How to fix the world in 10,000 not-so-easy steps

Our problems are just coordination problems, which means it's possible to solve them.

published: February 11, 2024 - last updated: May 13, 2024

Humanity faces many serious problems, many of them global, most of them getting ever more urgent. (opens new window) To me it seems clear: basically all these problems are, at some level of abstraction, just coordination failures (opens new window).

A coordination failure is a situation where people could work together to make something massively beneficial happen (opens new window), but for whatever reason aren't able to effectively make and execute plans. This is frustrating, because when coordination works it's the most powerful thing we can do (opens new window).

Coordination systems are the ultimate force-multiplier, and the main explanation for human success. This is why I have hope for our future: it's possible for us to drastically improve our coordination systems, and thereby enjoy improvements in every other aspect of human flourishing. It's coordination problems all the way down. Specifically I'm convinced democracy, because it theoretically considers the preferences of everyone, is likely to be the uniquely optimal way for people to work together.

Your immediate reaction might be skeptical: "sometimes people just too fundamentally disagree, and can't coexist"; "democracy just doesn't work". I don't blame you for that reaction, but I don't think that position is defensible. The basic concept of democracy isn't what's failing, but rather the absolutely awful implementations (opens new window) that dominate our conception of democracy.

It seems clear we could cooperate for mutual benefit much more often, because we're currently leaving a ton of easy improvements on the table. Even extremely obvious and simple systems like Approval Voting (opens new window) that work so well both in mathematical theory and experimental trials (opens new window) are rarely used. If we upgraded the tools we use to make cooperative decisions, especially if we were able to do so "optimally", I'm convinced many other problems would be solved shockingly quickly.

This essay lays out the plan I'm most convinced could massively improve the world, following the thesis that the highest-leverage route is to improve our coordination tools. I've tried my best to make this plan pragmatic and strategic and cautious, but at the same time extremely ambitious.

This is my personal plan, focused on what I'm capable of contributing with the time and resources and skills I have. But I'd be thrilled to have some help! Or to be convinced a different plan is more likely to succeed.

If you think people are just too dumb or irrational for any kind of democracy to function, or that some level of inherent hierarchy is natural and desirable, this essay isn't for you. But if you're committed to democracy as a moral imperative, and want to make it the best version of itself, we're on the same team. We might disagree about strategy, but at least it's possible for us have a productive discussion.

And of course maybe I'm wrong. Maybe these plans are doomed to failure. But I'm going to do my best to stay aware and keep listening and adjusting and iterating as long as I have the ability to.

My theory of change hinges on two hypotheses:

The above linked essays explain why I'm convinced by those hypotheses, so now I must validate them with experiments. I propose three phases intended to push in that direction, each hopefully general enough that even if my hypotheses turn out to be untrue the effort won't have been wasted. Since phases 2 and 3 are inherently more speculative and subject to revision, they're much less detailed than Phase 1.

§ Phase 1: Build Basic Tools

If effective democratic coordination and cooperative enterprises are what I imagine will ultimately bring about a fixed world, then we need to find a way to create those two things.

In this essay (opens new window) I explain why I think cooperatives aren't already dominant in the economy: ineffective tools for democratic coordination; and a lack of startup funding for those structures.

So we need to build tools for effective democratic coordination and cooperative startup funding!

This is a good enough time to share that I don't think blockchains are necessary or even very helpful for solving social problems (opens new window), at least given current blockchain paradigms. I'm also planning another essay discussing blockchains and governance specifically.

§ Effective democratic coordination

When I say a tool of democratic coordination is effective, I mean that when used by a group of people it will reliably solicit, select, and execute plans most likely to be maximally mutually beneficial for all of them.

(There are many ways of measuring "mutually beneficial for all of them", but broadly I think the idea of Pareto efficiency (opens new window) is the best we can do, given a requirement of some kind of envy freedom (opens new window). I'm planning another essay about these more detailed philosophical/mathematical questions.)

All three of those steps are important:

  • Solicit: Informally surface information about people's preferences, needs, problems, understanding of the world, and possible group plans.
  • Select: Formally choose which group plans to execute.
  • Execute: Actually do it. Here the three steps reoccur fractally for all sub-problems.

My concrete ideas are discussed either in other essays or an appendix of this essay. But here are the brief versions:

§ Cooperative startup funding

I have two separate ideas: one to better fund startup costs for normal "enterprise" cooperatives that provide something excludable (opens new window); and another to fund anti-rival goods (opens new window) such as open source software or social networks.

§ Phase 2: Build Durable and Nimble Institutions

Assuming effective tools for democratic coordination and cooperative startup funding have been built and seem to be working well, we can begin using them to build truly transformative institutions. The hope is to demonstrate at modest scale that effective democratic coordination can create responsive institutions that can compete with non-democratic ones.

Here are some vague ideas for the kinds of institutions I'd like to see:

§ Phase 3: Experiment and Expand

In general I'm dissatisfied with the level of ambition of existing cooperative enterprises. I believe that scale is now an unavoidable part of our society, and that cooperative enterprises should seek to expand and push at the boundaries of for-profit companies, especially abusive ones.

