What do these have in common: Elon’s Twitter, ants, General von Moltke’s strategy, brains, AI, prediction markets, and Wikipedia? The answer shows the path to the future of democracy.
you are totally right , I want to talk about the influence that the social media platforms like facebook and twitter have nowadays and how they can affect the opinions of people. The presidential elections of Tunisia back in 2019 prove it , a former law professor became the president of the country without any heavy campaign like other candidates who had by far more ressources to make a big campaign and also support from certain media outlets but the law professor succeeded because he was supported essentially by youth who made several facebook groups and pages where they talk that we must support him, it was a blow to the political establishment in Tunisia. https://www.bbc.com/news/world-africa-50087240 you can have an idea through this article.
Furthermore , this president Tomas stopped the parliament since July 25 because there are many problems and Tunisia is facing political and economic crises , I cannot go much in detail here , he says since many years that he has a view of governance that you are describing , he wants everyone to participate at the local level in his region in order to contribute to policy making but many are criticising him and his idea and I was among them but this article changed my opinion a little bit , I saw it as difficult to apply in reality but it can be possible in th future.
The president launched in the beginning of this year an online consultation where every citizen can contribute by answering questions and giving his view about the issues facing Tunisia and how we have to tackle them https://www.e-istichara.tn/home this is the website and now we have a referendum in July in order to accept the change of the constitution or not and he said that all the remarks from the platform will be taken into account in the new constitution and the economic reforms program , this is its website https://www.e-istichara.tn/home sorry for being long but your project reminded me of the views of the president that's why I wanted to explain , I'm still a lit bit skeptikal about how it can be applied on the ground but it will happen ..
A soap factory had a problem. They sometimes shipped empty boxes without the bar inside. This challenged their perceived quality with the buyers and distributors. Understanding how important these relationships were, the CEO of the company assembled his top people.
Six months and $8 million later, they had a fantastic solution - on time, on budget, and high quality. Everyone in the project was pleased.
They solved the problem by using a special scale that would sound a bell and flash lights whenever a soap box weighed less than it should. The line would stop, someone would walk over, remove the defective box, and then press another button to re-start the line. As a result of the new package monitoring process, no empty boxes were being shipped out of the factory.
A while later, the CEO decides to look at the first week report. Since the scales were put in place, no empty boxes had been shipped out of the factory. Each day about a dozen defective boxes were being removed, which was consistent with the projections. There were almost zero customer complaints and they were gaining market share. The CEO felt the $8 million was well spent.
However, the number of defective boxes picked up by the scales dropped to zero after three weeks. He filed a bug against it and after some investigation, the engineers came back saying the report was actually correct. The scales really weren't picking up any defects because all boxes that got to that point in the conveyor belt were good.
Puzzled, the CEO traveled down to the factory, viewed the part of the line where the precision scale was installed, and observed just ahead of the new $8 million dollar solution sat a $20 desk fan blowing the empty boxes off the belt and into a bin. He asked the line supervisor what that was about.
"Oh, that," the supervisor replied, "Bert, the kid from maintenance, put it there because he was tired of walking over, removing the box and re-starting the line every time the bell rang."
Jun 8, 2022·edited Jun 8, 2022Liked by Tomas Pueyo
Great article. Love the use of examples from different fields all pointing to the same things. And as usual the use of the visuals which makes it fun to read and more likely it will be read!
James Surowiecki's book Wisdom of Crowds sets out a couple of things that are necessary for better decisions among groups. Diversity, Independence and some mechanism for aggregating the ideas together. Without some of these conditions being met some of the examples you discuss would not work eg a Wikipedia article written only by Democrats or the work people contributed all had to be approved by me.
He also points out democracy is not just a mechanism to come up with the best policy. It is a mechanism for agreeing on how we can get along. In this regard, harmony is just as important (or more important) than the best decision. Free and fair elections where all our opinions are (sort of!) weighted equally leads to more citizen harmony and satisfaction.
Similarly when the German Govt decided to get out of coal-fired power, it didn't want to touch the issue because it would cop it in the neck whatever it decided. So it gave the problem to a commission made up of 28 stakeholders, told them to meet Germany's emissions targets and gave them funding to allocate. After nine months of negotiation they released a plan which was largely agreed on (27 of 28 stakeholders voted for it) despite decades of conflict. No one got everything they wanted. The German Govt got a plan without the grief it would have otherwise got and Germany was more settled. After all if you've a greenie and you've negotiated a plan on equal terms with a power company, unions etc, you've only got yourself to blame if you think you could have done better.
I think the aggregation mechanism is the key. If it's legitimate, it solves the harmony issue, because everybody buys into the way we rule ourselves.
The example you give abut Germany is great (I didn't know!), and it's an example of an aggregation mechanism. My sense is this is strictly better than much legislation today and we should do more of it, even if it's worse than more decentralized mechanisms
Good thinking here as always. I could see local legal-twitter-analogs that develop drafts of laws that legislatures and parliaments can approve or disapprove. That can serve as a check on twitter mobs. If a particular representative is always voting against the crowd sourced drafts, that can be used as a campaign issue next election.
A big roadblock is the “legaleze” language in which laws are written. I think you could either append the crowd-sourced laws with legal definitions or have translators work on the draft laws. Too many people will not engage if drafts are not in everyday language when they are being worked on.
1. Checks and Balances: you're concerned about 2 things: something being approved too fast, and something not reflecting the will of the people. In both cases, you should be able to do better than "all follow a bunch of people campaigning what kind of person they are and the bundle of opinions they have, and then give that person all the power to decide on our behalf what to do". Speed of approval can be algorithmic. Reflecting the will of the people can be asked to people directly.
2. Legalese: computer code that is fuzzy, sometimes purposefully, sometimes not. It's a middle ground between a mechanism and something human-readable. A better solution is to have these 2 layers: a layer of code, and a layer of clear explanation. Note that, in some cases, the fuzziness is a feature, not a bug. We will need a way to account for that.
First, a system that drives quick changes to a body of law is not a great idea. Stability and predictability are features of well-functioning government.
Keeping legislatures in power while the legal-pedias go through beta testing and live roll outs seems prudent. Maybe the Wikipedia and Linux examples worked because there was not much on the line when they were started. Do you think crypto currencies are a good model?
I like your description of legalese as code. The two-layer solution may work well.
I would never start by replacing the system wholesale. It doesn’t need to be. It can first replace local elections in an AB-tested way. As the system grows stronger, it corrects its errors, and ppl get used to it, it can move on to bigger levels of gov.
