Coordination Projects

A nice bunch of ideas :)

We have already discovered enough technology to live in utopias. What humanity lacks to move forward is not just more technology; it is more coordination.

I am often asked about the type of projects I would consider if I had more time to work on coordination. Here is a bunch. It is not a very orderly bunch, nor even the most important bunch of projects, but I've got to start sharing something.

This is going to be longer than my recent posts. (~4000 words)

Practical Phrases Repository

This idea is cheap, self-contained, and captures a convergent problem of coordination: maintaining a curated, up-to-date knowledge base exposed online.

—

I have written about Practical Phrases here. Phrases that let you do more things once you learn them.

Examples of Practical Phrases include "Yes, thank you." (I have met people who have rejected great things because they could not just say "Yes, thank you."), "I do not promise anything, but I will think more about it.", "I am sorry. I did not mean to say X." and "I am not comfortable with that".

Patrick McKenzie (patio11 on Twitter) is an expert at them, and people already asked for a repository of his Practical Phrases.

Practical Phrases are low-hanging fruits. They let people deal with more social situations at very little cost. A repository of such Practical Phrases could be a cheap boon for communication.

Grounding Self-Mastery

We have many standard external ability tests. They can be general, like IQ tests and the SAT, or specific, like the bar exam and the MCAT. They can be intellectual, like the above, or physical, like the CPAT (from firefighters) and the APFT (from the army).

Unfortunately, we do not have standard tests for self-mastery. These could test attributes like attention control, sustained focus, the ability to endure both pain and boredom, the ability to stay calm under a wide array of stimuli, self-discipline, self-awareness, control over one's memory, emotional resilience, and more.

—

This is a big problem. Certainly, measures can be gamed. But without measures, there's nothing to be gamed in the first place. "It's easy to lie with statistics, but it's hard to tell the truth without them."

Hence, very little collective effort is put into helping people master themselves more. And the little effort is un-grounded: no clear measures are improving due to those efforts. This is why I expect uselessness from most self-help literature, Buddhist traditions, and other philosophies. Sure, there is some information in there, but garbage hasn't been weeded out by a ruthless optimisation process. "Perfection is achieved, not when there is nothing more to add, but when there is nothing left to take away."

There are no clear measures of their practitioners' improvement that I can quickly check. I would love to know according to which testable metrics I am worse than a stoic sage, an enlightened Buddhist or a self-help guru.

For instance, if they're so good at focus, I should expect them to be better than most people at Geometry Dash and other rhythm games. Shall we test this?

—

Until preliminary metrics and tests are established, I see little hope for self-mastery as a field where we can progress.

This is important to me, as individuals lacking self-mastery are one of the major obstacles to group coordination.

Here are some tests I would investigate. Either because they are obvious or because I expect I could program them.

• While performing a boring task requiring constant attention, how much visual and auditive perturbation can be added before failure? I would also add tactile perturbations (pokes, tickles, pinches, etc.), but this is more expensive to produce reliably and thus makes for less universal tests.

• How long can one perform a boring task requiring constant and full attention? Or constant and low attention?

• How quickly can one learn to play a new game of a fixed degree of complexity?

• Pain tolerance.

• How quickly can one switch context between many assigned tasks?

• How well can one estimate how long it will take them to perform various tasks?

• While performing some unrelated task, how long can one maintain some pieces of information in their mind?

• How many things can one balance in parallel?

Coordination Research

Coordination is a rich topic with many questions.

Here is a sample.

Given that everyone and every organisation specialises, how can we make effective cross-disciplinary choices? More generally, how do we aggregate knowledge in groups? We all know many different things. How can we even make decisions that span different types of knowledge? What are the best practical procedures to make the best of knowing more? I would love a toolbox of techniques I could reuse in meetings or longer workstreams.

How do we build common knowledge? If we all agree that we want X, that Y is good, or that Z is true, how can we make this common knowledge as cheaply and quickly as possible at scale?

How can we run organisations effectively and quickly bootstrap new ones? Most organisations are neither communists nor capitalists in their structure. Very few are cooperatives, and I know of no organisation internally built around free-market principles.

They are better modelled by firm theory and organisational theory. This is true for both for-profits and non-profits, and this truth ranges from the smallest 10-person company to international conglomerates.

I need a survey of this, or at the very least, one reproducible method that scales for building a new organisation. I specifically want this in the context of spawning as many groups as possible working on this type of project.

How can we build a political programme that credibly satisfies the preferences of the majority, untangling values from the means of achieving them? How can we prevent all forms of irrational conflicts (irrational meaning that both parties would have been better off for not going through it)? Those are central national and international coordination goals. I would love to know if a group is tackling either of them directly.

