Balance of power

2025 Dec 30 See all posts


Balance of power

Special thanks to Gabriel Alfour, Audrey Tang and Ahmed Gatnash for feedback and review.

Many of us are afraid of Big Business. We like the products and services that companies provide, but dislike trillion-dollar monopolistic walled gardens, video games that turn into quasi-gambling, and companies manipulating entire governments for profit.

Many of us are afraid of Big Government. We like police and courts, public order, and various services, but we dislike governments arbitrarily picking winners and losers, restricting what people can say or read or think, violating human rights or starting wars.

Finally, many of us are afraid of the third corner of the triangle: Big Mob. We like independent civil society, charities and Wikipedia, but we dislike mobs lynching people, cancel culture, and things like the French Revolution or the Taiping Rebellion.

Basically, we like progress - whether in technology, economy or culture - but we fear the three historically most powerful generators of such progress.

A common response to this conundrum is the idea of balance of power. If we need powerful forces in society, then they should be balanced - either each force balanced within itself (eg. competition between companies), or between each other, or ideally both.

Historically, much of this balance would come automatically: there are natural diseconomies of scale that arise as a result of distance, or the need to coordinate very large numbers of people to do global-scale tasks. In this century, however, this is an assumption that no longer holds true: all of the above forces are getting much stronger at the same time, and can no longer avoid frequently interacting.

In this post, I will expand on this theme, and suggest some strategies for preserving this increasingly delicate feature of the world going forward.


In a previous post, I described this emerging world, where "Big X is here to stay for all X", as "the dense jungle".


How we fear big government

There is a good reason to fear government: a government has a lot of guns, and it can use those guns to hurt you. A government has the power to ruin you that far exceeds anything Mark Zuckerberg or cryptocurrency sellers could do even if they wanted to. For this reason, centuries of liberal political theory have focused on the problem of "taming the leviathan" - getting the benefits of government providing law and order, without the downsides of a king being able to do whatever he wants to his subjects.

Much of that theory can be summarized in one sentence: the government should act like a game, not like a player. That is: as much as possible, the government should be a reliable playing field that productively resolves disputes between people on its territory, and not an agent that actively pushes its own goals.

There are different versions of this ideal:

Even outside of governments traditionally considered "liberal", a similar idea applies. There has been a recent finding that, among governments classified as authoritarian, the ones that consistently deliver less economic growth are the ones that are "personalistic", as compared to those that are "institutionalized".

It is not always possible to avoid the government having characteristics of a player, particularly because of the possibility of external conflict: if a player goes to war against a game, the player wins. But even when the possibility of the government becoming a player is needed, it is often tightly controlled: see the Roman custom of electing a dictator, who would have great power for a short term, but then return to normal once the emergency ends.

How we fear big business

One succinct way to split up criticism of corporations is as follows:

The first issue (corporations being evil) happens because corporations are just very good optimization machines, and as they get more powerful (both in capability and in size), their goal of maximizing profit diverges more and more from the goal of their users and wider society. You can often see this pattern in industries that start out organic and hobbyist, but then become more and more profit-oriented - and at the same time in conflict with their users' interests - over time. Like, say:


Left: percent of coins directly allocated to insiders in newly launched cryptocurrencies, ~2009-21. Right: THC concentration in marijuana, ~1970-2020.


You also see the same pattern in video games: a space that originally focused on fun and fulfillment now increasingly focuses on built-in slot machine mechanisms to maximally extract money from players. Even major prediction markets have started to show a worrying tendency to focus not on pro-social goals like making better news media or improving governance, but on sports betting.

Those examples come more from increases in capability, coupled with competitive pressure. There is a different set of examples that come from increases in size. In general, as a corporation gets larger, it gains more ability to benefit from bending its surrounding environment (including economy, politics and culture) to its will. A company that is 10x larger will benefit 10x more from bending its environment to a particular degree, and so it will perform such actions at all more often than a smaller company - and when it does, it will do so with 10x the resources.