I hear your point on stability and predictability, but I’m not sure I agree. Of course, companies don’t invest if a country is not stable and they don’t know what will happen with their investment in the future. But that doesn’t mean a country’s legislation doesn’t change. Rather, that its changes are predictable. If the system works well, you should be able to see things coming, and what things are coming should in general increase the quality of the rule of law. What’s better, a shitty NEPA regulation in place for 50 years, or continuous NEPA updates in the margins because the big pbms were solved once and for all a long time ago through a system that works?
Another aspect to consider is that legislatures/ parliaments were set up as a check against the executive. If the legislature loses the power to enact laws then the executive becomes relatively more powerful. And absolute power ....
1. The executive could only execute the laws approved by this mechanism
2. Part of the executive could be also crowdsourced. Eg the DMV would probably be better if it was managed in a decentralized way. A DAO could be employing its workers and pay their salaries, you don't need an agency managing all of it
Jun 10, 2022·edited Jun 10, 2022Liked by Tomas Pueyo
Yeah, I wonder how well this would work in practice. Fundamentally, societies that aren't personalist (where those with arms only owe personal loyalty to a man) work only because enough key people believe in a mythos. For instance, officers of the US military swear an oath to uphold the US constitution, and not to the President or any person. Even more importantly, I believe enough of them take that oath seriously because they believe in the constitution (they believe in that mythos). Some countries may have the same legalese but if the men with arms regard pieces of paper as only pieces of paper, any law or constitution is meaningless. So you would need an DAO that has enough mythos, or rather, enough people with arms have to believe that DAO is endowed with some mythos that they are willing to risk their life for to defend.
Ultimately, force of arms/threat of violence carries the day. No DAO is able to fire a gun.
"...lawmaking will become fast, prolific, and intelligent, and we won’t all be angry all the time at the terrible politicians we’ve had to elect." - I am ever optimistic Tomas!
Tomas, I feel like you are the only writer on the internet laying out a positive vision for the future. I mean that in two senses - you reject normative analysis; and your predicted post-democratic ideology-transcending world government sounds like progress to me!
Do you have a proposed KPI? Aggregate land value? An 'Optimism Index'?
Ah I love that you say that. I had never thought of myself as that, but I'm glad you think that! The future world is great, we just need to steer it in the right direction.
I think there are some good metrics. Indeed, GDP has been a good proxy, but optimizes for time worked and money spent, and doesn't account well for intangibles like health and education. My current sense is that a blended metric like HDI or the like could be better.
Fascinating, as always... One thought about your conclusion on the future of democracy and how laws could be written in the future. Is there a fear it could play out badly for minorities?
I think it would be orders of magnitude better. Right now minorities can be ignored (the rule of 50%+1) or in charge (lobbies). This should solve both issues, the same was as an article on Wikipedia doesn't get approved just because a majority thinks it should be a certain way, or because a rich person wants an article to be a certain way.
The consultation aims to give the opportunity to all citizens, whether they are residents in Tunisia or abroad, to express themselves freely and transparently about the recovery of the democratic transition in Tunisia and this, according to an unconventional mechanism, in a participatory approach allowing a common conception of the future of Tunisia. Ideas can be proposed for the development of new visions and approaches that could allow the management of public affairs in their different aspects.
The consultation also aims to make the Tunisian citizen a real actor in the process of developing a new conception of the fundamental choices linked to the political system in Tunisia.
This consultation will allow the Tunisian people, the sole holder of power, to identify the major political and economic reforms to which they aspire, and will set up a democratic framework for deliberation on various proposals that would help to face the various current challenges Tunisian citizens are confronted to in various fields.
This is what is written in the website of the consultation Tomas , making everyone participate and this is what you explained ..
The challenge is not as much the participation (everybody is participating all the time), but rather the condensation of insights. Usually, 95% is trash, 4.9% is interesting, and 0.1% is gold. I hope the process they suggest will work to unearth and gather the gold.
That's the objective , there is a place in the consultation where you can write freely about the topics asked , the question is how they will manage to gather the best answers and suggestions from thousands of people ..
Very interesting post! I'm interested in the concept of mission command that you discuss in your first point.
Moltke the Elder's version of mission command was very much necessary due to the possibilities and limitations of military technologies you mention. Decentralization was critical when one could communicate without controlling.
Fast forward to 2022. Western militaries have the technology and capability to both communicate rapidly with technology and keep persistent communications throughout operations. In military circles, this has developed into the problem of the "8,000 mile screwdriver". In other words, military commanders have so much tech available that the tech actually enables them to micromanage to their liking (to the detriment of the organization, at times).
The U.S. military in particular loves the concept of mission command. But it struggles to implement true decentralization due to both the bias towards control and the capability of doing so. It's interesting to me because, of all the examples you sketch out, this one is a case of technology impeding decentralization.
Do you see any parallels or lessons to be learned for your broader argument, or is this just a one-off case that is specific to military contexts?
Internet is a decentralization mechanism, but AI is a centralization mechanism. It concentrates all the data in one model, that then can make amazing decisions. Look at China: they're relying on AI to craft the perfect authoritarian state, and I fear they'll succeed.
In the future, we might ask: "Siri, what's the optimal policy for drug regulation in the US?" and it might be able to spit it out.
From what you're saying, this is what the US military is experiencing right now.
Thanks for the response! I'll try to respond clearly. Apologies if these thoughts seem incoherent.
I agree that AI is a centralization mechanism that complements the Internet's decentralizing tendencies. However, this looks different for how U.S. military operations tend to be run. There isn't an AI that makes optimal decisions for military organizations. Instead, that decision making is vested with a single individual, the commander.
The concept of mission command decentralizes the execution of military operations and allows decision making at lower echelons while retaining the originator's (the commander's) overall decision to do X.
Mission command is great at decentralizing execution (tell them what you expect and get out of the way if they are doing it right). Except that now, commanders have the ability to manage military activities at a distance. A general can communicate directly to a sergeant from thousands of miles away and tell that sergeant specific instructions.
The U.S. military struggles to allow the decentralization mission command calls for because single individuals - commanders - have the ability to interfere at any point of an operation. Technology enables this micromanaging. Only the self-discipline of individuals to -not- use the technology to micromanage enables mission command to work.
Going back to your first point, would mission command work better if there was an objective and optimized decision maker such as an AI? Or at least a commander that was assisted by AI in their decision making?
This may be more of a cautionary tale - if the human element isn't addressed, the benefits of network effects and decentralized systems aren't going to materialize.
These are great examples to dissect what works and why. Let's see.
It sounds like the key question in decision-making is the ability to aggregate data. Data is harder to aggregate when
1. There's a lot of data
2. The data is hard to communicate
3. Centralized data is hard to digest
If we take wikipedia, the amount of data that we're talking about is massive. So much so, that it's hard to digest. So it makes sense to decentralize it.
