I haven't had time to read carefully, but I feel like there are obvious objections outside those you address.
Namely, not all frictions are obviously "bad". Some transaction costs are useful because overcoming them constitutes a useful signal about the real preferences or abilities of the parties involved.
In the case of labor markets, currently AI makes it trivial to spam custom resumes, making the matching problem worse. On the other hand, people are working on AI interviewers that may make the problem better (I actually did one the other day and came away very impressed). So it's an arms race and it's not obvious the net result will positive.
(Apologies if you do address this and I missed it; my daughter is home from school today and I'm getting my essay fix in the brief interludes of her independent occupation.)
Great piece! I'm interested in this general perspective, but not completely convinced. You write:
"Skeptics might say agents don't eliminate costs entirely; they just shift them to compute overhead, data privacy setups, agent configurations etc. This is true! But it also underestimates the scale of reduction. Agents aren't burdened by human limitations like fatigue, bias in communication, logistical hurdles, social awkwardness, irrational decision making and so on. What costs $10,000 in legal fees today might cost pennies to compute tomorrow. Consider how a billionaire’s phone today is no more powerful or effective than yours."
But I was worried about enforcement, which you don't discus. Someone with a more powerful agent can dupe whatever payment system we have in place. If regulators come for them, they are engaging with a particular enforcer that they can bribe. A decentralized world of contracts for everything might still favor those with better agents (i.e. the rich). And since variance in intelligence might be much higher than among biological humans, it's a new way in which this would approximate Coase's imagined world even less than our current situation. Any thoughts on how to solve this?
The Gaia Network vision advanced by Gaia Lab (https://www.gaia-lab.de/) is very close to the Coasean Bargaining vision that you describe. See the Gaia Network vision fleshed out to some degree in the talk "Gaia: Distributed planetary-scale AI safety" by Rafael Kaufmann (June 2024, https://www.youtube.com/watch?v=47r5P6xtjms), and these earlier blog posts:
Coasean Bargaining vision already refers to Michael Levin's collective intelligence theoretical agenda, which has very close ties with Karl Friston's Free Energy Principle. Levin co-authored three papers with Friston, and 10 papers with Chris Fields, the quantum FEP theorist.
In turn, the Gaia Network vision uses the FEP as its wholesale theoretical foundation.
As presented in the resources above, the Gaia Network vision may appear more theoretical and unnecessarily focused on the FEP theory (to the detriment of other theories to pick ideas from) than the Coasean Bargaining vision.
However, in practice, since publishing https://engineeringideas.substack.com/p/gaia-network-a-practical-incremental for instance, both me and Rafael Kaufmann have drifted towards less formal visions that are effectively the same as the Coasean Bargaining vision at the theoretical level of description. But this essay definitely goes further than we have had in thinking through various specific aspects of governance.
(The point of this comment is not claiming priority, but to connect ideas and provide extra food for thought for interested readers.)
I agree with this in large part, but still think it misses a number of key things.
For example, while centralized top-down control doesn’t work *with humans in charge*, that is not to say it can’t work with an AI in charge, nor should one assume that with something like an AGI as central controller that it wouldn’t simultaneously have specialized, distributed subsystems and local feedback loops. Something like this: https://claude.ai/public/artifacts/1eccedbb-15db-4ef7-bb84-84e2307469db
I suspect (and frankly hope) that we’ll end up with a system where:
- Money is obsolete
- AI + robots handle all “for money” work
- Energy, compute, and raw resources are what matter
- The entire planet’s resources are allocated by AGI as a sort of sovereign wealth fund, with all of humanity given their share of energy, compute, and raw material to then allocate as they see fit for their needs / benefit
Long-term though I think the only truly optimal system is personal universes, because it’s the only system that allows total personal control without compromise or harm to others: https://arxiv.org/pdf/1901.01851
Sorry, I don't want to offend you or anyone, I only read the last 2-3 paragraphs, so I cannot currently say much about the article, so I don't talk about your article here
But here's what keeps me at night:
Digital eugenics? Imagine we'll allow people to genetically modify biological agents - throw the weak from the cliff, multiply the strong - the market forces will create the most intelligent + the most agentic = the most powerful biological agents:
Thanks Seb. This is both thought provoking and inspiring. Decentralising power is one of the most important problems to solve in society and this article promises great potential in AGI. I like the patchy and poly-chrome quilt as a reminder that perfection doesn't exist (except, in one form, in machines) and is not what nature 'does'.