How can we transcend old moral frameworks and coordinate with people with whom we disagree and are fighting? For worse, we live in a world where we are constantly fighting. We fight for survival, for positional goods, in 0-sum games. We deceive and get deceived. We hurt and get hurt.

For better, we are never in total opposition to each other. This is why we have established laws for wars, for instance. And we should do more of that stuff.

Unfortunately, morality demands blood. Morality is unidimensional and brands people as wholly good, right, treacherous, part of our tribe, an enemy, impure, and so on.

Reality does not work that way: everyone is good in some ways and bad in others. The point is not that no one is 100% good or 100% bad on the one dimension of goodness. The point is that there are thousands of dimensions of goodness that are not reducible to each other, and some people might even be 100% good or bad on some of them!

This is what I mean by "We are never in total conflict with each other." We are never fully opposed in all dimensions. No treatment of coordination is complete without clarifying how we can coordinate while fighting each other.

—

Fundamentally, the main question behind coordination is: how can we make the most of being together?

Unfortunately, coordination is not studied as a unified field. Other fields mention coordination, such as economics, game theory, decision theory, or political philosophy. But it is not studied in its own right. The closest Wikipedia article I found that describes coordination is in French, from management theory.

Coordination is, within an organisation, one of the key management functions that consists of ensuring, for a group of people and tasks, a conjunction of efforts towards a common objective.

This coordination may require a dedicated coordinator whose goal is to modify the environment to maximise interactions between group members. […]

Coordination should be studied as its own field. This project is not about developing the entire field, as this depends mostly on whether others buy those claims. But within the project, we should at least aim for:

• Establishing the main questions of the field, to define its scope. These questions should clarify what is part of studying coordination and what isn't.

• Surveying the answers to these questions from other fields. This should build some common knowledge about what we already know about coordination.

• Building measures of progress on these questions. The measures should feature internal intermediary targets (which should mostly make sense to people studying the field) and external major milestones (new affordances coming from the field, which should make sense to anyone).

• An honest attempt at answering the questions and evaluating ourselves on those measures of progress.

Good Neutral Policies

Pareto Improvements

The concept of a Pareto improvement comes from economics. A Pareto improvement is a deal in which no party is worse off for taking it, and at least one party is better off. Pareto improvements can take multiple forms...

Sharing information. One of the wonders of the information economy is that information is free to duplicate. If I tell someone else a fact, a maths theorem, or the notes of a song, I don't lose that knowledge. As a result, sharing information makes everyone better off by default.

Liberalism. One of the major tenets of liberalism is the focus on Pareto improvements. The idea is that as long as we ensure that interactions between people are not done under coercion and that no interaction between two parties directly hurts a third party, all interactions are Pareto improvements!

"Growing the pie". This is a metaphor for "increasing the amount of available resources" instead of "redistributing the resources" through non-coercive or coercive means. The idea behind that metaphor is that increasing available resources is almost definitionally a Pareto improvement. Everyone keeps what they have (thus, no one is worse off), and more resources are available (thus, at least one party is better off).

Limits of the Pareto Improvement Model

Unfortunately, in groups that are large enough, Pareto improvements are impossible. There are many reasons why that is the case.

Merchants of misery. In large enough groups, merchants of misery appear. Merchants of misery are people who profit from people's suffering, hopelessness and overall misery. Merchants of misery can be nice people who alleviate other's suffering (such as doctors or morticians), plain fucking assholes (such as cult leaders and war criminals), or anywhere in between. Their common denominator is that as people get better, their businesses get worse.

Destructive people. In large enough groups, destructive people appear. They are people who kill, injure, and traumatise others or destroy and damage their properties. They can do so by mistake, as in the unilateralist curse. They can do so because of a lack of coordination, as in the prisoner's dilemma. Or they can just be sadistic psychopaths who genuinely enjoy spreading pain.

Selfish assholes. Selfish assholes are a special breed of destructive people. They are completely fine with having others suffer whenever it furthers their interests. If they are stupid, they might immediately become criminals and go to jail. But if they are smart, they will happily temporarily comply with laws and norms until they accumulate enough power to not have to do so anymore. At this point, if it benefits them, they will happily subvert the laws and weaken them.

Uncontrolled externalities. In economics, externalities are costs and benefits linked to an interaction which are unpriced and usually borne by third parties. The most infamous one is carbon emissions. But externalities can be much more mundane. A loud party will induce negative externalities borne by the neighbours.

—

One of the worst negative externalities is power accruing to destructive people in general and selfish assholes specifically. Because of this, we can not just make all knowledge public, give everyone access to all the fruits of technology (from recreational drugs to weapons), and so on and so forth.