I also believe there are large benefits to aggregation and unification, and I point to the immense power of for-profit conglomerates as proof. I believe cooperative enterprises should seek to gain the benefits of scale and aggregation for their members.

My hypothesis is that Adaptive Democracy will be the key to unlocking that kind of nimble scale without compromising the transparency and safety inherent in cooperative ownership.

I'm convinced that if democratically controlled institutions are working at modest scales, we should continue to test them at larger scales. This applies both to cooperative enterprises, political parties, and (slowly and carefully) government itself.

At larger levels of scale it finally might make sense to experiment with some kind of decentralized ledger technology (opens new window) to make voting and record-keeping very difficult to fake or control. But it's essential we don't fall into the trap of irrational financialization or untethered value capture (opens new window) that existing blockchain communities have fallen into.

§ My Personal Tasks

  • Create reusable assurance threshold mechanisms.
    • Create commons obligation licenses.
      • Write rough draft based on existing licenses.
      • Find collaborators capable of making them legally sound.
    • Write essay detailing threshold open platforms. Since this idea is just a policy of a company we don't need binding forms of it yet 🤷
  • Create govscript.
  • Evangelize adaptive grants concepts.

Thank you for hearing me out! I hope you'll reach out with feedback, ideas, and whatever kind of support you're excited to offer 😊

§ Appendix: govscript

I've already written an essay exploring the idea of Computable Governance Code (opens new window), and argued that building a programming language for governance isn't an arrogant attempt to create an "Everything Equation" capable of making all of society fully programmable (opens new window). I fully realize it's often impossible to distill human rules into perfectly precise and unambiguous logical statements. But for all the rules that are fully computable, such as the definitions of forms or quorum requirements or fee structures etc, we ought to be able to state them precisely, and then allow the rules themselves to be used by computers to automatically administer those rules.

Even voting and the structure of elections can be boiled down to computable rules about when things happen, what records are kept, how those records are aggregated, and who is allowed to see or change what. In situations where some fuzzy rules must be interpreted by humans, the computable rules can specify which humans will be empowered to do so, how they will enter their decisions into the system, and what inputs they must give the system to document their rationale.

The goal of govscript is to make it easy for any group of people to easily create a well-managed institution for themselves, no matter how big or small, and freely reuse open source libraries of governance rules created and validated by others.

govscript is what I'll work on next, since it's the most general and flexible idea I have. With a functioning implementation of govscript it will be easy to implement and experiment with Adaptive Democracy.

I've wanted to build a database for a long time and have many ideas about how to do so in a more flexible and fully automated way. govscript itself will merely be a translation/usability layer on top of that database that makes it possible for non-technical people to more easily understand how to use it in common governance use-cases.

govscript seems like it must be able to:

  • Define data types and functions.
  • Define collections of stored data, including those consisting of just lone values.
  • Specify allowed inputs from users, including all the details of authorization.
  • Specify reoccurring events for different automatic actions (think cron jobs (opens new window)).
  • Keep a fully immutable log of all events over the entire life of the database (opens new window). Privacy constraints such as through zero-knowledge proofs will be eventually important but aren't immediately necessary.

§ Appendix: Adaptive Grants

As already mentioned, I've written an entire essay exploring a more accountable form of crowdfunding. This quote gives the basic idea:

  • For the entire project duration, only one month (or whatever period) of funds are released to the creator at a time. This means the maximum they can run away with is a single period of budget. Creators are as usual expected to give updates on their progress.
  • Every period the backers can cast an adaptive vote weighted by the size of their backing on whether the project should continue.

Why add this increased accountability? To make people more comfortable backing riskier projects by less known creators:

... it's generally well-understood that a crowdfunding campaign is unlikely to be successful if the creators don't already have a large following and a strong reputation. I think this is a problem.

The ability to "build a reputation" doesn't necessarily correlate with the ability to successfully complete an impactful project. There are many deeply skilled, creative, and determined people who don't have the opportunity (or forethought) to do their work in public. Most of the work that most people do at their day jobs is private and counts for nothing in the public square that is the internet.

And even worse, there are many people who are very skilled at building a reputation that definitely don't have the skill to complete an impactful project! It's pretty easy to put together a team of marketers and designers and hype a product that you definitely can't build.

Although I think this more flexible and accountable form of crowdfunding is a good enough idea to try, I'm not sure it makes sense to build a new platform. It isn't clear to me if I should simply evangelize my proposed improvements to existing platforms instead of starting a new one. I'm thinking about this.

Regardless, here are the advantages I would intend to build in a new crowdfunding platform:

  • Cooperative ownership, where employees are given membership voting weights, and project creators are awarded membership voting weights in proportion to the funds taken from their successfully funded projects.
  • The flexible and accountable Adaptive Grants structure.
  • Incentives toward ACH payments for lower fees.
  • Smarter budget calculators that factor in ratios between fixed and per-item costs.
  • Individually selectable rewards, rather than inflexible "tiers".