Objectively Coined Opinions: A New Architecture for a Trustworthy World
Whitepaper Supplement M: A Juxtaposition of "Coasean Bargaining at Scale"
Publication Date: Wednesday, October 29, 2025
Abstract
This document provides a formal analysis of the essay "Coasean Bargaining at Scale," conducted through the lens of the Objectively Coined Opinions (OCO) protocol. The essay argues that complex social problems can be solved efficiently (as per the Coase Theorem) if "transaction costs" are low, and that new technologies like AI can facilitate this large-scale negotiation. This analysis concludes that OCO is the concrete, engineered, and gamified manifestation of this exact theory, applied to the problem of civilizational sense-making. OCO is a machine architecturally designed to lower the transaction costs of public discourse to near zero, creating the first scalable market for achieving objective consensus.
Gateway Question: How does the economic theory of "Coasean Bargaining at Scale" converge with or diverge from the core mechanics and economic engine of Objectively Coined Opinions?
1.0 Arena Judgment: ELEVATE (Convergence)
Justification (The "Why" behind the "What"): The convergence is structural and total. OCO is not just like a Coasean system; it is a Coasean system, engineered for the domain of ideas.
1.1 OCO as a "Low Transaction Cost" Environment
The Coase Theorem requires low transaction costs. Public discourse, in its "wild west" state, has infinitely high transaction costs (social friction, ad hominem, trolling, platform bias).
OCO Alignment: OCO is an engine for systematically annihilating these costs.
Search Costs: The game's structure, which brings a randomized set of peer ideas directly to each player, eliminates the cost of searching for diverse viewpoints.
Bargaining Costs: The Anonymous Arena and mandatory justification mechanics create a clean, efficient "bargaining" process, free from social friction and ad hominem attacks. It forces the debate onto the ideas themselves.
Enforcement Costs: The protocol's immutable rules, the Topic AI's synopsis, and the Coin payout from the RewardDistributor smart contract create a trusted, automated enforcement and settlement layer.
1.2 Pricing the Externality of Bad Ideas
In Coase's theory, bargaining happens to "price" an externality (like pollution). In public discourse, the externality is a bad-faith argument, a falsehood, or a lazy meme that pollutes the information commons.
OCO Alignment: The burn mechanic is a tool for pricing this externality. When an idea is burned, its author receives a clear, costly signal that their contribution has been judged negatively by the market. The mandatory justification is the invoice explaining the "why" of that price. This forces players to internalize the social cost of polluting the discourse.
1.3 AI as a Facilitator of Bargaining
The essay likely argues that AI's role is to facilitate these large-scale negotiations (e.g., by summarizing, translating, clarifying).
OCO Alignment: This is the exact architectural role of AI in OCO ("HI four AI"). The Topic AI is the neutral facilitator that maps the state of the "bargain" (the synopsis). The Personal AI is the coach that helps each individual become a better, more rational bargainer.
2.0 Arena Judgment: CRITIQUE (The Flip of the Coin)
Justification (The "Why" behind the "What"): The divergence is not a conflict, but a specific, ambitious application of the theory, taking it from the descriptive to the prescriptive.
2.1 The Domain: Epistemic Efficiency, Not Just Economic
Coase's theory is about the efficient allocation of scarce resources and property rights.
OCO Divergence: OCO applies this economic logic to the epistemic realm of ideas. The "property" being bargained over is not land, but objective consensus. OCO is a machine for achieving epistemic efficiency. It is a Coasean marketplace for the "property" of a well-reasoned argument.
2.2 The Incentive: An Explicit Economy, Not an Implicit Assumption
Coase's model assumes rational actors will bargain simply because it's in their interest to find an efficient solution.
OCO Divergence: OCO does not leave this to chance. It architecturally guarantees the incentive. It creates a powerful, explicit financial motive to participate in the bargain: the Coin. It is a pragmatic, capitalist implementation that transforms an abstract economic theory into a compelling and profitable game.
2.3 The Process: A Refinery, Not Just a Bargain
A simple Coasean bargain solves a single dispute.