As a result, we must look not only at how wealthy individuals are but also at the overall distribution of wealth. If someone with too much wealth is a selfish asshole, then they will start to unravel the tissue of civilisation itself. The easiest way to deal with this problem is, of course, to ensure that no single individual has too much unilateral power.

—

The list above is not exhaustive, but it is a good starting point.

Good Neutral Policies

I do not have a compact formal definition for "a good neutral policy". Nevertheless, here is my attempt.

A good neutral is a policy that, at first approximation, is a Pareto improvement that factors in the limitations mentioned above. In other words:

• At its core, it is not about shifting money from some people to others or about raising or lowering taxes.

• It is not a revolution, and it is easily reversible.

• It strives to be a Pareto improvement, even though it may be detrimental to destructive people and merchants of misery.

• It does not increase extreme inequality.

• It does not likely have more unintended consequences than most policies.

I would be very interested in a political platform that centres on such policies. It seems crucial to me to exhaust all such policies before moving toward redistributive policies. Before arbitraging conflicts between people, groups or classes, we should aim for what is good for everyone.

—

If I had to build such a platform, I would investigate the following...

Housing policy. Building or renting a home is a cost on existence. We should aim to make that cost as low as possible.

My default approach would be to shift the burden of proof at all layers to the entities and organisations that want to prevent new developments. Ideally, aside from a minority of reserves and heritage sites, upzoning and building should get approved by default.

Energy policy. Energy is upstream of most industries. We should make it as cheap as possible. My mainline would be to invest in nuclear energy and efficient renewable sources.

Local Experiments. Many ambitious policies should not be tried at the full scale of a country. We should get into the habit of trying them at local scales instead. We could experiment with a couple of Cities with a libertarian or a communist emphasis and study what happens. We could campaign for time-limited laws that we need to renew.

Social Media. Without resorting to full-blown state censorship, old media still obeyed many rules, whether it was newspapers, radio stations or TV channels. Social media bypassed all those rules, sucked the revenue out of traditional media, and is now the main source of information for many people. There ought to be some regulation that is better than this status quo.

Judiciary Enforced Public Commitments. This is one I am least confident in its utility, but I'd like to see some investigation. Around me, I see a big breakdown of trust in authorities. Some of it is for good reasons: many authorities can get away with a lot, enough that it makes sense to doubt them. From politicians and police officers to researchers and journalists.

To deal with this, I would be interested in experimenting with a regime where people can submit themselves (or their actions) to higher scrutiny and liability.

For instance, someone could make a public claim under that regime. Then, if the claim gets later falsified, the burden of proof would be on the person making the claim to present documentation establishing that they were making their best guess. If the commitment was broken, they would then suffer judiciary penalties.

Community Software

I wrote about a month ago about the need for a better alternative to Reddit. To give some context, I used free message board software quite often when I was younger. Whenever I'd spin up a new friend group or a game group (clan/tribe/crew/guild/etc.) online, I would create a message board with some ready-made software, install a chat plugin, write down the guidelines and boot up a small community.

These tools took care of a lot! Members and profiles, permissions and groups, collaborative (although not real-time) document editing, short-form real-time chat, and long-form long-lived discussions.

I do not have this anymore. Much less a version built on extensible open-source software that people let you deploy for free. I think this is a non-trivial reason behind the rise of centralised software like Discord, Slack or Reddit: the free alternatives plainly suck.

For the 2024 version of this, I would want some all-in-one software that features…

• Mobile friendliness. At the very least, responsive design or a Progressive Web App. At best, actual mobile apps.

• Channel-based real-time chats. Like IRC, Discord or Slack.

• Threaded long-lived discussions. Like Reddit.

• Real-time document editing. Like Framadoc or Google Docs.

• File-tree-based document management, with search over the indexed documents. Like Google Drive or Notion.

• Todo-list management. A simple version of this would look like Todoist. But a more complete version would let one send to and keep track of requests from other team members, a feature sorely missing in Slack.

• No AI. No bloatware. No extra-shit. The feature set above revolves around collaborative text editing in various shapes. Whether it is real-time chats, long-lived discussions, documents, or to-do lists. That's all it should be! There can be rendering of the text as images when URLs are included (have you heard of BBCode?), but the core thing should be text.

I would have paid solid money for a managed solution for my company. If someone was building a prize pool, I would contribute some money. It would be even better if I could deploy it myself for the internet communities I've been meaning to run over the last few years.

This has been a strong bottleneck to both my non-work and work-related efforts. I expect I am not the only one.

