Arguing About How the World Should Burn

After eavesdropping on a thousand Twitter arguments and reading just as many thinkpieces, I’ve noticed that there are two main ways of conceptualizing community governance. Both are normative. Both primarily arise when it comes to conflicts over free speech — or who it’s okay to punch.

glitched woman hitting neo-nazi with handbag

A glitched version of the famous photo of an old lady clobbering a Neo-Nazi with her purse. Original photo taken by Hans Runesson in Växjö, Sweden circa 1985.

One mode is to focus on content, and the other is to focus on process. These are two different paradigms that shape people’s reactions to a given controversial issue. For example, let’s say a major news event happens. How do you go about selecting the articles you’re going to read about it?

  1. Choose the outlets or individual writers who share your worldview, and see what they have to say.
  2. Seek out authors who signal thoughtfulness across the political spectrum, perhaps with an emphasis on original reporting.

Here’s another example: Computer scientist Curtis Yarvin applies to speak at a technical conference, and his talk is accepted in a blind evaluation process. After Yarvin’s speaking slot is announced, the conference organizers find out that he has an alter-ego as a blogger called Mencius Moldbug. His blog promotes views that much of the conference’s larger community finds abhorrent.

The content approach is to sever ties with Yarvin — because the content of his character has been judged to be objectionable. His work and their personality are deemed toxic or actively harmful. On the other hand, the process approach is to point out that Yarvin’s submission was selected blindly, on its own merits, and affirm that he will be ejected only via the process laid out in the pre-established code of conduct.

To put it in more abstract terms, content people focus on ends over means, and process people focus on means over ends. This is an imperfect way to summarize the principle, because the reason why process people focus on means is that they think this approach leads to better ends. But “ends over means and vice versa” will do as catchy shorthand. (The distinction is similar to the conflict between deontological ethics and consequentialism, although my emphasis is less on the philosophy and more on the praxis.)

Nate Soares encapsulates what the two sides of any argument miss about each other:

When I was younger, I’d regularly see people taking actions that I would never take unless I was acting maliciously. I would automatically, on a gut level, assume that the other person must be malicious. Only later did my models of other people become sufficiently diverse to allow me to imagine well-intentioned people taking actions that I would only take if I were being malicious, via differences in ways of modeling the world, choosing actions, or coping with feelings of defensiveness / insecurity / frustration / etc. that stem from benign motives.

Soares started out with the assumption that everyone perceived the world the way that he did. Then became aware that he was committing the typical mind fallacy — “the mistake of modelling the minds inside other people’s brains as exactly the same as your own mind.” (As another example, Democrats do this when they claim that Republicans want poor people to die.) He was able to adjust his methods for judging others to account for diverse paradigms. (Theoretically, Democrats could do this by recognizing that Republicans have different priorities and assessments of policy effects, while sharing the goal of making the world better for everyone.)

Neither the content approach nor the process approach is objectively correct, although it’s perfectly valid to prefer one or the other. Proponents of both are trying to optimize for community health as they conceive it, but based on different criteria. Content is about correctness whereas process is about fairness.

Friendly Fire Defined by Framework

I believe that effective principles for governing society have to account for individuals and groups who disagree with each other all participating. The rules should be robust even when there are rival factions. But not everyone agrees with me — some believe it is better to eject those with conflicting paradigms. Neither position is correct, per se. Rather, we hold different values that optimize for different types of communities.

The internet social justice movement sometimes baffles its critics because of its focus on ends. The uncharitable way of describing their attitude goes, “It’s okay when we do it.” What’s missing is the addendum “because our ends are just.” There is no hypocrisy when certain feminist circles shun an outspoken misogynist but welcome an outspoken misandrist, because they’re not working with a process mindset. They are not optimizing toward a “fair” process; that is explicitly not what they want nor what they deem desirable.

Take the issue of whether saying “men are trash” is hate speech. The social justice position is as follows, articulated by Libby Watson:

“Men are trash” is simply not the same thing as sexist attacks on women, because you don’t face the systemic oppression that makes sexist slurs so toxic. If you’re a man and someone calls you a stupid trash man, wow, I’m sorry you had a bad day. If someone calls a woman a bitch or a slut, that’s tinged with millennia of oppression.