In law-making, there's a lot of data (all the pbms that all citizens go through, plus all the universe of potential solutions). The data is hard to communicate (you can send it to the gov, but then how does it know where to go?). And digesting it is hard (could lawmakers get all the insight from all the relevant ppl, and then make decisions alone? Hardly).
What AI does, however, is take a ton of data, centralize it, and digest it in a way that it can then automatically take the next data point and spit out an answer. That's why it's a centralization mechanism.
Now for the military: mission command assumes (1) there's too much data, (2) it's hard to communicate it, and / or (3) it's hard to digest.
I think we all agree that (1) is true in the battle in general.
In the case of von Moltke, (2) was true (telegraph back and forth second to second was not viable), and (3) too (how would you make sense of all these telegraphs?).
For the modern US army, (2) is probably not a problem (satellite imagery, sensors, data feeds...)
So if (3) is easy, even if there's a lot of data, it can be properly digested, and then it makes sense to concentrates all the decision-making in one place.
This would require, however:
a. Visualizations that allow a great mapping of the situation in real time (input)
b. The ability to produce orders extremely fast for all the teams that need them (output)
a. I don't know how good the visualizations are, you tell me. My assumptions is that in Sci Fi they can be amazing, but not sure about today's military.
b. I don't know how much time the command center has for analysis, how fast they can craft responses based on the situation, and how many of these decisions they need to make per minute.
If viz are good and the # of decisions they need to make is small, then it makes sense to do command & control. That's basically what Napoleon did.
If, however, viz are not good, or there's too many decisions that need to be made per minute, then it makes more sense to decentralize.
Maybe what's happening is that viz are good enough that *high-level decisions* can be made quickly? And so those move back to the command center as command & control? While viz is not good enough to handle every lower-level decision, so those remain as mission command?
This analysis is very intriguing. I like your three main points of reference re: data aggregation (Is there a lot of data? Can that data be communicated easily? And can it be aggregated easily?). I'm keeping this one in my pocket for future use.
The answer to your question on visualizations in military command centers is: maybe. It's all dependent on the specific unit and the echelon. For example, in the Pentagon working for the Joint Chiefs of Staff, you have access to extremely cutting edge systems, data analytics, professional data experts, and individuals at the peak of their military careers. A smaller unit in the special operations community may not have the vast resources, but that team may have the latest systems on hand that can help them receive and understand orders. And they have the expertise and seniority to make their own visualizations rapidly.
The mismatch here is with a standard company, platoon, or squad. These units don't generally have the resourcing to have great visualization tools on hand. They are nominally digital, but these networks are threatened and generally have limited bandwidth. And they are always on the move. (I would still use mission command here, as they have the expertise to operate with minimal interference, but that's another thread).
Perhaps the mismatch is when a higher level unit uses a C2 philosophy vs. a mission command philosophy without considering the capabilities of all units involved. Viz is good for the high-level unit, but not good enough at the lower unit, but the headquarters is still doing C2.
I'd also say that, in conflict, the # of decisions that need to be made can quickly overwhelm a single individual. Good commanders sort of recognize which high level decisions they must make. But in the total amount of decisions that must be made at any given moment, those decisions are probably <1% of the total amount of decisions. The other 99% could be critical, and can be directly influenced, but can be dealt with at a lower level.
As always, you’ve provided info that is new to me and makes me think. This is insight into a paradigm shift that is coming. I always greatly appreciate the heads up. Thank you, Tomas.
I remember as an MBA in 1986 reading in our case studies about the Japanese concept of Kaizan, "gradually improving productivity by involving all employees and by making the work environment more efficient." That was revolutionary then and what you outline here is gonna be cool to see.
Not sure whether to be eagerly hopeful or incredibly terrified by the prospect of decentralized autonomous organizations making the rule of law on the fly (relatively). Interesting times ahead!
I live in the museum otherwise known as England. It is a physical museum, but of more relevance to this article is the fact that it is a social museum and a museum of thought as well. Its use of DAOs was the foundation of the British Empire.
My very brief take on England’s Geohistory would be something on the lines of:
One of the last areas in Europe to get Mesopotomian/Greek/Roman knowledge so the local tribes were vulnerable to invaders with better technology. Somewhat protected by sea, but also at the mercy of other sea powers because of that. This creates a massive incentive to develop naval power.
Flat, fertile lands in England are an asset that everyone wants, so conflict drives development of both military technology and social evolution. Communities are forced to band together for protection so the “nation” expands to the limits of geography: the Welsh and Scottish borders.
England’s subsequent success depended on the unifying mythos of nationalism and loyalty to the monarch. Not necessarily to a particular monarch, but to the idea of the monarchy. This was greatly helped by the fact that from the time of the Magna Carta onwards English monarchs were gradually forced to share more power with the nobles and then parliament. A society with a strong sense of hierarchy and obedience developed and a microcosm of that society inhabited the DAOs otherwise known as ships.
Before the development of radio there could be a massive gap in both space and time between the commander-in-chief and the action required. Orders had to be in the form of “do x by whatever means you see fit” and individual captains had to exercise their judgement in how best to achieve those aims. I am a big fan of Patrick O’Brian and his meticulously researched (but fictional) naval series about the British navy in the time of Napoleon. Part of the secret of British success was the naval education system both on land and at sea that prepared officers for obedience, but also nurtured a degree of independence.
As you have explained to some who question the relevance of Geohistory, the history of a nation can explain a large part of its psyche today. Britain still has strong elements of the very hierarchical and obedient society the enabled the empire to flourish. It worked in part because of the concept of noblesse oblige. The populace gave their loyalty and obedience and in return the nobility had some sense of honour and duty. The problem with success can be persistent belief in the things which produced that success (or were perceived to produce that success) and a reluctance to change as the world changes.
I had written the above several weeks ago, but got stuck at 75% done trying to make sense of the British attitude today. Despite their former success using DAOs, recent Britain has a strong bias towards command and control and is often regarded as a nanny state. The leaders of both sides of politics almost universally attended Oxbridge. Medical practice in England is largely dictated by centralised bureaucracy producing guidelines (which get interpreted as rules) and formularies to dictate prescribing practices. There is little trust in the knowledge and experience of individual doctors.
To get somewhere near a postable 90% I’ll just go back to the usual framework for a useful model of reality: there is a balance between command and control and the DAO concept. As your discussions above highlight, certain circumstances favour one over the other and I think you have the parameters right. Constant Conflict Theory predicts that whichever works better will eventually be widely adopted. Those who seek power and wish to impose their ideas on others will obviously favour command and control. Those who seek optimisation of results will favour listening, trusting others enough to cede some control, and allowing people and systems to develop and grow stronger.
