Cross-posted from my Substack.
In this article, I highlight some of the ideas from Musthtaq Khan’s interview for 80000hours podcast about institutional economics[1], political economy, his “political settlement” framework, and the methodology of economics, and connect these ideas to the concepts in scale-free regulative development[2], morphogenesis[3], and cognitive science[4][5].
Khan points to the fact that socioeconomies are non-ergodic dynamical systems with hysteresis, i.e., memories of their own[1]:
To say that all structures are created by individuals, and therefore if the structure of society in India is different from the one in the United States, then we have to look at the individual incentives that created those structures, I think is a non-starter. It confuses the path dependence of history and the complexity of how structures are built up. Individuals today in India may not have any capacity of changing that structure to look like the one in the U.S. or Norway, not because they have some information deficit or anything like that, but because a structure itself has a reality and a meaning which affects the way individuals behave.
Here’s the exactly parallel observation that Levin[3] makes about morphogenesis:
Development is thus incredibly reliable, producing bodies to very tight tolerance despite considerable deviations and noise at the level of gene expression and cellular activity (Gonze et al. 2018; Eritano et al. 2020; Simon, Hadjantonakis, and Schroter 2018). This robustness, and its occasional failure in the case of birth defects immediately suggests teleonomic perspectives because only goal-directed agents can make mistakes; biophysics alone cannot make mistakes – every micro-scale process proceeds according to the laws of physics and chemistry. Developmental defects are mistakes relative to the correct outcome toward which they strive.
This means that from the perspective of scale-free teleonomy, socioeconomies must be “goal-directed agents that can make mistakes”. The causal (or “creation”) link between individual behaviour and institutions (socioeconomic structures) is bidirectional rather than unidirectional from individual behaviour to institutions.
Khan’s position is that viewing economy exclusively through the lens of individual choice (behaviour) and emergence of supply, demand, and market prices is overly reductive and misses opportunities for intervention and explanation. This non-reductionist approach[6] to economics is, of course, shared by Levin with his non-reductionist approach to biology and medicine[3]:
These attempts to view morphogenesis a not merely an emergent physical process but a goal-directed control loop have led to many new discoveries and novel capabilities in the prediction and control of anatomical outcomes that had not been discovered from prior bottom-up approaches, and which offer numerous advantages for regenerative medicine.
In the passage quoted above, Khan also seems to suggest that just increasing the level of individual education in the society, though it may help, could be insufficient to change the economic structure and build the economy full of high-capability organisations[1]:
Individuals today in India may not have any capacity of changing that structure to look like the one in the U.S. or Norway, not because they have some information deficit or anything like that, but because a structure itself has a reality and a meaning which affects the way individuals behave.
Individual education (economics) seems to correspond to the knowledge encoded in genes (biology)[3]:
Because genomes encode micro-level protein hardware, not directly specifying growth and form, it is essential to understand not only molecular mechanisms necessary for morphogenesis, but also the information-processing dynamics that are sufficient for the swarm intelligence of cell groups to create, repair, and reconstruct large-scale anatomical features.
However, Khan distinguishes between public or university education with the actual know-how of how to organise a factory, a hospital, a logistics company, and so on, which cannot be learned in school but mostly through practice[1]:
[…] this has nothing much to do with what you learn in business school. It is to do with how you organize a whole team of people to operate seamlessly as an organic whole. And it sounds to us to be rather obvious, but this is an incredibly difficult thing to achieve. Take the example of a hospital in a developing country. Hospitals in developing countries have doctors who are very skilled. In fact, most of them would love to leave and take a job in an advanced country where they would perform perfectly well. They have all the machines that you require for a hospital. They have the drugs, or many or most of them. And yet their capacity to deliver good health services is very poor. The reason is not the quality of the people or the quality of the machines. It’s how it’s organized. Are you doing the cleaning properly? Are you managing the flow of tests so that the right tests go at the right time to the right doctor for the right patient? Are you managing your entry so that the beds are kept just about full enough, but not overly full? Are you managing your quality control and your ordering of spare parts? And this is where it fails. This is where universities don’t work in some countries, hospitals don’t work in other countries. Not because they don’t have professors. I’m from a country where universities don’t work very well, but there are many people like me who are from that country who are quite good professors. But the problem is not the professor. The problem is not the machine or the desk or the whiteboard. The problem is the organization, and how all of this is put together.
