Thought Experiments in Governance

All have flaws, but the search is not for perfection, it is for innovation.

Photo by Alex on Unsplash

At some point in our history, we stopped treating governance as something you can play with.

If you take David Graeber and David Wengrow seriously in The Dawn of Everything, human societies used to experiment quite wildly with how they organised themselves. There were cities that lived like monarchies in one season and switched to egalitarian councils in another. There were ritual hierarchies that flipped upside down at particular times of year, so that those at the bottom spent a while at the top. People tried centralized authority, then abandoned it, then brought it back in a different form. Governance was not a single track. It was a repertoire.

Then, somewhere along the way, we fell in love with the idea that there is one true form. The nation state. The professional bureaucracy. The constitutional democracy. The party system. The market as allocator of almost everything. We argue about which tweaks are better, but the basic structure is treated as sacred. We do reforms, not experiments. We patch, we tinker, we defend, but we rarely say, what if we are wrong about the entire shape of this thing.

So here is the premise. If people once experimented with governance on purpose, then there is nothing metaphysical blocking us from doing that again. The capacity is still there. What is missing is practice and permission.

I want to suggest that we consciously restart the habit of governance experimentation. Not in the abstract sense of another white paper about “rethinking democracy,” but in the literal sense of prototyping weird systems, running them in bounded ways, seeing how they work, noticing how they fail, and feeding that back into the next round. Not because any one design will save us, but because the act of experimentation itself is a way out of being stuck.

To make that less vague, I want to walk through a handful of strange governance models. Each one has a name, a basic idea, a reason it is interesting, and a set of obvious flaws. None of them is ready for general use. All of them are dangerous in their own ways. That is the point. Knowing the flaws is not a reason to abandon experimentation. It is exactly what tells us where to look next.

The first sketch is what I will call Field Knot Governance. Instead of thinking in terms of citizens and territories, this model treats the world as overlapping fields of consequence. Housing is a field. Watersheds are a field. Schooling, elder care, heat stress, pollination, all of these are fields. There are no mayors or parliaments in this picture, only temporary knots that form inside a field when it is under visible strain. People, tools, and organisations show up as capacities, not as abstract individuals. A knot gathers those capacities, makes sense of the situation, adjusts patterns, then dissolves.

Field Knot Governance has an obvious appeal. It honours the fact that problems are contextual. It tries to prevent power from freezing into permanent offices. It makes coordination something that happens in intense bursts around real issues, instead of through a standing layer of professional managers. At the same time, once you strip away the pretty language, it clearly inherits the classic strengths and weaknesses of adhocracy. Flexible teams can respond quickly, but informal elites rise just as quickly. People with more time, confidence, or social capital will dominate knots without ever being officially in charge. It becomes hard for ordinary participants to tell whether they are genuinely shaping outcomes or simply feeding information into someone else’s pattern machine. In other words, Field Knot Governance is a useful experiment to imagine, but it brings the risk of a permanent class of serial “fixers” who quietly capture the whole show.

Seeing that clearly is not a defeat. It tells us that any fluid, knot based model must take seriously who gets to keep showing up, how fatigue and attention are distributed, and how we prevent charisma and free time from becoming the real constitution. The experiment gives us design constraints, even if it never leaves the page.

The second model is Constraint Lattice Governance. Instead of issuing policies, budgets, or programmes, this design imagines governance that only defines structural patterns that are not allowed to exist. You do not write a law that says employers must do X and Y. You define a forbidden pattern that says no actor may both set safety standards and profit from violating them. You do not pass a policy that says “more social housing.” You write a rule that says no entity may own more than a certain fraction of habitable units in a given area. The focus moves from behaviours to shapes of power.

In theory, that sounds elegant. Rather than micromanaging what people do, you prune away the structural traps that reliably produce harm. You ban certain ownership configurations, certain concentrations of decision rights, certain kinds of opacity. In practice, it runs straight into the hard limits of prohibition. If you create a rule that says “pattern P is not allowed to exist,” and the surrounding system strongly incentivizes pattern P, then pattern P does not vanish. It retreats into more sophisticated disguises. You get pyramid structures hidden behind shell companies. You get de facto monopolies dressed up as franchise networks. You get regulatory bodies that look independent on paper and act as captives in reality. Prohibition does not remove the thing, it wraps it in a black box and raises the profit of anyone skilled enough to operate inside that box.