Very interesting. And you have many similar themes. Thx for sharing!
On the DAO, that’s interesting. I’ll have to think about it.
Hypothesizing: maybe the UK was decentralized vs Europe, but much less than say the US. And centuries of success reinforced the centrality of London for power. I don’t know!
Having thought about it I would go with the ant analogy. The British ships and army were to some extent clone-like and sent out into the world to bring back useful stuff to the mother country. Strict rules governed each unit and every man had a specific role. There was strong loyalty to King and country and alignment of incentive in that all had to work together to survive, but also ships could win prize money which was shared out to the crew.
In the section Peer Review you describe releasing an article that is 90% correct and then depending on crowd sourcing to advance the 90% to 99%. This may be appropriate if you want to produce an evolving system that values only future goals. Many systems, medicine most prominently, value both current and future targets. You can not morally release a therapy that is 90% effective when you know 99% is achievable albeit at great cost. To do so makes you a monster like Elizabeth Holmes. Many other (most?) systems besides medicine may resist sacrificing current day performance for future advances. Democracy may be one of them.
Jun 10, 2022·edited Jun 10, 2022Liked by Tomas Pueyo
There is a puzzling inconsistency in the story: you wrote, "Decentralization was enabled by faster communications"; but in fact faster communications make decentralization less necessary. Ancient Sumeria, ancient Greece, and the Roman Empire had much less centralized decision-making than the US today, because communication across them took weeks.
It would be more-accurate to say that faster communications increases the geographic scale that any system of government can achieve. So today's communications could enable the true democracy of Athens, supposing that were desirable, on the scale of a modern nation. But it also means you can create larger centralized totalitarian governments than ever before, as we saw in the 20th century.
I've been thinking about your comment for a few hours already. You hit on something important and quite fundamental. So let me think out loud, tell me what you think.
First, I think I mix two concepts in the article: subsidiarity (which is hierarchical) and decentralization (not hierarchical).
In hierarchical decision-making, high-level decisions are made at the top, low-level decisions at the bottom. The principle of subsidiarity simply says that as many decisions should be pushed down as possible. In what circumstances is this valid?
1. If low-level nodes don't need to coordinate with nodes from another part of the network.
2. If there's a lot of information on the ground
3. If, when information is transmitted up, there is a loss of fidelity
4. If it's hard to aggregate the information from many different sources.
Both Napoleon and von Moltke were hierarchical, but in von Moltke's time, subsidiarity worked better. Why?
1. Low-level nodes don't need to coordinate with nodes in other parts of the network. They need to fight with their own unit. They do, however, need to coordinate at the battalions / armies level. Which was the big drive for Napoleon to coordinate everything, and one of the reasons why von Moltke's armies sometimes suffered (they counted on other forces to support them). So I'd say this pushed more for authoritarian hierarchy than subsidiarity
2. Lots of info on the ground: this was certainly true. With the trains and telegraphs, many more teams could act across a much vaster expanse. Battles included 100ks of ppl, not thousands.
3. Loss of fidelity: probably high loss of fidelity when transmitting information, because of the distances. But there was the telegraph, so maybe that counteracted the other trend?
4. Hard to aggregate the information from all the sources: probably true given just the sheer amount of data that had to be transmitted, and the fact that the telegraph was not a high bandwidth technology at the time.
So maybe what happened is that the dramatic increase in information, due to the train and telegraph enabling many more ppl interacting across many more areas, meant that it was just impossible to centralize that data quickly and then to digest it, whereas it was still possible with Napoleon.
Maybe in fact this is one of the reasons why Napoleon was considered such a good military leader: at that point, battles required coordinating vast amounts of teams across different areas, and that was a mess to do, but Napoleon was able to get the data to him quickly, and his brain was able to process it fast enough to give orders?
So that would be for hierarchies: subsidiarity vs. authority (or mission command vs. command and control).
Now the question becomes: when is a decentralized process better than hierarchy (either authority or subsidiarity).
For example, nearly all companies are hierarchical. The few that are decentralized are famous because of it: the holacracy of Valve (maker of Steam), and at some point Zappos. So clearly decentralization is not always better. Why does hierarchy work better than decentralization in companies?
It sounds to me like there are forces that push decision-making to be hierarchical vs. decentralized.
One of them is the importance of alignment. The more a lot of ppl need to be single-mindedly rowing in the same direction, the more this pushes for centralized coordination to make sure one decision is made, quickly, and everybody follows it. Companies need this
Another one is, again, the amount of information at the node level. The more there is, the harder it is to convey all this information to a central node, and the more you want those close to the info to make the decisions.
Another one is the need to coordinate across all nodes. If you can easily divide people into autonomous teams, they don't need to coordinate as much between them, and decentralization is less useful.
Another one, and I have the intuition that this is key, is the ability for any message to find the right recipient. If it's very hard to get all the info to the right ppl, then instead you'll use the hierarchy to aggregate it up and communicate it down. Decentralization works well when (a) the right person can be found, OR (b) the message between nodes can pass through.
(a)
If we take wikipedia, every person might write an article, but then other ppl interested in that article can easily find it. You only see the articles you're most interested and knowledgeable in.
On Twitter, you can find the right messages because you start following the ppl you're interested in.
(b)
Capitalism has some of it: you only pay attention to the industry / goods you're interested in. But it also has another crucial value: the mechanism of prices conveys information across nodes.
If a miner has a harder time finding some ore, the ore he does find will be sold at a higher price, which the furnace owner will have to pay, and then pass through to their own customers, who then pass through the price increase all the way to the end consumer, who might then buy it or not, sending back the message through the entire chain. The information doesn't lose much fidelity at every step.
Neurons do the same: they pass action-potentials from one to the other, without much signal loss.
Twitter does the same: the retweet has no signal loss.
In wikipedia, obviously, there's no signal loss from one person to another: the article is fixed and accessible to all.
So maybe the main handicap of decentralization is the loss of signal when signals are relayed across nodes. You can definitely see this in democracy: people can't communicate with each other efficiently. Same arguments all the time, too emotional, lack of depth, different priorities, lack of clarity of the areas of expertise of each...
If this is true, the key factor for decentralizing democracy would be a lossless way to convey political insight.
you are totally right , I want to talk about the influence that the social media platforms like facebook and twitter have nowadays and how they can affect the opinions of people. The presidential elections of Tunisia back in 2019 prove it , a former law professor became the president of the country without any heavy campaign like other candidates who had by far more ressources to make a big campaign and also support from certain media outlets but the law professor succeeded because he was supported essentially by youth who made several facebook groups and pages where they talk that we must support him, it was a blow to the political establishment in Tunisia. https://www.bbc.com/news/world-africa-50087240 you can have an idea through this article.