It is obviously unfair for “men are trash” to be considered acceptable when “women are trash” wouldn’t be. But that’s not the point. As another example of intentional double standards, here’s what a leftist wrote in response to the Nazi-punching debate that boiled over in response to Richard Spencer getting clocked:

An anti-fascist outlook has no tolerance for “intolerance.” It will not “agree to disagree.” To those who argue that this would make us no better than Nazis, we must point out that our critique is not against violence, incivility, discrimination or disrupting speeches in the abstract, but against those who do so in the service of white supremacy, hetero-patriarchy, class oppression and genocide. The point here isn’t tactics, it’s politics.

See? Ends over means.

In the same vein, there is no hypocrisy when free-speech activists advocate for both outspoken misogynists’ and outspoken misandrists’ rights, with the aim of ensuring liberty for all. Their focus is on the process of protecting free speech rather than the content of what people are choosing to say. Many of the same people believe that no one should be assaulted, regardless of how controversial their political views are.

Slate Star Codex’s 2014 classic “In Favor of Niceness, Community, and Civilization” makes the quintessential process argument against a content position:

Civilization didn’t conquer the world by forbidding you to murder your enemies unless they are actually unrighteous in which case go ahead and kill them all. Liberals didn’t give their lives in the battle against tyranny to end discrimination against all religions except Jansenism because seriously fuck Jansenists. […] Liberalism does not conquer by fire and sword. Liberalism conquers by communities of people who agree to play by the rules, slowly growing until eventually an equilibrium is disturbed.

Tragically, content-focused groups and process-focused groups often have common goals, despite their disparate approaches. The shared ground gets obfuscated when partisans can’t identify the root of their differences, or attempt to see the world through the other side’s eyes. The smoke of the battle covers their overlapping territory, if you will.

The different approaches are sometimes practiced according to ingroup-outgroup divides. Ingroup members are subject to process-based evaluations, whereas outgroup members are dismissed entirely because of who they are. Think about what I said earlier, that the content approach prioritizes correctness while the process approach prioritizes fairness. Members of the outgroup don’t conform to your norms and values, and that’s the basis on which you judge then. Ingroup members share your cultural assumptions and expectations, and you hold social power over each other. It makes sense that you’d want the governance process within your ingroup to incorporate some kind of due process, lest you be unfairly ejected.

For example, tribal blood feuds were governed according to complex rules, like the Albanian Kanun, although the parts of that code intended to deescalate have been neglected, perhaps because modernity has broken up the community institutions that managed the process. The norms of the Nuer tribe (see slide 24 here) also qualify.

The Dordrecht Confession, a set of doctrines followed by Anabaptist groups like the Amish, says that “in shunning as well as in reproving, such moderation and Christian discretion must be used, that it may conduce, not to the destruction, but to the reformation of the sinner. […] Therefore, we must not count them as enemies, but admonish them as brethren, that thereby they may be brought to a knowledge of and to repentance and sorrow for their sins, so that they may become reconciled to God, and consequently be received again into the church[.]”

An erring ingroup member is put through a process of atonement and reconciliation. Anarchist accountability processes are meant to work the same way.

However, some ingroups maintain opposite policies. Consider this anecdote about President Trump and ousted FBI Director James Comey:

As they ate, the president and Mr. Comey made small talk about the election and the crowd sizes at Mr. Trump’s rallies. The president then turned the conversation to whether Mr. Comey would pledge his loyalty to him.

Mr. Comey declined to make that pledge. Instead, Mr. Comey has recounted to others, he told Mr. Trump that he would always be honest with him, but that he was not “reliable” in the conventional political sense.

A loyalty pledge is content-based: “Because you are who you are, I will support you and stand by you.” A promise to be honest is process-based: “I will base my stances on what I know to be true.

Deconstructing the Emperor’s Clothes

You can look at the content paradigm and the process paradigm as competing egregores. Each paradigm can be experienced by individuals, but it becomes something larger and more powerful when many individuals experience the framework in tandem, binding each other to it through social cohesion incentives. Sarah wrote in “Weaponized Sacredness”:

Egregoric entities, whether gods or demons or dictators at the center of a cult of personality, are powerful entities, even as they are imaginary — wholly created by and consisting of thoughts, speech, and behaviors. The human mind is a powerful entity, and becomes more powerful in coordination with others. To say that an “imaginary” entity, existing only in human minds, has agency, is not much stranger than suggesting that humans themselves, inscrutable piles of preferences that we are, have agency.