Constraint Lattice Governance is still worth thinking with, because it shifts attention from individual misconduct to structural design. It also teaches a hard lesson. A governance system that relies mainly on banning forms, even abstract forms, is almost guaranteed to create a theatre of compliance alongside a hidden world of real decisions. That does not mean we never forbid anything. It means that prohibition should not be our main tool, because it produces exactly the opacity that makes power hardest to challenge.

The third model pushes in a different direction. Call it Mutual Exterior Governance. Here, every community is governed only by its neighbours. Imagine a map of many small communities, each with a few thousand people. The baseline rules for Community A are written and updated by an exterior ring of other communities, not by A itself. Community A, in turn, sends delegates to help write the rules for some other community. No unit is ever allowed to legislate for itself. No one is sovereign over their own charter.

The attraction here is that it directly attacks one of the most stubborn failure modes of self rule. When the same people who benefit from a system also write its rules and police them, capture is not an accident, it is an attractor. Mutual Exterior Governance blocks that path. You cannot write laws that entrench your own local privilege, because you never hold the pen when it comes to your own basic framework.

The risk is obvious as soon as you say it out loud. If neighbouring communities are divided along ethnic, religious, or class lines, a ring of neighbours can coordinate to punish communities they consider an out group. It is easy to imagine racial blocs or sectarian coalitions using this sideways power to crush minorities. You can try to soften that with careful graph design, with reciprocity rules, with rotation and oversight. You will not remove the possibility. Mutual Exterior Governance shifts the location of capture without guaranteeing that it fades. That is still useful knowledge. It reminds us that taking away self rule does not automatically remove oppression. It simply moves the playing field, and any experiment in that direction would need serious protections against coalition targeting.

The fourth model is Blind Puzzle Governance. Instead of shifting who makes rules, this one tries to change what those decision makers can see. Every governance question is turned into an abstract problem. For example, a real world decision about allocating budget between three neighbourhoods is transformed into a puzzle where Group X, Group Y, and Group Z have certain baseline indicators, certain risks, and certain potential gains. Decision makers see those indicators. They do not see whether Group X is their own class or their political enemies, and they do not see the names of the places involved. The system requires 3 layers of people involved in the decision process, Framers, Solvers, and Binders. Framers translate messy reality into puzzles with anonymized variables. Solvers choose options based on those variables and on agreed fairness metrics. Binders take those decisions and apply them back to the real world.

The hope is to break identity based coalitions by removing the information they need to target or protect specific groups. You cannot pass a budget that rewards your supporters and punishes your enemies if you do not know which abstract profile belongs to which side. Blind Puzzle Governance partially achieves that in theory. It clearly relocates politics to the framing layer. Which variables matter. How disadvantage is encoded. What counts as a fair distribution. Those are not neutral choices. If framers, solvers, and binders collude, they can quietly rebuild identity into the puzzle. Even without explicit collusion, bias can be smuggled in through categories and metrics. The model is especially interesting for narrow domains like transport scheduling or certain kinds of procurement, where problems really can be stated as puzzles. As a general approach to governance, it raises as many problems as it resolves. Again, that is not a failure. It teaches us that anonymisation without deep attention to framing is a polite fiction.

The fifth model is Avatar Snap Governance. Here, instead of anonymised problems, there is an anonymised population. The system generates a layer of synthetic citizens, avatars whose profiles are statistically derived from the real population. Each real person is secretly linked to one or more avatars. All governance decisions are made for avatars only. Decision makers see that Avatar X is rich, Avatar Y is poor and sick, Avatar Z lives in a dangerous floodplain. They never see which avatars are tied to themselves or their allies. At intervals, the consequences that have accumulated for each avatar snap back into the life of the real person behind it. If you consistently vote for rules that make life worse for precarious profiles, you cannot know whether you are quietly constructing your own future conditions.