OCO Divergence: OCO is a refinery. It is a multi-round, iterative process designed to create a new, high-value product (a refined consensus) from low-value raw material (raw opinion), not just to settle a single dispute between two parties.
3.0 Synopsis (for the Topic-Specific AI)
The essay on "Coasean Bargaining at Scale" provides the definitive economic theory for OCO's core mechanics. OCO is a direct, architectural application of this theory, using a gamified protocol to dramatically lower the transaction costs of societal sense-making. The protocol prices the externality of bad arguments and uses AI as a neutral facilitator, all as the theory would suggest. OCO's primary innovation is in applying this economic bargaining model to the epistemic realm of ideas, transforming it from a simple dispute-resolution tool into a positive-sum engine for creating a new asset—objective consensus—and incentivizing the entire process with a real-world currency. The essay is the economic blueprint; OCO is the factory.
Hey Seb, I agree with much in your positive vision, e.g. the desire to avoid singletons; the huge importance of negotiating agents, and of personal assistants we can trust.
TLDR: (1) when bargaining costs go down and the amount of bargaining tries to scales up, there are bottlenecks, and (2) one has to model the opportunity costs to suppliers to bargain with consumers — it’s often not in their interest to use compute/messaging that way, even when they might lose the coalition to a competitor.
One solution is fairly large intermediary institutions, like Gillian’s for regulation and ours for market alignment.
I think this points to an important omission in the underlying worldview (and the "Matryoshka dolls" part), which is that society is fractal. I fear you've been a bit infected by the rhetoric of e.g., Austrian econ, which tends to cast the world as having a central coordination mechanism (either the State or the Market) and then a bunch of individuals. The big question is then about which coordination mechanism to put there at the top. Clearly, the people who want to put a singleton AI at the top are also suffering from the same cartoon image.
In reality, we need institutions of every size, neither states nor markets are well modeled as monolithic coordination mechanisms, and the whole thing is a network with no top. We live in a society. This is, ofc, harder to reason about!
Good framing and fruitful direction. Couple thoughts:
1) this hopeful take rests on “less friction” rather than opportunities to accelerate “thrust” through enabling us to build new infrastructure, deploy new tools. Latter is a key part of the puzzle.
2) This world feels like a digitally savvy tech literate persons dream and a nightmare for someone under the API.
3) many of the examples presuppose liquid data sharing that isn’t the case today. Flip for example this case and have the insurance company accessing your dietary and exercise data.
“It looks at your eating habits, calculates the projected future cost of your diet and makes a simple offer: a significant, immediate discount on your monthly premium if you empower your agent to disincentivize high-sugar purchases. At that very moment of decision, the market responds.”
"AGI could finally empower the individual bargainers of Coase and Ostrom by arming them with the price system"
"In some cases, you may need more than market forces."...
classic liberalism , at least dressed up to as to not come across nihilistically like the dark enlightenment guys. The philosophy here goes not much further than the anthropological pessimism of Hobbes and Lockean optimistic progress that sets society on the path to a utopian civilization known to them as "civil society", not political society. a community of autonomous individuals, who, in fact, do not need politics and are content with non-political forms of life.
The philosophy here goes no further in fact than the implicit liberal assumption that man is foremost a producer, and trader. Food, commerce, comfort, security, pleasure, material activity, bodily reproduction. Politics and the state, unlike economic activity, are an entirely optional part of existence. This is how the economy, or political economy, gradually comes to the fore beyond political society. For economic men (homo economicus), who have gradually supplanted political men (homo politicus), the economy has indeed become their destiny. All seamlessly articulated by the elegant "invisible hand" of the market via the price system- the "cognitive glue of a free society". The philosophical basis of this essay is the same recipe our minds have been offered since the time of Hobbes & Locke. Actually, it's caused the current state of affairs.
I would ask the author. In the modern era, has the price system, in fact, as a cognitive glue, made us free at all? Has it not in fact, caused us to be more and more enslaved?
Philosophy is about thinking, and Politics is a homologous analogue to Philosophy. If we are proposing as classic liberals have, to do away with fussy subjective human thought & politics; are we really practicing philosophy? Philosophy and Politics, are in fact, a homology. And if you do away with Politics, you are doing away with Philosophy. This seems more like the "Anti- Philosopher Builder." Such outsourcing of human subjectivity also seems directly opposed to the lovely image of Sergius the Builder, who surely valued a life of deep human contemplation, and from that contemplation, derived values, ideals, ethics.