Furthermore , this president Tomas stopped the parliament since July 25 because there are many problems and Tunisia is facing political and economic crises , I cannot go much in detail here , he says since many years that he has a view of governance that you are describing , he wants everyone to participate at the local level in his region in order to contribute to policy making but many are criticising him and his idea and I was among them but this article changed my opinion a little bit , I saw it as difficult to apply in reality but it can be possible in th future.
The president launched in the beginning of this year an online consultation where every citizen can contribute by answering questions and giving his view about the issues facing Tunisia and how we have to tackle them https://www.e-istichara.tn/home this is the website and now we have a referendum in July in order to accept the change of the constitution or not and he said that all the remarks from the platform will be taken into account in the new constitution and the economic reforms program , this is its website https://www.e-istichara.tn/home sorry for being long but your project reminded me of the views of the president that's why I wanted to explain , I'm still a lit bit skeptikal about how it can be applied on the ground but it will happen ..
I guess this story is what you are talking about:
A soap factory had a problem. They sometimes shipped empty boxes without the bar inside. This challenged their perceived quality with the buyers and distributors. Understanding how important these relationships were, the CEO of the company assembled his top people.
Six months and $8 million later, they had a fantastic solution - on time, on budget, and high quality. Everyone in the project was pleased.
They solved the problem by using a special scale that would sound a bell and flash lights whenever a soap box weighed less than it should. The line would stop, someone would walk over, remove the defective box, and then press another button to re-start the line. As a result of the new package monitoring process, no empty boxes were being shipped out of the factory.
A while later, the CEO decides to look at the first week report. Since the scales were put in place, no empty boxes had been shipped out of the factory. Each day about a dozen defective boxes were being removed, which was consistent with the projections. There were almost zero customer complaints and they were gaining market share. The CEO felt the $8 million was well spent.
However, the number of defective boxes picked up by the scales dropped to zero after three weeks. He filed a bug against it and after some investigation, the engineers came back saying the report was actually correct. The scales really weren't picking up any defects because all boxes that got to that point in the conveyor belt were good.
Puzzled, the CEO traveled down to the factory, viewed the part of the line where the precision scale was installed, and observed just ahead of the new $8 million dollar solution sat a $20 desk fan blowing the empty boxes off the belt and into a bin. He asked the line supervisor what that was about.
"Oh, that," the supervisor replied, "Bert, the kid from maintenance, put it there because he was tired of walking over, removing the box and re-starting the line every time the bell rang."
Ha! I love it.
It is indeed connected. Principle of subsidiarity
Great article. Love the use of examples from different fields all pointing to the same things. And as usual the use of the visuals which makes it fun to read and more likely it will be read!
James Surowiecki's book Wisdom of Crowds sets out a couple of things that are necessary for better decisions among groups. Diversity, Independence and some mechanism for aggregating the ideas together. Without some of these conditions being met some of the examples you discuss would not work eg a Wikipedia article written only by Democrats or the work people contributed all had to be approved by me.
He also points out democracy is not just a mechanism to come up with the best policy. It is a mechanism for agreeing on how we can get along. In this regard, harmony is just as important (or more important) than the best decision. Free and fair elections where all our opinions are (sort of!) weighted equally leads to more citizen harmony and satisfaction.
Similarly when the German Govt decided to get out of coal-fired power, it didn't want to touch the issue because it would cop it in the neck whatever it decided. So it gave the problem to a commission made up of 28 stakeholders, told them to meet Germany's emissions targets and gave them funding to allocate. After nine months of negotiation they released a plan which was largely agreed on (27 of 28 stakeholders voted for it) despite decades of conflict. No one got everything they wanted. The German Govt got a plan without the grief it would have otherwise got and Germany was more settled. After all if you've a greenie and you've negotiated a plan on equal terms with a power company, unions etc, you've only got yourself to blame if you think you could have done better.
Cheers
Very good point.
I think the aggregation mechanism is the key. If it's legitimate, it solves the harmony issue, because everybody buys into the way we rule ourselves.
The example you give abut Germany is great (I didn't know!), and it's an example of an aggregation mechanism. My sense is this is strictly better than much legislation today and we should do more of it, even if it's worse than more decentralized mechanisms
Good thinking here as always. I could see local legal-twitter-analogs that develop drafts of laws that legislatures and parliaments can approve or disapprove. That can serve as a check on twitter mobs. If a particular representative is always voting against the crowd sourced drafts, that can be used as a campaign issue next election.
A big roadblock is the “legaleze” language in which laws are written. I think you could either append the crowd-sourced laws with legal definitions or have translators work on the draft laws. Too many people will not engage if drafts are not in everyday language when they are being worked on.
Two very good points.
1. Checks and Balances: you're concerned about 2 things: something being approved too fast, and something not reflecting the will of the people. In both cases, you should be able to do better than "all follow a bunch of people campaigning what kind of person they are and the bundle of opinions they have, and then give that person all the power to decide on our behalf what to do". Speed of approval can be algorithmic. Reflecting the will of the people can be asked to people directly.
2. Legalese: computer code that is fuzzy, sometimes purposefully, sometimes not. It's a middle ground between a mechanism and something human-readable. A better solution is to have these 2 layers: a layer of code, and a layer of clear explanation. Note that, in some cases, the fuzziness is a feature, not a bug. We will need a way to account for that.
First, a system that drives quick changes to a body of law is not a great idea. Stability and predictability are features of well-functioning government.
Keeping legislatures in power while the legal-pedias go through beta testing and live roll outs seems prudent. Maybe the Wikipedia and Linux examples worked because there was not much on the line when they were started. Do you think crypto currencies are a good model?
I like your description of legalese as code. The two-layer solution may work well.
Indeed.
I would never start by replacing the system wholesale. It doesn’t need to be. It can first replace local elections in an AB-tested way. As the system grows stronger, it corrects its errors, and ppl get used to it, it can move on to bigger levels of gov.
I hear your point on stability and predictability, but I’m not sure I agree. Of course, companies don’t invest if a country is not stable and they don’t know what will happen with their investment in the future. But that doesn’t mean a country’s legislation doesn’t change. Rather, that its changes are predictable. If the system works well, you should be able to see things coming, and what things are coming should in general increase the quality of the rule of law. What’s better, a shitty NEPA regulation in place for 50 years, or continuous NEPA updates in the margins because the big pbms were solved once and for all a long time ago through a system that works?