Avatar Snap Governance takes the old philosophical idea of a veil of ignorance and tries to turn it into a procedural rule. You are never allowed to know exactly who you are governing, and the system guarantees that you will eventually end up living under the choices you made for someone statistically similar to you, whether you know it or not. The dangers are large. Whoever runs the avatar system and the snap mechanism holds immense meta power. People might reject the whole thing as manipulative. Mapping consequences back from synthetic citizens into material conditions will always be fuzzy and unfair at the edges. Once again, the value is not in treating it as a blueprint. The value is in how it forces us to ask whether we really believe our own moral claims about treating others as we would wish to be treated, and what it would mean to enforce that structurally instead of leaving it to personal virtue.

Field Knot Governance, Constraint Lattice Governance, Mutual Exterior Governance, Blind Puzzle Governance, and Avatar Snap Governance. Five strange shapes. None of them clean. None of them safe. Each of them pressing on a different stuck place in our current imagination.

The question is what we do with shapes like this. One option is to treat them as intellectual curiosities and then go back to arguing about minor electoral reforms and slight tweaks to parliamentary procedure. Another option is to treat them as early sketches in a practice of experimentation. That second option does not mean we build any of these models at full scale. It means we accept that serious experimentation will produce designs with obvious flaws and that those flaws are not disqualifying. They are information.

What would it mean to experiment responsibly with governance today. First, it would mean admitting that our current forms are not neutral baselines. Representative democracies captured by parties and money are one experiment that has largely frozen into dogma. Technocratic bureaucracies are another. Authoritarian capitalism is another. They are not the natural end state of political evolution. They are local equilibria, and they are showing their cracks.

Second, it would mean shrinking the scale of courage required. You do not start by refounding a nation state. You start with city districts, digital communities, cooperatives, and specific domains like school governance or land use boards. You treat each experiment as something that must be reversible, or at least correctable, within a known time frame. No permanent offices. No immortal councils. Every structure comes with a sunset condition and a simple way to step back toward something more familiar if it really hurts.

Third, you build in the expectation of failure. A governance experiment that does not throw up obvious flaws is probably not doing anything new. The question is not whether a design has drawbacks. The question is whether people can see the drawbacks in time, and whether those who live under it can say no, enough, stop, without getting crushed. Trying out a very gentle version of Mutual Exterior Governance in a handful of municipalities, with a strong veto and exit option, is a very different thing to imposing it on an entire country by force.

Fourth, you make the learning public. When a model fails, you document how it failed. Did Field Knot processes burn people out and turn into insider games. Did Constraint Lattice rules create shadow empires behind the law. Did Blind Puzzle allocations erase important cultural knowledge. Did Avatar Snap mechanisms feel random and punitive. Did Mutual Exterior rings encourage spite between neighbouring communities. That knowledge should feed into the next round, not be buried to protect anyone’s reputation.

Finally, you treat governance design less like writing sacred scripture and more like maintaining open source software in a world that is already on fire. You fork, you merge, you maintain, you deprecate. You keep an archive of patterns that worked in some context, ready to be reactivated or reworked elsewhere. You resist the urge to declare any version final.

Graeber and Wengrow’s point was not that ancient societies were wiser or kinder. It was that they were less stuck. They were willing to treat hierarchy as a seasonal costume rather than a destiny. They did not assume that living one way now meant living that way forever. We do not have their exact conditions. We do have their capacity to refuse inevitability.

Reawakening experimentation in governance is not mainly a technical project. It is a cultural one. It means telling ourselves that we are allowed to play with the rules of how we decide and who decides. It means admitting that the worst thing we can do, in the face of cascading crises, is insist that the only mature response is to keep doing the same thing, only more efficiently.

We are not going to invent the perfect system in a single generation, let alone in a single essay. What we can do is normalise the idea that governance is not a finished product. It is a live question. If our ancestors could treat it that way, then so can we. Getting stuck is not a law of nature. It is a habit. And habits can be broken.