Humanity Priorities

The Universal Declaration of Human Rights is a beautiful document but a bit old. It dates back to a world where few countries had vast infrastructure or went through their demographic transition. It predates the Internet and most of what we would consider modern technologies.

It is also very deontological, outlining many positive and negative rights but not much in terms of goals.

I believe there should be an attempt at a modern version of this document, synthesising the humanist ideals of the UDHR with our new concerns, factoring in both rights-oriented and goals-oriented components, and surveying people from all over the world directly from their smartphones.

This is a massive undertaking. I would just start by shitting a 2-pager, opening a worldwide forum for whoever is interested in discussing it, and possibly explicitly targeting people from different countries, spiritualities and political affiliations if I have some ins or extra time.

No BS Methodology

I follow a no-BS methodology whenever I need to solve problems for which no canonical methodology exists.

It is not my methodology: I have seen chunks of it in many places. For instance, it prominently features the following...

Hamming Questions. The two Hamming Questions are "What are the most important problems in your field?" and "Why aren't you working on them?".

While the Hamming Questions were asked in a research context, in my experience, they are awfully relevant to generic problem-solving. As in "What are the most immediate steps to take to solve your problem?" and "Why aren't you taking them?".

Too often, when faced with a problem, people do not tackle the most immediate steps needed to solve it, and the reasons why are not great.

Getting Things Done (GTD). GTD is a great book that drills the idea of writing things down instead of ruminating about them. Its core idea is that once something is written and part of a process that will process it, one does not need to consider it anymore and might as well consider it done or obsolete.

I strongly advocate writing things down and taking notes. Any serious thinking, planning, reasoning or reflection must involve notes. Otherwise, they are only free fleeting mental words.

Minimum Viable Product (MVP), Minimum Viable Programs and the Unix Philosophy. The main idea behind the MVP is that there is a pared-down version of any project that can solve the most basic version of its goal.

In the context of software ideas, the MVP can even be a landing page describing the product with a "Subscribe for more information" button. Just the minimal implementation of an idea.

The MVP should be as simple, cheap and easy to test as possible. One of the main ways I see great ideas and projects die is on the altar of extraneous complexity.

Beyond the MVP, thinking of an idea's "degradations" is a core skill. This involves creating many intermediary milestones, from the MVP to the full-blown idea. Ideally, the milestones can be refined in short steps that fit in a to-do list.

Asking others clear questions. This is a synthesis between two very different skills.

The first one is not shying away from asking questions. This means being ready to integrate other people into one's thinking processes. Too often, people think solving problems should be done alone, even in work teams.

The second skill is clarifying questions so that they are easy to reply to. This process of clarifying questions to a future or imaginary interlocutor has the added benefit of emulating rubber duck debugging. It is similar to what happens when an essayist writes down their thoughts. As Nicolas Boileau wrote: "What is clearly conceived is clearly expressed, and the words to say it come easily". (I swear it sounds better in French.)

—

I often summarise this methodology as "Don't be smart and don't be stupid" in a spirit similar to Keep It Simple Stupid. I have found that most projects die not for lack of smartness but for not avoiding simple, preventable mistakes. And trying to be smart is itself a preventable mistake.

I would be interested in applying this methodology to building the research field of coordination or to the existing fields of moral philosophy and complexity theory. Outside of research, I would happily demonstrate it through any other projects mentioned here.

Trusted Public Third Parties

I love Wikipedia, Metaculus and Prediction Markets. The reason why is that they are third parties that I can roughly trust. It's not that I trust them to be correct or even well-calibrated. It is that I trust that there is a process that has been followed when I see something that comes from them.

I have the same fondness for due process in court. I believe such institutions are key to civilisation and want to build more of them.

Here are a couple of ideas for similar institutions.

Private courts. For contracts that would otherwise be non-enforceable in a regular court of law. It could enforce its contracts by taking some stake in escrow from the participants and burning it in case of issues.

Due process for journalism. I would love a newspaper that defines and publishes its methodology for producing its articles. I would love to read articles which expose the research that went behind, such as explicitly flagging what's from other newspapers, the exclusives, all the requests for interviews and the like.

Debate organisers. The previous ideas are costly and require investigation before working on them. If I wanted to start right in on trusted public third parties, I would work on organising debates. As much good as I think podcasts do, I believe that direct debates between people who strongly disagree are much more valuable.

This requires good, impartial moderators who can be trusted to uphold ground rules and ensure the debate stays interesting instead of veering off toward some hyper-contextual third-order counter-counter-counter-argument. I would start with people who are not too far apart in their thinking and, over time, extend it to people who are more vehemently opposed.

Conclusion

People have been asking me to write such a list for a while, so I put down what I had in mind these last few days. I might write another list like this with new ideas in the future.