Another aspect to consider is that legislatures/ parliaments were set up as a check against the executive. If the legislature loses the power to enact laws then the executive becomes relatively more powerful. And absolute power ....
Very good point.
1. The executive could only execute the laws approved by this mechanism
2. Part of the executive could be also crowdsourced. Eg the DMV would probably be better if it was managed in a decentralized way. A DAO could be employing its workers and pay their salaries, you don't need an agency managing all of it
Yeah, I wonder how well this would work in practice. Fundamentally, societies that aren't personalist (where those with arms only owe personal loyalty to a man) work only because enough key people believe in a mythos. For instance, officers of the US military swear an oath to uphold the US constitution, and not to the President or any person. Even more importantly, I believe enough of them take that oath seriously because they believe in the constitution (they believe in that mythos). Some countries may have the same legalese but if the men with arms regard pieces of paper as only pieces of paper, any law or constitution is meaningless. So you would need an DAO that has enough mythos, or rather, enough people with arms have to believe that DAO is endowed with some mythos that they are willing to risk their life for to defend.
Ultimately, force of arms/threat of violence carries the day. No DAO is able to fire a gun.
NEPA = National Environmental Policy Act?
Yes.
Just one example.
I’m an environmentalist but I’ve heard the current regulatory framework gives it ample power to stop projects.
I agree that this is a poor law, even as amended. It frustrates anyone who wants to balance cost and benefit.
"...lawmaking will become fast, prolific, and intelligent, and we won’t all be angry all the time at the terrible politicians we’ve had to elect." - I am ever optimistic Tomas!
Tomas, I feel like you are the only writer on the internet laying out a positive vision for the future. I mean that in two senses - you reject normative analysis; and your predicted post-democratic ideology-transcending world government sounds like progress to me!
Do you have a proposed KPI? Aggregate land value? An 'Optimism Index'?
Ah I love that you say that. I had never thought of myself as that, but I'm glad you think that! The future world is great, we just need to steer it in the right direction.
I think there are some good metrics. Indeed, GDP has been a good proxy, but optimizes for time worked and money spent, and doesn't account well for intangibles like health and education. My current sense is that a blended metric like HDI or the like could be better.
|WDYT?
Fascinating, as always... One thought about your conclusion on the future of democracy and how laws could be written in the future. Is there a fear it could play out badly for minorities?
I think it would be orders of magnitude better. Right now minorities can be ignored (the rule of 50%+1) or in charge (lobbies). This should solve both issues, the same was as an article on Wikipedia doesn't get approved just because a majority thinks it should be a certain way, or because a rich person wants an article to be a certain way.
The consultation aims to give the opportunity to all citizens, whether they are residents in Tunisia or abroad, to express themselves freely and transparently about the recovery of the democratic transition in Tunisia and this, according to an unconventional mechanism, in a participatory approach allowing a common conception of the future of Tunisia. Ideas can be proposed for the development of new visions and approaches that could allow the management of public affairs in their different aspects.
The consultation also aims to make the Tunisian citizen a real actor in the process of developing a new conception of the fundamental choices linked to the political system in Tunisia.
This consultation will allow the Tunisian people, the sole holder of power, to identify the major political and economic reforms to which they aspire, and will set up a democratic framework for deliberation on various proposals that would help to face the various current challenges Tunisian citizens are confronted to in various fields.
This is what is written in the website of the consultation Tomas , making everyone participate and this is what you explained ..
I love it! And I had no idea. Thanks for sharing!
The challenge is not as much the participation (everybody is participating all the time), but rather the condensation of insights. Usually, 95% is trash, 4.9% is interesting, and 0.1% is gold. I hope the process they suggest will work to unearth and gather the gold.
That's the objective , there is a place in the consultation where you can write freely about the topics asked , the question is how they will manage to gather the best answers and suggestions from thousands of people ..
Very interesting post! I'm interested in the concept of mission command that you discuss in your first point.
Moltke the Elder's version of mission command was very much necessary due to the possibilities and limitations of military technologies you mention. Decentralization was critical when one could communicate without controlling.
Fast forward to 2022. Western militaries have the technology and capability to both communicate rapidly with technology and keep persistent communications throughout operations. In military circles, this has developed into the problem of the "8,000 mile screwdriver". In other words, military commanders have so much tech available that the tech actually enables them to micromanage to their liking (to the detriment of the organization, at times).
The U.S. military in particular loves the concept of mission command. But it struggles to implement true decentralization due to both the bias towards control and the capability of doing so. It's interesting to me because, of all the examples you sketch out, this one is a case of technology impeding decentralization.
Do you see any parallels or lessons to be learned for your broader argument, or is this just a one-off case that is specific to military contexts?
AI does the same thing.
Internet is a decentralization mechanism, but AI is a centralization mechanism. It concentrates all the data in one model, that then can make amazing decisions. Look at China: they're relying on AI to craft the perfect authoritarian state, and I fear they'll succeed.
In the future, we might ask: "Siri, what's the optimal policy for drug regulation in the US?" and it might be able to spit it out.
From what you're saying, this is what the US military is experiencing right now.
WDYT?
Thanks for the response! I'll try to respond clearly. Apologies if these thoughts seem incoherent.
I agree that AI is a centralization mechanism that complements the Internet's decentralizing tendencies. However, this looks different for how U.S. military operations tend to be run. There isn't an AI that makes optimal decisions for military organizations. Instead, that decision making is vested with a single individual, the commander.
The concept of mission command decentralizes the execution of military operations and allows decision making at lower echelons while retaining the originator's (the commander's) overall decision to do X.
Mission command is great at decentralizing execution (tell them what you expect and get out of the way if they are doing it right). Except that now, commanders have the ability to manage military activities at a distance. A general can communicate directly to a sergeant from thousands of miles away and tell that sergeant specific instructions.
The U.S. military struggles to allow the decentralization mission command calls for because single individuals - commanders - have the ability to interfere at any point of an operation. Technology enables this micromanaging. Only the self-discipline of individuals to -not- use the technology to micromanage enables mission command to work.
Going back to your first point, would mission command work better if there was an objective and optimized decision maker such as an AI? Or at least a commander that was assisted by AI in their decision making?
This may be more of a cautionary tale - if the human element isn't addressed, the benefits of network effects and decentralized systems aren't going to materialize.
These are great examples to dissect what works and why. Let's see.
It sounds like the key question in decision-making is the ability to aggregate data. Data is harder to aggregate when
1. There's a lot of data
2. The data is hard to communicate
3. Centralized data is hard to digest
If we take wikipedia, the amount of data that we're talking about is massive. So much so, that it's hard to digest. So it makes sense to decentralize it.