Sarah noted later in the essay, “Sometimes a sacred entity is personified, as with gods or demons or the centers of cults of personality. At other times, the sacred entity is composed only of abstract ideas, refusing to personify itself. Why might it benefit such an entity to hide its nature?”

My guess is that certain egregores — or certain sociocultural interpretative frameworks — tend to lose their potency when the bearers recognize them. When you figure out how they function, you see the way that you’ve acted as a cog in the meaning-making machine, and you’re able to decide whether you still want to serve that role.

For me, the content-versus-process dichotomy worked like that. In the past I’ve held the content stance. More recently I’ve held the process stance. In fact, I still tend to gravitate toward and defend the process crowd by default, despite being aware of the blinding effect of dogged adherence to a single simple framework. But recognizing the structure of both dogmas loosened the hold of the entire duality.

It helps to realize that neither paradigm will grant liberty to its adherents. Social governance and freedom are incompatible. I don’t mean that ideologically (to the extent that any assertion about politics can be non-ideological). I mean that the whole purpose and function of governance is to impose constraints. Personally I don’t think this is inherently bad — but I do think it’s inherent.

In a society of one, you are at perfect liberty to do whatever you want to. You can follow any whim that strikes you, take any action you want, and yell whatever you want. You may be constrained by available resources, but not by rules or status games. No one will try to get you fired, or toss you in jail, or even shush you. Your choices are limited only by your environment and your own capabilities.

A society of two is immediately more complicated. You have to decide how to cope with each other — rivals, partners, or an uneasy indifference? What if you want to be partners but they take advantage of your goodwill and act as a rival? What if you want mutual deference but they try to encroach on your territory? A society of three introduces the possibility of factions. And on it goes, as you add people.

Some constraint on one’s options is acceptable to most of us. We’re social creatures; we need companionship. Cooperating and specializing is also good for productivity, and therefore survival. When you decide to have governance, to codify the norms of your community, you’ve already decided to limit people’s freedom. Your next decision is what constraints you’re going to impose. What will you punish and what will you tax? What will you encourage and subsidize?

Content and process are two dueling answers to this question, championed by fighters who mostly don’t understand the cause they’ve been marshalled to defend. Those who espouse the content approach want to push people out when they don’t buy into the norms and beliefs of the majority. Those who espouse the process approach want to compel everyone to obey the same rules. Neither side will ever be satisfied as long as the other exists.

Get Ribbonfarm in your inbox

Get new post updates by email

New post updates are sent out once a week

About Sonya Mann

Sonya Mann is a writer of various sorts. She runs a cyberpunk newsletter called Exolymph and spends a lot of time on Twitter.


  1. Thanks for framing this discussion in a way that’s so cogent!

    To give my own two cents, I think it’s helpful to to point out that both content-based and process-based social norms have pretty well-documented failure modes. For example, the rules in content-based systems can be very difficult to predict and can change rapidly in ways that appear arbitrary and capacious to outsiders (think basically every criticism of the ever-changing list of acceptable terms to use in social justice spaces). Meanwhile, process-based systems can become calcified into rituals that are obviously bad, but everyone accepts because, hey, they’re following the rules.

    This link is a good illustration of the latter:

    • Eugine Nier says

      Another failure mode of process oriented norms, and a big asymmetry between the two, is that process oriented groups have a hard time defending themselves against infiltration by content oriented people. If a group starts out with process oriented people, any person who completes or is approved by the process can join, including content oriented people. However, once they join they will ignore the process in favor of including more people oriented towards the same content in the group.

      • YES. This is a huge social problem that’s hard to guard against. I’m only aware of a few groups that manage this successfully.

    • Very good point!

  2. David Sims says

    Awesome article Sonya, the content versus process dichotomy is a really clean framework for interpreting the dogged motivations of both sides.