But maybe, as the author suggests, AGI could *finally* empower us as individual bargainers. Just one more technological breakthrough will finally solve the human problem....
Our government could already use taxes rather than blunt regulations, but chooses not to. There wouldn't be "public health externalities" for an individual eating an unhealthy diet if people paid their own costs for health, imposing a tax was already analyzing things from the wrong end. The economic justifications for such policies are a figleaf on what's politically popular, which would apply to universal balloon ownership if that is what were popular instead. As Bryan Caplan put it "Adverse hazard, moral selection, whatever" https://www.econlib.org/archives/2009/07/adverse_hazard.html
Speaking of another GMU economist, Alex Tabarrok came up with dominant assurance contracts long ago, before AI was on the horizon. The reason we rely on government instead isn't because we lacked AI, rather little of what the government provides are actually public goods (in the sense of being non-excludable and non-rivalrous). Economic notions of efficiency are just at odds with the political beliefs of the public https://www.econlib.org/archives/2008/02/are_externaliti.html Even inequality isn't actually valued in an economic sense in accordance with its political popularity. If you consider, for example, how much inequality there is between siblings in the same family, and how little parents do to correct that https://www.overcomingbias.com/p/unequal_inequalhtml it raises questions about our willingness to pay for equality https://www.betonit.ai/p/be-relatively-rich-the-easy-way
This is a great story about a world where we set prices on goods and bads. Unfortunately, much of this world is possible today, and humans *hate* it. I'm not convinced that outsourcing the decisionmaking to an AI Agent actually solves the problem.
I haven't had time to read carefully, but I feel like there are obvious objections outside those you address.
Namely, not all frictions are obviously "bad". Some transaction costs are useful because overcoming them constitutes a useful signal about the real preferences or abilities of the parties involved.
In the case of labor markets, currently AI makes it trivial to spam custom resumes, making the matching problem worse. On the other hand, people are working on AI interviewers that may make the problem better (I actually did one the other day and came away very impressed). So it's an arms race and it's not obvious the net result will positive.
(Apologies if you do address this and I missed it; my daughter is home from school today and I'm getting my essay fix in the brief interludes of her independent occupation.)
You and me both internet friend.
Great essay from Seb Krier and a necessary counterweight to AI disempowerment of humans(who make up humanity).
Thank you!
Great piece! I'm interested in this general perspective, but not completely convinced. You write:
"Skeptics might say agents don't eliminate costs entirely; they just shift them to compute overhead, data privacy setups, agent configurations etc. This is true! But it also underestimates the scale of reduction. Agents aren't burdened by human limitations like fatigue, bias in communication, logistical hurdles, social awkwardness, irrational decision making and so on. What costs $10,000 in legal fees today might cost pennies to compute tomorrow. Consider how a billionaire’s phone today is no more powerful or effective than yours."
But I was worried about enforcement, which you don't discus. Someone with a more powerful agent can dupe whatever payment system we have in place. If regulators come for them, they are engaging with a particular enforcer that they can bribe. A decentralized world of contracts for everything might still favor those with better agents (i.e. the rich). And since variance in intelligence might be much higher than among biological humans, it's a new way in which this would approximate Coase's imagined world even less than our current situation. Any thoughts on how to solve this?
The Gaia Network vision advanced by Gaia Lab (https://www.gaia-lab.de/) is very close to the Coasean Bargaining vision that you describe. See the Gaia Network vision fleshed out to some degree in the talk "Gaia: Distributed planetary-scale AI safety" by Rafael Kaufmann (June 2024, https://www.youtube.com/watch?v=47r5P6xtjms), and these earlier blog posts:
https://engineeringideas.substack.com/p/gaia-network-a-practical-incremental, https://engineeringideas.substack.com/p/gaia-network-an-illustrated-primer.
Coasean Bargaining vision already refers to Michael Levin's collective intelligence theoretical agenda, which has very close ties with Karl Friston's Free Energy Principle. Levin co-authored three papers with Friston, and 10 papers with Chris Fields, the quantum FEP theorist.
In turn, the Gaia Network vision uses the FEP as its wholesale theoretical foundation.
