizing capitalism and structures of domination more broadly. (Kevin Carson deserves significant credit for doing the work of presenting a comprehensive account of capitalism that has collective action problems at its center). This is particularly frustrating because throughout history we see again and again how limits to control shape how societies operate and technology develops – from the earliest states imposing crops on peasants that were easier to monitor at the expense of poorer nutrition-per-hours-of-labor-expended to capitalists selecting productive technologies that prioritized control over workers at the expense of efficiency.

Leftists will sorta admit that this is a thing, but outside of Carson there are no real attempts to build theories of capitalism that have it at their core. See for example in 2015 when the late David Graeber admitted that “The Right, at least, has a critique of bureaucracy. It’s not a very good one. But at least it exists. The Left has none.” Such an admission speaks volumes about how badly anticapitalists have done on this front.

There’s an immense opportunity cost to failing to integrate and popularize such frameworks. I seriously believe that the failure of leftists to understand collective action problems has been an obstacle on par with all the armies, infrastructure, institutions, and propaganda that uphold the status quo. Indeed it might very well be a bigger failure because the systems that maintain power also face their own internal collective action problems. Collective action problems cut both ways, complicating attempts at expanding and squashing freedom.

There’s million reasons for why anticapitalist movements failed in the last century (advertising, post-Fordist workplaces, global supply chains, neoliberal think-tanks, consumerism, etc). And yes all of that certainly shaped things, but the basic point that collective action is hard should really be the null hypothesis for why any attempts to change things failed or resulted in unintended consequences (as well as why authoritarian regimes or movements made critical mistakes, for example!). Trying to change things is hard enough, but it’s even harder when you shoot yourself in the foot by adopting poor models of the world that promise things that never come.