    I think I’ve held a similarly conceptual division between these two camps (though not as well formed or expressed as yours) however rather than viewing them as distinct and seperate, I see them as representing a symbiotic adversarial relationship. You note the tragedy that these divergent groups can often share common goals; what if this isn’t entirely tragic, but indicative of an underlying collaboration. Not a predetermined or considered collaboration, but a structurally induced collaboration in the same way two businesses compete for market share in a capitalist system in order to encourage innovation through competition.

    This would also address the fact that the rules and freedoms advocated by the process camp are not immune to the influence of the content camp – even with their divergent world models. I think this is becuase the process camp represent a set of predefined liberties that, through historical experience and conceptual validity, present a model that will optimise equitable fairness. But this legacy does impart a degree of institutionalism on them and lend to their being equated with the status quo.

    I could ramble on, but I’ll be strict with myself and try to wrap up my thoughts in one last paragraph. Essentially, I’ve always viewed the process camp as equivalent to incrementalists – they share liberal ideals and goals, but balance this with an ingrained pragmatism that sees change as best realised and sustained through a slow and.steady grind, rather than a bang. On the other hand, content people appear to actively occupy the role of extremist. By this I mean, they agitate and aggrevate with the underlying belief that such confronting views and behaviour will force awarenesss and change in their desired direction. They’re unashamed envelope pushers. The real fascination I find here is that envelope pushers do hold an advantage in their position because their actions on the fringe can disrupt and reframe what is perceived as being moderate, and I think this links back with the idea that the process camp is invariably influenced by the content camp over time.

    Anyhow, this one got me thinking – it was great, good work! :)

  3. Alastair Roberts says

    Thought-provoking post. The content-process distinction also often seems to intersect with the distinction between utopian and practical people. The content approach also tends to be married to a Whiggish view of history, which assumes that the arc of history is moving in its general direction and that it is both on the right and more powerful ‘side of history’. By contrast, the process approach is much more dubious about the inevitability of desirable and just outcomes from the game of history and is more concerned to ensure that it is played by fair rules that protect the losing side. Such a vision of history is more amenable to compromise, negotiation, accommodation, provisionality, etc., without necessarily requiring the abandonment of a vision of and pursuit of more radical social goods.

    • Alastair Roberts says

      In my experience, which has caused me to become rather more cynical on these fronts over the last year or so, content or process positions often rely more upon a sense of what benefits one’s side than on deeper principle. It is interesting to observe how the defenders of free speech tend to be the people who are the weakest in a given situation. Placed in situations where they have the advantage, these same people reveal themselves to be content people, masquerading as process people. There are also a number of process people (Freddie deBoer may be an example here) who strongly favour playing by ‘fair rules’, but consistently stress that they do so because they are convinced that they serve their desired ends. Their commitment to the means as such is not always clear to me, as if the supposed connection between those means and their preferred ends were disproven, it is by no means clear that they would stand by the means. It seems to me that the most committed process people either operate more with a sort of Rawlsian veil of ignorance, being uncertain about how things will pan out and wanting to ensure that no one gets mistreated in the process, or with an even more strong deontological commitment to the means themselves, even beyond their ensuring desired fairness.

      • Interesting point — it reminds me of epigenetics. Self-interest could be the underlying principle that is expressed differently depending on the context.

    • Alastair Roberts says

      A few further things.

      First, More’s speech to Roper in A Man for All Seasons is a great process versus content argument.

      Second, appeals to ‘process’ and perhaps even more so ‘content’ are seldom as straightforward as they seem, as such positions usually serve purposes beyond their ostensive ones. See SlateStarCodex’s ‘The Ideology is not the Movement’, for instance. I would beware of taking people’s commitment to these things at face value. More often than not, the driving impetus for such commitments lies elsewhere and the appeal to process or content is designed to give leverage to other ends.

      Third, Edmund Burke argues, rightly I believe, that many of the restraints upon us, and not merely our liberties, should be reckoned among our rights and the grounds of our freedom. So governance is not necessarily a limit upon freedom, unless freedom is defined as autonomy. Under certain conditions limits can make us freer, enabling us to achieve things that we couldn’t achieve otherwise.