As presented in the resources above, the Gaia Network vision may appear more theoretical and unnecessarily focused on the FEP theory (to the detriment of other theories to pick ideas from) than the Coasean Bargaining vision.
However, in practice, since publishing https://engineeringideas.substack.com/p/gaia-network-a-practical-incremental for instance, both me and Rafael Kaufmann have drifted towards less formal visions that are effectively the same as the Coasean Bargaining vision at the theoretical level of description. But this essay definitely goes further than we have had in thinking through various specific aspects of governance.
(The point of this comment is not claiming priority, but to connect ideas and provide extra food for thought for interested readers.)
I agree with this in large part, but still think it misses a number of key things.
For example, while centralized top-down control doesn’t work *with humans in charge*, that is not to say it can’t work with an AI in charge, nor should one assume that with something like an AGI as central controller that it wouldn’t simultaneously have specialized, distributed subsystems and local feedback loops. Something like this: https://claude.ai/public/artifacts/1eccedbb-15db-4ef7-bb84-84e2307469db
I suspect (and frankly hope) that we’ll end up with a system where:
- Money is obsolete
- AI + robots handle all “for money” work
- Energy, compute, and raw resources are what matter
- The entire planet’s resources are allocated by AGI as a sort of sovereign wealth fund, with all of humanity given their share of energy, compute, and raw material to then allocate as they see fit for their needs / benefit
Long-term though I think the only truly optimal system is personal universes, because it’s the only system that allows total personal control without compromise or harm to others: https://arxiv.org/pdf/1901.01851
Sorry, I don't want to offend you or anyone, I only read the last 2-3 paragraphs, so I cannot currently say much about the article, so I don't talk about your article here
But here's what keeps me at night:
Digital eugenics? Imagine we'll allow people to genetically modify biological agents - throw the weak from the cliff, multiply the strong - the market forces will create the most intelligent + the most agentic = the most powerful biological agents:
Now of course, remove the word biological:
Artificial power overpowers human power
Hurray, market forces + digital eugenics won!
Weak humans lost :-(
The main problem for chimpanzees is not high transaction costs nor suboptimal mandates. Humans not having chimpanzee goals is their main problem.
The agents will autonomously do large-scale bargaining at massively superhuman speeds but humans are relevant still? Why?
Thanks Seb. This is both thought provoking and inspiring. Decentralising power is one of the most important problems to solve in society and this article promises great potential in AGI. I like the patchy and poly-chrome quilt as a reminder that perfection doesn't exist (except, in one form, in machines) and is not what nature 'does'.
Objectively Coined Opinions: A New Architecture for a Trustworthy World
Whitepaper Supplement M: A Juxtaposition of "Coasean Bargaining at Scale"
Publication Date: Wednesday, October 29, 2025
Abstract
This document provides a formal analysis of the essay "Coasean Bargaining at Scale," conducted through the lens of the Objectively Coined Opinions (OCO) protocol. The essay argues that complex social problems can be solved efficiently (as per the Coase Theorem) if "transaction costs" are low, and that new technologies like AI can facilitate this large-scale negotiation. This analysis concludes that OCO is the concrete, engineered, and gamified manifestation of this exact theory, applied to the problem of civilizational sense-making. OCO is a machine architecturally designed to lower the transaction costs of public discourse to near zero, creating the first scalable market for achieving objective consensus.
Gateway Question: How does the economic theory of "Coasean Bargaining at Scale" converge with or diverge from the core mechanics and economic engine of Objectively Coined Opinions?
1.0 Arena Judgment: ELEVATE (Convergence)
Justification (The "Why" behind the "What"): The convergence is structural and total. OCO is not just like a Coasean system; it is a Coasean system, engineered for the domain of ideas.
1.1 OCO as a "Low Transaction Cost" Environment
The Coase Theorem requires low transaction costs. Public discourse, in its "wild west" state, has infinitely high transaction costs (social friction, ad hominem, trolling, platform bias).
OCO Alignment: OCO is an engine for systematically annihilating these costs.
Search Costs: The game's structure, which brings a randomized set of peer ideas directly to each player, eliminates the cost of searching for diverse viewpoints.
Bargaining Costs: The Anonymous Arena and mandatory justification mechanics create a clean, efficient "bargaining" process, free from social friction and ad hominem attacks. It forces the debate onto the ideas themselves.