Mathematically, this is the same argument as why monopolists price above marginal cost and increase their profit at the expense of societal deadweight loss: in that case, the "environment" is the market price, and monopolists are "bending the environment" by restricting the quantity they sell. How much bending you can do is proportional to your market share. But expressed in these more general terms, it's a powerful argument that applies in a wide variety of situations (eg. corporate lobbying, De Beers style cultural manipulation campaigns...).

The second issue (corporations being lame) involves corporations being boring and sterile and risk-averse, and creating outcomes that are homogeneous on very large scales - both within individual corporations and between them.


Architectural monoculture is one archetypal form of corporate lameness


The word "soulless" is interesting because it has a connotation that is halfway between "evil" and "lame". "Soulless" feels like a natural way to describe corporations addicting people for clicks or creating cartels to raise prices or polluting rivers, but it also feels like a natural way to describe corporations making every city in the world look exactly the same, or making ten Hollywood movies that are exact replicas of each other, or...

I would argue that these two types of soullessness both come from two factors: commonality of motive, and commonality of agency. Corporations are all highly motivated by the profit motive, and many powerful actors with the same strong motivation will inevitably pull in the same direction, without strong countervailing forces to pull in other directions.

Commonality of agency comes from a company being large, which gives it added incentive to shape its environment. One $1 billion company will do much more environment-shaping than a hundred $10 million companies. It also creates more homogeneity: Starbucks adds far more to any "feeling of urban homogeneity" than a hundred of its 100x-smaller competitors put together.

Investors can amplify both dynamics. While a (non-sociopathic) startup founder would be happier if their company grows to $1b and helps the world than if it grows to $50b and breaks it ($49b of yachts and planes are NOT worth the world hating you), investors are much further removed from non-financial consequences of their decisions. As the market becomes competitive, investors willing to seek out the $50b get higher returns, and the ones who are content to only take $1b get lower (if not negative) returns and have a hard time attracting capital. Additionally, investors who have shares in many portfolio companies can often passively encourage those companies to operate at least partially like a merged collective super-agent. An important limiting factor to both dynamics is the limit to investors' ability to surveil what is going on inside their portfolio companies and "hold people accountable".

Meanwhile, market competition addresses commonality of agency, but it only addresses commonality of motive to the extent that the different competitors have different, not-just-profit-seeking, motives. Often, they do: companies frequently sacrifice on profits in the name of releasing innovations openly to the public, or upholding deeply held values, or aesthetics. But this is not guaranteed.

If commonality of motive and commonality of agency create soullessness, then what is "soul"? I would argue that the definition of "soul" here is nothing other than pluralism: in this context, it is the set of things in corporations that is not homogeneous between them.

How we fear Big Mob

When people talk positively about "civil society" - the part of society that is not profit-motivated and is not government - they always talk about it as being made up of a huge number of independent institutions that are all doing different things. When I ask AI to explain "civil society" to me, it gives these kinds of examples:



When people talk negatively about populism, they tend to envision the opposite: a single charismatic leader who is able to arouse millions of people to directly listen to them and join in a giant mob all pursuing one single goal. Populism is about the "common people", but more specifically it is about a fiction that the common people are united - and often, united in support of a leader and in opposition to a hated outgroup.

When people do criticize civil society, the argument is invariably that it's failing at its mission to be "a huge number of independent institutions that are all doing different things", and is instead driving some emergent common agenda, eg. "The Cathedral".

Balance between forces

In all of the cases above, we talked about balance of power within each of the three big "forces". But we can also have balance of power between forces. A major example is the balance of power between government and business.

Capitalist democracy can be described as a theory of balance of power between Big Government and Big Business: entrepreneurs both get legal tools to challenge aggressive government action, and get a concentration of capital with which they can act independently, but at the same time governments can regulate corporations.

Palladium-ism lionizes billionaires, but specifically when they go off and do unconventional crazy things in pursuit of their own detailed visions, rather than directly seeking profit (see eg. [1] [2] [3]). In this way one can view Palladium-ism as an attempt to thread the needle and get the best parts of capitalism without the worst.