Structures like this could be used by both enterprise cooperatives to fund startup costs by offering crowdfunding backers heavily discounted memberships (or whatever else) in exchange for their backing; and startup open source projects for some combination of discounted assurance threshold licenses or voting weights or whatever else.

§ Appendix: Assurance threshold mechanisms

Most information goods are anti-rival (opens new window), meaning they get more valuable the more people adopt them. This means it's inherently counterproductive to meaningfully exclude participation to just supporting members, creating a pretty tricky commons sustainability problem. We want creators of open information goods to be able to prosperously support themselves without having to wrap the good in some kind of antagonistically incentivized company (opens new window).

The best solution I can think of is an adaptation of assurance contracts (opens new window) that I call threshold conditional licenses. The basic idea:

  • The information good is released to the public from the beginning, but using some kind of pro-socially restrictive license, such as GPL, AGPL, or creative commons share-alike or non-commercial.
  • Supporting members are given the right to use the project according to a more normal permissive license (the license is "conditional" on membership).
  • If the project rises above a certain threshold of monthly support the project switches to a more permissive license for everyone.
  • If the project falls below that threshold at any time the project switches back to the conditional license.

    Basically: during any period where the threshold isn't satisfied, supporting members are the only people who have permissive license access to the project.

The same general structure could be applied to platforms such as social networks that are intrinsically anti-rival, by making full access to the platform conditional either on supporting membership or the threshold being satisfied. This is a bit messier but seems like it could work.

Also I'm fairly convinced this concept would work best when paired with cooperative ownership and voting rights for supporting members. Otherwise it will be difficult for potential members to trust they won't be "rug-pulled" (opens new window). But of course you're welcome to try whatever you want.

In order to enable people to experiment with this structure there must be reusable versions of these licenses. Since I'm not a lawyer, I can't do that on my own, so I'm merely going to reach out to people who might know what I should do next, or who could collaborate to make this happen.

§ Appendix: Democratic forum for fact and truth seeking

This is one of my least developed ideas, and I haven't ironed it out to my satisfaction.

The hope is to use Adaptive Democracy and better "signal/noise" incentive systems to create a forum where those who help build durable truth and consensus across difference are rewarded with increased reputation, which is used to increase reach (opens new window). Should allow people to share information, fact-check things, and have actually productive discussions that help everyone better understand each other and the world. Essentially hoping to solve the sorting/prioritization/evaluation problem for knowledge/ideas in a values-aligned way.

Here are some vague bullet points:

  • Again, inspired by bridge-building structures (opens new window) such as Polis (opens new window).
  • Information is broadly categorized into three types:
    • "Signal" is defined as information that is both surprising/new (opens new window) and positively consensed (consensus across multiple normally-disagreeing groups). In general those who create signal are given increased reputation.
    • "Noise" is defined as information that is either unsurprising (already digested by the community) or merely not consensed (only affirmed by people in narrow groups). In general noise is not given increased reputation.
    • "Spam/Abuse" is defined as information made in bad faith, by being harmful or merely negatively consensed (consensus of badness across many different normally-disagreeing groups). In general spam/abuse are punished with decreased reach, or possibly exile from the forum.
  • Moderation of decisions such as exile from the platform or spam/abuse are made using Adaptive Democracy, meaning members in good-standing select constitutional documents outlining how moderation works.
  • Reputation is used as "preference weights" on "tags" representing specific judgments/categorizations of information. This means categorization/prioritization is always driven by intentional decisions of humans rather than their unthinking impulses, with the hope of biasing the system away from cheap virality and lurid stimulus, and toward more values-aligned action (system two thinking rather than system one (opens new window)).
  • Reputation can be context-specific, so for example someone who creates signal in the area of Anthropology has increased reputation/reach in that area but not necessarily others.
  • Users can post information of a few different types:
    • Questions seeking technical help or factual knowledge.
    • Prompts for a position on an open-ended question, where answers are given as unified "essays" describing different positions.
    • Announcements, where a user is seeking attention of some kind.
  • Users can slowly build a web of validated knowledge:
    • They can post "facts" to the site, but all facts must have happened to them, not someone else. So rather than saying "it rained on Feb 2nd" one would say "I took this picture of it raining on Feb 2nd". The purpose of this is to make the breadth of all factual claims narrow enough for some realistic process to be able to adjudicate that fact if necessary. It's easier to ask "did this happen to you" than "did this happen".
    • Facts can be challenged and adjudicated by some democratic trial process, the details of which I'm not sure of yet! (opens new window)
  • Inputs are categorized using possibly overlapping tags, which is more flexible and realistic than the Stack Exchange (opens new window) family of sites where questions must be strictly categorized into clear site domains.
  • LLMs would be useful in such a system, but their outputs should only be suggestions to humans, who ultimately make all final decisions.

Want to hear from me in the future?

Usual fine print, I won't spam you or sell your info.
Thank you! Come again.