  4. Romeo Stevens says

    Some other common dual-process frames carving at the similar, though slightly different joints:
    Systems vs Goals
    Inputs vs Outputs
    Content vs Context (I think the word context can load some very different intuitions than process)
    Forward chaining vs Backward chaining (effectuation vs theory of constraints maybe?)

    more interesting overlaps?

    • Oh dang systems versus goals is particularly good, that might have been better terminology to use in this case!

      • I actually don’t think so. Systems and goals are both “process” in your sense of the word — ways of structuring behavior (habit formation versus planning) independently of the content of what the behavior is about. Content in your sense is only partly about being goal directed. It is mainly about processes being subservient to judgments rather than substitutes for them. You can be goal directed in a very procedural way. It’s harder, but you can be “content” focused in a systems way too. For example, redlining was a “system” that had a “content” effect (denying blacks housing).

        • I disagree with Dr. Rao. Goals are not processes, they are objectives or outcomes. The general applicability of the distinction is that goal focus can overlook the persistent application of processes needed to get there. However, I agree that it’s not the best way to think about this subject.

  5. Consequentialist vs Deontological models can sometimes be thought of via a utility function optimization perspective which leads to the following distinction:

    If you have a single goal, the consequentialist / ends-over-means / content-based approaches are correct, because they optimize for that single goal. Consequentialist models, however, are hard to solve for. In the face of uncertainty (and often even without it), how do you decide what action or series of actions is optimal? Course corrections when you get new information can lead to “discontinuities” in this approach, examples of which have become famous paradoxes like in the Trolley problem (What if one of the people on the tracks is a close friend?)

    If, however, you have multiple goals (say, competing goals or subgoals leading up to a final goal), or your goals may shift over time, the deontological / means-over-ends / process-based approach seems like the better one. Instead of trying to predict the consequences of your actions, pick your next action according to a set of virtues or processes that approximately lead closer toward your current goal or are shared by your multiple goals, and move a small step in that direction.

    Eventually in the multiple goal setting, you will come to a point where competing goals lead you towards different actions. At this point, you should probably pick a goal to optimize and switch to the consequentialist point of view.

    I obviously have my own biases, but in real-world situations, I would say that the process-based approach is superior but can end in some kind of conflict and can necessitate a switch to the content-based approach.

  6. Daniel C says

    I’m reminded of something Lowtax said in a recent interview, on striking a balance between rule-based and content-based moderation:

    “I find Twitter’s situation to be of their own making. They never concretely set out a set of rules. When I first started the forums, I wrote four pages of rules and a catch-all at the end: If there’s something else we don’t like, we’re going to ban you. We have every right to ban you and that’s it. With Twitter, they never defined anything. They never said what’s allowed, what isn’t allowed, what will happen. They just kind of floated around. If something got really out of hand they would get rid of it, but since they had no concrete rules, they had no active moderation, people didn’t know what was or what wasn’t allowed. They dug their own grave and now they’re way too far into it to dig out.”

  7. To add to Romeo’s list of similar distinctions:

    The bias-variance tradeoff: the fundamental problem of how a model or intelligent agent can generalize from past experience. A high bias / low variance model (“process”) has few parameters and is therefore simple, easy to understand, and can be computed quickly. It also cannot adequately represent the full complexity of reality and makes poor predictions even when tested against familiar experiences. A low bias / high variance model (“content”) is complex, has many parameters, makes arbitrary exceptions for every edge case, and is relatively slow and expensive to compute. This model predicts past experience very well, but fails to generalize to new experiences. It is often possible to trade a small amount of increased variance for a large reduction in bias, which usually leads to better predictions (read: better decisions) in novel situations.

    Also in Robert Kegan’s Constructive Developmental Theory, “content” roughly corresponds to stage 3, while “process” is roughly stage 4. Stage 5 has to do with finding a synthesis of the two modes (or perhaps by default engaging in a dialectic that can find many syntheses) in order to avoid a “dogged adherence to a single simple framework” while still understanding the utility of prior modes.

  8. Jim Stone says

    I wonder about correlations with other personality measures.

    Is content-orientation associated with low openness and high neuroticism (to use big-5 categories)? Is process orientation associated with high openness and low neuroticism?