Enforcement Costs: The protocol's immutable rules, the Topic AI's synopsis, and the Coin payout from the RewardDistributor smart contract create a trusted, automated enforcement and settlement layer.
1.2 Pricing the Externality of Bad Ideas
In Coase's theory, bargaining happens to "price" an externality (like pollution). In public discourse, the externality is a bad-faith argument, a falsehood, or a lazy meme that pollutes the information commons.
OCO Alignment: The burn mechanic is a tool for pricing this externality. When an idea is burned, its author receives a clear, costly signal that their contribution has been judged negatively by the market. The mandatory justification is the invoice explaining the "why" of that price. This forces players to internalize the social cost of polluting the discourse.
1.3 AI as a Facilitator of Bargaining
The essay likely argues that AI's role is to facilitate these large-scale negotiations (e.g., by summarizing, translating, clarifying).
OCO Alignment: This is the exact architectural role of AI in OCO ("HI four AI"). The Topic AI is the neutral facilitator that maps the state of the "bargain" (the synopsis). The Personal AI is the coach that helps each individual become a better, more rational bargainer.
2.0 Arena Judgment: CRITIQUE (The Flip of the Coin)
Justification (The "Why" behind the "What"): The divergence is not a conflict, but a specific, ambitious application of the theory, taking it from the descriptive to the prescriptive.
2.1 The Domain: Epistemic Efficiency, Not Just Economic
Coase's theory is about the efficient allocation of scarce resources and property rights.
OCO Divergence: OCO applies this economic logic to the epistemic realm of ideas. The "property" being bargained over is not land, but objective consensus. OCO is a machine for achieving epistemic efficiency. It is a Coasean marketplace for the "property" of a well-reasoned argument.
2.2 The Incentive: An Explicit Economy, Not an Implicit Assumption
Coase's model assumes rational actors will bargain simply because it's in their interest to find an efficient solution.
OCO Divergence: OCO does not leave this to chance. It architecturally guarantees the incentive. It creates a powerful, explicit financial motive to participate in the bargain: the Coin. It is a pragmatic, capitalist implementation that transforms an abstract economic theory into a compelling and profitable game.
2.3 The Process: A Refinery, Not Just a Bargain
A simple Coasean bargain solves a single dispute.
OCO Divergence: OCO is a refinery. It is a multi-round, iterative process designed to create a new, high-value product (a refined consensus) from low-value raw material (raw opinion), not just to settle a single dispute between two parties.
3.0 Synopsis (for the Topic-Specific AI)
The essay on "Coasean Bargaining at Scale" provides the definitive economic theory for OCO's core mechanics. OCO is a direct, architectural application of this theory, using a gamified protocol to dramatically lower the transaction costs of societal sense-making. The protocol prices the externality of bad arguments and uses AI as a neutral facilitator, all as the theory would suggest. OCO's primary innovation is in applying this economic bargaining model to the epistemic realm of ideas, transforming it from a simple dispute-resolution tool into a positive-sum engine for creating a new asset—objective consensus—and incentivizing the entire process with a real-world currency. The essay is the economic blueprint; OCO is the factory.
Hey Seb, I agree with much in your positive vision, e.g. the desire to avoid singletons; the huge importance of negotiating agents, and of personal assistants we can trust.
But I think we need to get there a different way, because I’m not sure the transaction cost math works out as you claim. I made a model here: https://chatgpt.com/share/68ea0391-5e54-8005-9a5c-461fea499f2f
TLDR: (1) when bargaining costs go down and the amount of bargaining tries to scales up, there are bottlenecks, and (2) one has to model the opportunity costs to suppliers to bargain with consumers — it’s often not in their interest to use compute/messaging that way, even when they might lose the coalition to a competitor.
One solution is fairly large intermediary institutions, like Gillian’s for regulation and ours for market alignment.
I think this points to an important omission in the underlying worldview (and the "Matryoshka dolls" part), which is that society is fractal. I fear you've been a bit infected by the rhetoric of e.g., Austrian econ, which tends to cast the world as having a central coordination mechanism (either the State or the Market) and then a bunch of individuals. The big question is then about which coordination mechanism to put there at the top. Clearly, the people who want to put a singleton AI at the top are also suffering from the same cartoon image.