In law-making, there's a lot of data (all the pbms that all citizens go through, plus all the universe of potential solutions). The data is hard to communicate (you can send it to the gov, but then how does it know where to go?). And digesting it is hard (could lawmakers get all the insight from all the relevant ppl, and then make decisions alone? Hardly).
What AI does, however, is take a ton of data, centralize it, and digest it in a way that it can then automatically take the next data point and spit out an answer. That's why it's a centralization mechanism.
Now for the military: mission command assumes (1) there's too much data, (2) it's hard to communicate it, and / or (3) it's hard to digest.
I think we all agree that (1) is true in the battle in general.
In the case of von Moltke, (2) was true (telegraph back and forth second to second was not viable), and (3) too (how would you make sense of all these telegraphs?).
For the modern US army, (2) is probably not a problem (satellite imagery, sensors, data feeds...)
So if (3) is easy, even if there's a lot of data, it can be properly digested, and then it makes sense to concentrates all the decision-making in one place.
This would require, however:
a. Visualizations that allow a great mapping of the situation in real time (input)
b. The ability to produce orders extremely fast for all the teams that need them (output)
a. I don't know how good the visualizations are, you tell me. My assumptions is that in Sci Fi they can be amazing, but not sure about today's military.
b. I don't know how much time the command center has for analysis, how fast they can craft responses based on the situation, and how many of these decisions they need to make per minute.
If viz are good and the # of decisions they need to make is small, then it makes sense to do command & control. That's basically what Napoleon did.
If, however, viz are not good, or there's too many decisions that need to be made per minute, then it makes more sense to decentralize.
Maybe what's happening is that viz are good enough that *high-level decisions* can be made quickly? And so those move back to the command center as command & control? While viz is not good enough to handle every lower-level decision, so those remain as mission command?
This analysis is very intriguing. I like your three main points of reference re: data aggregation (Is there a lot of data? Can that data be communicated easily? And can it be aggregated easily?). I'm keeping this one in my pocket for future use.
The answer to your question on visualizations in military command centers is: maybe. It's all dependent on the specific unit and the echelon. For example, in the Pentagon working for the Joint Chiefs of Staff, you have access to extremely cutting edge systems, data analytics, professional data experts, and individuals at the peak of their military careers. A smaller unit in the special operations community may not have the vast resources, but that team may have the latest systems on hand that can help them receive and understand orders. And they have the expertise and seniority to make their own visualizations rapidly.
The mismatch here is with a standard company, platoon, or squad. These units don't generally have the resourcing to have great visualization tools on hand. They are nominally digital, but these networks are threatened and generally have limited bandwidth. And they are always on the move. (I would still use mission command here, as they have the expertise to operate with minimal interference, but that's another thread).
Perhaps the mismatch is when a higher level unit uses a C2 philosophy vs. a mission command philosophy without considering the capabilities of all units involved. Viz is good for the high-level unit, but not good enough at the lower unit, but the headquarters is still doing C2.
I'd also say that, in conflict, the # of decisions that need to be made can quickly overwhelm a single individual. Good commanders sort of recognize which high level decisions they must make. But in the total amount of decisions that must be made at any given moment, those decisions are probably <1% of the total amount of decisions. The other 99% could be critical, and can be directly influenced, but can be dealt with at a lower level.
Fascinating. Thanks for sharing.
Do read my comment responding to Phil Getz
https://unchartedterritories.tomaspueyo.com/p/future-of-democracy-decentralized/comment/7055988?s=w
You two have made me think a lot about this.
As always, you’ve provided info that is new to me and makes me think. This is insight into a paradigm shift that is coming. I always greatly appreciate the heads up. Thank you, Tomas.
Made my day!
I remember as an MBA in 1986 reading in our case studies about the Japanese concept of Kaizan, "gradually improving productivity by involving all employees and by making the work environment more efficient." That was revolutionary then and what you outline here is gonna be cool to see.
Ah, good one to add to the list! Thx
Not sure whether to be eagerly hopeful or incredibly terrified by the prospect of decentralized autonomous organizations making the rule of law on the fly (relatively). Interesting times ahead!
This is a moment of experimentation explosion, like in the 1500s. There were many wars of religion, but it also produced the Renaissance.
I live in the museum otherwise known as England. It is a physical museum, but of more relevance to this article is the fact that it is a social museum and a museum of thought as well. Its use of DAOs was the foundation of the British Empire.
My very brief take on England’s Geohistory would be something on the lines of:
One of the last areas in Europe to get Mesopotomian/Greek/Roman knowledge so the local tribes were vulnerable to invaders with better technology. Somewhat protected by sea, but also at the mercy of other sea powers because of that. This creates a massive incentive to develop naval power.
Flat, fertile lands in England are an asset that everyone wants, so conflict drives development of both military technology and social evolution. Communities are forced to band together for protection so the “nation” expands to the limits of geography: the Welsh and Scottish borders.
England’s subsequent success depended on the unifying mythos of nationalism and loyalty to the monarch. Not necessarily to a particular monarch, but to the idea of the monarchy. This was greatly helped by the fact that from the time of the Magna Carta onwards English monarchs were gradually forced to share more power with the nobles and then parliament. A society with a strong sense of hierarchy and obedience developed and a microcosm of that society inhabited the DAOs otherwise known as ships.
Before the development of radio there could be a massive gap in both space and time between the commander-in-chief and the action required. Orders had to be in the form of “do x by whatever means you see fit” and individual captains had to exercise their judgement in how best to achieve those aims. I am a big fan of Patrick O’Brian and his meticulously researched (but fictional) naval series about the British navy in the time of Napoleon. Part of the secret of British success was the naval education system both on land and at sea that prepared officers for obedience, but also nurtured a degree of independence.
As you have explained to some who question the relevance of Geohistory, the history of a nation can explain a large part of its psyche today. Britain still has strong elements of the very hierarchical and obedient society the enabled the empire to flourish. It worked in part because of the concept of noblesse oblige. The populace gave their loyalty and obedience and in return the nobility had some sense of honour and duty. The problem with success can be persistent belief in the things which produced that success (or were perceived to produce that success) and a reluctance to change as the world changes.
I had written the above several weeks ago, but got stuck at 75% done trying to make sense of the British attitude today. Despite their former success using DAOs, recent Britain has a strong bias towards command and control and is often regarded as a nanny state. The leaders of both sides of politics almost universally attended Oxbridge. Medical practice in England is largely dictated by centralised bureaucracy producing guidelines (which get interpreted as rules) and formularies to dictate prescribing practices. There is little trust in the knowledge and experience of individual doctors.