Although both were crucial in setting the conditions to enable it, ultimately Starship was created neither by profit motive nor by government mandate.


My own views toward philanthropy are similar to Palladium-ism in some ways. I have said many things that stridently support billionaire philanthropy, and I want more of it. But the type of it that I want is the type that counterbalances other forces in society. Markets are often not willing to fund public goods, governments are often not willing to fund things that are not yet elite consensus, or things whose beneficiaries are not concentrated in any single countries. Some things are in both categories at the same time, and so get passed over by both. Wealthy individuals can fill the gap.

But there is a way in which billionaire philanthropy can become harmful: when it stops being a counterbalance to government, and instead takes over the government. This has happened in Silicon Valley in the last few years: powerful tech CEOs and VCs have become much less libertarian and supportive of "exit", and more engaged in directly pushing the government to align with their preferred ends - and in exchange, making the world's most powerful governments even more powerful.


I much prefer the thing on the left (2013) to the thing on the right (2025), because the thing on the left is an expression of balance of power, whereas the thing on the right represents two extremely powerful factions who should be balancing each other instead merging together.


You can also have balance of power between the other two forces in the triangle. The Enlightenment-era idea of the Fourth Estate was precisely about civil society as a check on government power (meanwhile, even in the absence of any censorship, lots of power goes in the other direction: the government funds schools and universities and can do a lot to shape the content of especially the former). The media reports on business, meanwhile successful business figures frequently fund the media. These mechanics are all healthy and add robustness to society, as long as one direction of flow does not overpower the others.

Balance of power and economies of scale

If there is one argument that explains both the rise of America in the 20áµ—Ê° century and the rise of China in the 21st, it's a simple one: economies of scale. This is something that people in both places often bring up as a criticism of places like Europe: there are many small and medium-sized countries with different cultures and languages and institutions, and so it's difficult to create businesses that operate across the continent. Meanwhile, in a large homogeneous country, you can easily scale to hundreds of millions of people.

Economies of scale are a big deal. And at the level of humanity, we want economies of scale because they are by far the most effective way to make progress. But economies of scale are a double edged sword: if I have 2x the resources that you do, I will be able to make more than 2x the progress. Hence, next year, I will have eg. 2.02x the resources that you do. Hence, eventually, the most powerful actor controls everything.


Left: proportional growth. Small differences at the start become small differences at the end. Right: growth with economies of scale. Small differences at the start become very large differences over time.


Historically, there have been two pressures counterbalancing economies of scale, and preventing this kind of effect:

  1. Diseconomies of scale. Large institutions are very inefficient in many ways: internal conflicts of interest, communication costs, costs because of physical distance.
  2. Diffusion. People move between companies and between countries and take their ideas and talents with them. Poorer countries are able to trade with richer countries and get catch-up growth. Industrial espionage happens everywhere. Innovations get reverse-engineered. You can use one social network to advertise another social network.

If the cheetah is ahead of the turtle, the first effect makes the cheetah go slower, and the second effect acts like a rubber hand pulling the turtle forward closer to the cheetah. But recently, a few key forces are affecting this balance:

Basically, economies of scale are going up, and while diffusion of ideas is probably higher than before due to internet communication, diffusion of control is lower than before.

The conundrum: how do we have a flourishing civilization in the 21ˢᵗ century, with rapid progress, without extreme concentration of power?

The solution: mandate more diffusion.

What does "mandate more diffusion" mean? First, a few government policy examples:

We can also come up with other ideas in this direction: for example, we could imagine governments making mechanisms inspired by the EU carbon border adjustment mechanism, but charging the tax on (domestic or foreign) products in proportion to some measure of how proprietary a product is: if you share the technology with us, including by open-sourcing it, the tax drops to zero. Another idea that should be brought back is Harberger taxes on intellectual property.

But there is also a more "chaotic" strategy that we should use much more: adversarial interoperability.

As Cory Doctorow explains:

[Adversarial interoperability is] when you create a new product or service that plugs into the existing ones without the permission of the companies that make them. Think of third-party printer ink, alternative app stores, or independent repair shops that use compatible parts from rival manufacturers to fix your car or your phone or your tractor.