In reality, we need institutions of every size, neither states nor markets are well modeled as monolithic coordination mechanisms, and the whole thing is a network with no top. We live in a society. This is, ofc, harder to reason about!
Good framing and fruitful direction. Couple thoughts:
1) this hopeful take rests on “less friction” rather than opportunities to accelerate “thrust” through enabling us to build new infrastructure, deploy new tools. Latter is a key part of the puzzle.
2) This world feels like a digitally savvy tech literate persons dream and a nightmare for someone under the API.
3) many of the examples presuppose liquid data sharing that isn’t the case today. Flip for example this case and have the insurance company accessing your dietary and exercise data.
“It looks at your eating habits, calculates the projected future cost of your diet and makes a simple offer: a significant, immediate discount on your monthly premium if you empower your agent to disincentivize high-sugar purchases. At that very moment of decision, the market responds.”
"AGI could finally empower the individual bargainers of Coase and Ostrom by arming them with the price system"
"In some cases, you may need more than market forces."...
classic liberalism , at least dressed up to as to not come across nihilistically like the dark enlightenment guys. The philosophy here goes not much further than the anthropological pessimism of Hobbes and Lockean optimistic progress that sets society on the path to a utopian civilization known to them as "civil society", not political society. a community of autonomous individuals, who, in fact, do not need politics and are content with non-political forms of life.
The philosophy here goes no further in fact than the implicit liberal assumption that man is foremost a producer, and trader. Food, commerce, comfort, security, pleasure, material activity, bodily reproduction. Politics and the state, unlike economic activity, are an entirely optional part of existence. This is how the economy, or political economy, gradually comes to the fore beyond political society. For economic men (homo economicus), who have gradually supplanted political men (homo politicus), the economy has indeed become their destiny. All seamlessly articulated by the elegant "invisible hand" of the market via the price system- the "cognitive glue of a free society". The philosophical basis of this essay is the same recipe our minds have been offered since the time of Hobbes & Locke. Actually, it's caused the current state of affairs.
I would ask the author. In the modern era, has the price system, in fact, as a cognitive glue, made us free at all? Has it not in fact, caused us to be more and more enslaved?
Philosophy is about thinking, and Politics is a homologous analogue to Philosophy. If we are proposing as classic liberals have, to do away with fussy subjective human thought & politics; are we really practicing philosophy? Philosophy and Politics, are in fact, a homology. And if you do away with Politics, you are doing away with Philosophy. This seems more like the "Anti- Philosopher Builder." Such outsourcing of human subjectivity also seems directly opposed to the lovely image of Sergius the Builder, who surely valued a life of deep human contemplation, and from that contemplation, derived values, ideals, ethics.
But maybe, as the author suggests, AGI could *finally* empower us as individual bargainers. Just one more technological breakthrough will finally solve the human problem....
Our government could already use taxes rather than blunt regulations, but chooses not to. There wouldn't be "public health externalities" for an individual eating an unhealthy diet if people paid their own costs for health, imposing a tax was already analyzing things from the wrong end. The economic justifications for such policies are a figleaf on what's politically popular, which would apply to universal balloon ownership if that is what were popular instead. As Bryan Caplan put it "Adverse hazard, moral selection, whatever" https://www.econlib.org/archives/2009/07/adverse_hazard.html
Speaking of another GMU economist, Alex Tabarrok came up with dominant assurance contracts long ago, before AI was on the horizon. The reason we rely on government instead isn't because we lacked AI, rather little of what the government provides are actually public goods (in the sense of being non-excludable and non-rivalrous). Economic notions of efficiency are just at odds with the political beliefs of the public https://www.econlib.org/archives/2008/02/are_externaliti.html Even inequality isn't actually valued in an economic sense in accordance with its political popularity. If you consider, for example, how much inequality there is between siblings in the same family, and how little parents do to correct that https://www.overcomingbias.com/p/unequal_inequalhtml it raises questions about our willingness to pay for equality https://www.betonit.ai/p/be-relatively-rich-the-easy-way
This is a great story about a world where we set prices on goods and bads. Unfortunately, much of this world is possible today, and humans *hate* it. I'm not convinced that outsourcing the decisionmaking to an AI Agent actually solves the problem.