To get somewhere near a postable 90% I’ll just go back to the usual framework for a useful model of reality: there is a balance between command and control and the DAO concept. As your discussions above highlight, certain circumstances favour one over the other and I think you have the parameters right. Constant Conflict Theory predicts that whichever works better will eventually be widely adopted. Those who seek power and wish to impose their ideas on others will obviously favour command and control. Those who seek optimisation of results will favour listening, trusting others enough to cede some control, and allowing people and systems to develop and grow stronger.
Very interesting. And you have many similar themes. Thx for sharing!
On the DAO, that’s interesting. I’ll have to think about it.
Hypothesizing: maybe the UK was decentralized vs Europe, but much less than say the US. And centuries of success reinforced the centrality of London for power. I don’t know!
Having thought about it I would go with the ant analogy. The British ships and army were to some extent clone-like and sent out into the world to bring back useful stuff to the mother country. Strict rules governed each unit and every man had a specific role. There was strong loyalty to King and country and alignment of incentive in that all had to work together to survive, but also ships could win prize money which was shared out to the crew.
In the section Peer Review you describe releasing an article that is 90% correct and then depending on crowd sourcing to advance the 90% to 99%. This may be appropriate if you want to produce an evolving system that values only future goals. Many systems, medicine most prominently, value both current and future targets. You can not morally release a therapy that is 90% effective when you know 99% is achievable albeit at great cost. To do so makes you a monster like Elizabeth Holmes. Many other (most?) systems besides medicine may resist sacrificing current day performance for future advances. Democracy may be one of them.
There is a puzzling inconsistency in the story: you wrote, "Decentralization was enabled by faster communications"; but in fact faster communications make decentralization less necessary. Ancient Sumeria, ancient Greece, and the Roman Empire had much less centralized decision-making than the US today, because communication across them took weeks.
It would be more-accurate to say that faster communications increases the geographic scale that any system of government can achieve. So today's communications could enable the true democracy of Athens, supposing that were desirable, on the scale of a modern nation. But it also means you can create larger centralized totalitarian governments than ever before, as we saw in the 20th century.
I've been thinking about your comment for a few hours already. You hit on something important and quite fundamental. So let me think out loud, tell me what you think.
First, I think I mix two concepts in the article: subsidiarity (which is hierarchical) and decentralization (not hierarchical).
In hierarchical decision-making, high-level decisions are made at the top, low-level decisions at the bottom. The principle of subsidiarity simply says that as many decisions should be pushed down as possible. In what circumstances is this valid?
1. If low-level nodes don't need to coordinate with nodes from another part of the network.
2. If there's a lot of information on the ground
3. If, when information is transmitted up, there is a loss of fidelity
4. If it's hard to aggregate the information from many different sources.
Both Napoleon and von Moltke were hierarchical, but in von Moltke's time, subsidiarity worked better. Why?
1. Low-level nodes don't need to coordinate with nodes in other parts of the network. They need to fight with their own unit. They do, however, need to coordinate at the battalions / armies level. Which was the big drive for Napoleon to coordinate everything, and one of the reasons why von Moltke's armies sometimes suffered (they counted on other forces to support them). So I'd say this pushed more for authoritarian hierarchy than subsidiarity
2. Lots of info on the ground: this was certainly true. With the trains and telegraphs, many more teams could act across a much vaster expanse. Battles included 100ks of ppl, not thousands.
3. Loss of fidelity: probably high loss of fidelity when transmitting information, because of the distances. But there was the telegraph, so maybe that counteracted the other trend?
4. Hard to aggregate the information from all the sources: probably true given just the sheer amount of data that had to be transmitted, and the fact that the telegraph was not a high bandwidth technology at the time.
So maybe what happened is that the dramatic increase in information, due to the train and telegraph enabling many more ppl interacting across many more areas, meant that it was just impossible to centralize that data quickly and then to digest it, whereas it was still possible with Napoleon.
Maybe in fact this is one of the reasons why Napoleon was considered such a good military leader: at that point, battles required coordinating vast amounts of teams across different areas, and that was a mess to do, but Napoleon was able to get the data to him quickly, and his brain was able to process it fast enough to give orders?
So that would be for hierarchies: subsidiarity vs. authority (or mission command vs. command and control).
Now the question becomes: when is a decentralized process better than hierarchy (either authority or subsidiarity).
For example, nearly all companies are hierarchical. The few that are decentralized are famous because of it: the holacracy of Valve (maker of Steam), and at some point Zappos. So clearly decentralization is not always better. Why does hierarchy work better than decentralization in companies?
It sounds to me like there are forces that push decision-making to be hierarchical vs. decentralized.
One of them is the importance of alignment. The more a lot of ppl need to be single-mindedly rowing in the same direction, the more this pushes for centralized coordination to make sure one decision is made, quickly, and everybody follows it. Companies need this
Another one is, again, the amount of information at the node level. The more there is, the harder it is to convey all this information to a central node, and the more you want those close to the info to make the decisions.
Another one is the need to coordinate across all nodes. If you can easily divide people into autonomous teams, they don't need to coordinate as much between them, and decentralization is less useful.
Another one, and I have the intuition that this is key, is the ability for any message to find the right recipient. If it's very hard to get all the info to the right ppl, then instead you'll use the hierarchy to aggregate it up and communicate it down. Decentralization works well when (a) the right person can be found, OR (b) the message between nodes can pass through.
(a)
If we take wikipedia, every person might write an article, but then other ppl interested in that article can easily find it. You only see the articles you're most interested and knowledgeable in.
On Twitter, you can find the right messages because you start following the ppl you're interested in.
(b)
Capitalism has some of it: you only pay attention to the industry / goods you're interested in. But it also has another crucial value: the mechanism of prices conveys information across nodes.
If a miner has a harder time finding some ore, the ore he does find will be sold at a higher price, which the furnace owner will have to pay, and then pass through to their own customers, who then pass through the price increase all the way to the end consumer, who might then buy it or not, sending back the message through the entire chain. The information doesn't lose much fidelity at every step.
Neurons do the same: they pass action-potentials from one to the other, without much signal loss.
Twitter does the same: the retweet has no signal loss.
In wikipedia, obviously, there's no signal loss from one person to another: the article is fixed and accessible to all.
So maybe the main handicap of decentralization is the loss of signal when signals are relayed across nodes. You can definitely see this in democracy: people can't communicate with each other efficiently. Same arguments all the time, too emotional, lack of depth, different priorities, lack of clarity of the areas of expertise of each...
If this is true, the key factor for decentralizing democracy would be a lossless way to convey political insight.
Thank you so much for your analysis, ideas, and inspiration. Your substack is the best thing I read.
I’m glad to hear!