Basically, interact with technology platforms, social media websites, businesses and countries in ways that let you benefit from the value that they are creating, without their permission.

Some possible examples:

In general, much value capture in web2 is at the level of the user interface, and so if you can make alternative interfaces that are still interoperable with the platform and its other users using the existing interface, then you can remain part of the network, but opt out of its value capture.


Sci-Hub, a tool of mandatory diffusion that has arguably done a lot to improve fairness and open access in science.


A third strategy to increase diffusion is to get back to Glen Weyl and Audrey Tang's ideas on Plurality. They describe these ideas as being about enabling "cooperation across difference" - ways to have better discussions, and collaboration, between people who disagree or have different goals, and gain the efficiencies of being part of a larger group, without having the downsides of being a larger group that is then a single goal-directed agent. Ideas like this can enable open-source communities, collections of countries, and other groups that are not actors to have higher levels of diffusion between each other, allowing them to share more economies of scale and remain competitive with more internally organized centralized behemoths.

Note that this is all structurally similar to Piketty's r > g concept and his desire to solve it with a global wealth tax (and on the other end, stronger public services). The key difference is that instead of focusing on "wealth", we are going one step upstream and focusing on the generators of unbounded wealth concentration - diffusing not dollars, but the means of production.

I would argue that this approach is better because it more directly targets the dangerous thing (extreme growth coupled with exclusion), and if done well it can even be efficiency-increasing. It also has the advantage that it does not limit itself to targeting one type of power. While a global wealth tax may prevent concentration of power among billionaires, it would do nothing against powerful dictatorial governments or other transnational entities, and it would perhaps leave us even more defenseless against them. A global decentralized strategy of forcing technological diffusion - telling people "either you grow with us, and share access to your secret sauce and your network on a reasonable schedule, or you grow entirely alone, and we shut you out" - would address power concentration in a different way.

D/acc: making the world safe for multipolarity

A theoretical risk of pluralism is the vulnerable world hypothesis: the possibility that we live in a world where as technology advances, a growing number of actors will be able to cause catastrophic harm to everyone. The less coordinated the world is, the more likely it is that one of them will end up wanting to do this. The only solution, according to some, is to concentrate power more. This post argued for concentrating power less.

d/acc ([1] [2]) is a complementary strategy that makes concentrating power less safer. It involves building defensive technology that keeps pace with offense, in a way that is open and available to everyone, reducing the need to concentrate out of fear over security.


The cube of d/acc technologies


A pluralist morality

Slave morality says: you are not allowed to be powerful.

Master morality says: you are commanded to be powerful.

A synthesis morality focused on balance of power might say: you are not allowed to be hegemonic, but you are encouraged to be impactful, and to empower others.

This is another way of phrasing the "power to" vs "power over" distinction, which has existed for centuries.

One way to have power to without power over is to have high diffusion toward the outside world. Another way to have power to without power over is to build something which minimizes its ability to be used as a lever of power.

In Ethereum, one good example of this is the decentralized staking pool Lido. Lido has about 24% of the total ETH staked supply today, but people are much less scared of it than of they would be of almost anything else having 24% of the stake. This is because Lido is not a single actor: it is an internally decentralized DAO with several dozen operators, and a "dual governance" design that gives staked ETH holders the ability to veto decisions. Lido deserves credit for putting significant effort in this direction. At the same time, of course, the Ethereum community has been insistent that even with these safeguards, Lido should not control all of Ethereum's stake - and today it's very far from that.

More projects should explicitly think about not just a "business model" - how they bring in resources to support the work that they are doing - but also a "decentralization model" - how they avoid concentrating power in themselves, and the risks associated with having such power. There are some cases where decentralization is easy: relatively few mind the dominance of the English language, or much less that of open protocols like TCP, IP and HTTP. In other cases, decentralization is hard because the use case demands intentionality and ability to act in some situations. Figuring out how to get the upsides of flexibility without the downsides will be an important challenge for a long time to come.