In biology, this organisational and managerial know-how corresponds to the capabilities to perform complex, fine-tuned control that neurons, immune cells, and other types of cells acquire in the organism in the process of development while being surrounded by other cells (which could serve as their affordances) and having certain control goals[7][8]. This knowledge is not encoded in the genes!
Khan[1]:
The idea that a policy is a black box that the government just announces and everyone starts following it is a total mistake. A government is just one organization amongst many organizations in society. And there is an interplay going on between governments, political parties, opposition parties, trade unions, churches, mosques, the people, different kinds of agencies and forums… All of them are trying to influence this policy outcome, but also — and this is critical — implementing it on a daily basis. And if the vast majority of your organizations and society are happily violating the rules and not checking each other, then it’s not going to be an implementable policy.
These horizontal networks (other synonyms and terms related to horizontal networks that Khan uses: institutions, social structures, power structures, social organizations, legal systems, incentives, etc.) determine the ecological niches and affordance landscapes within which the players (economic entities, agents) compete and cooperate with each other.
In biological morphogenesis, the equivalents of these horizontal networks and institutions are biochemical and bioelectric networks of communication between cells, tissues, and organs. According to Levin, bioelectric networks are the medium for encoding, communicating, and negotiating the high-level morphological (also metabolic and physiological) goals of the organism that subsystems (organs, tissues, and cells) tap into[3]:
Memory (implemented by bioelectric networks or other mechanisms) is central to teleonomy as a mechanism for encoding future goal states. More generally, however, bioelectric states are a medium that binds individual cells toward large-scale goals – it underlies scale-up (Figure 5) and emergence of higher-level teleonomic individuals (Levin 2019), much as it does to create brains with emergent unified mental content out of a collection of individual neuronal cells. This is why disruptions of bioelectric communication, in the absence of genetic alterations or carcinogens, can initiate cancer in vivo - a shrinking of the size of goals from morphogenetic activity of normal maintenance to unicellular goals of maximum proliferation and migration (metastasis) (Levin 2021b); conversely, forcing appropriate bioelectric communication can normalize cells despite strong expression of oncogenes that otherwise induce tumors (Chernet and Levin 2014, 2013). The framework focused on inflating or shrinking the scale of the teleonomic activity leads directly to novel capabilities, in this case in the context of the cancer problem (Levin 2021b; Moore, Walker, and Levin 2017).
Once the high-level goal (setpoint) is achieved, the bioelectric and biochemical networks maintain the organism in homeostasis. Khan describes how the network of incentives helped to establish and then maintain the effectiveness of skill-training programmes in Bangladesh. This could be seen as maintaining a balance in “workforce metabolism” by the socioeconomy, thanks to the horizontal network of checks and balances.
Another key point in Khan’s political settlement theory is that advanced institutions (such as corporate law) are only demanded by sufficiently advanced (and hence high-capability) organisations. However, these organisations also don’t pop out without the institutions: the latter are necessary for the development of advanced organisations. Thus, socioeconomic development is necessary a gradual, spiral-like process[1]:
[…] powerful organizations which need rules for their own reproduction. They need rules for complex contracting. They need rules to raise finance in complex ways. They need to organize large numbers of people who are not known to them. They’re nameless, faceless people. So you need to have a contract-based rule of law system.
And so in advanced countries, generally speaking, most powerful organizations want a rule of law. And the difference in developing countries is that the powerful organizations are networks, which are informal patronage networks, kin networks, clientelist networks, tribal networks, religious networks, or even companies which are not that capable themselves, and their interactions with each other do not require a rule of law. And therefore a lot of their activity of lobbying, pressuring, and so on is informal. So I don’t think this is very much to do with culture or other kinds of things like that, although they do matter. It’s largely to do with the very nature of development, that developing countries have a preponderance of organizations that don’t need a rule of law. And yet, the fact that they don’t need a rule of law can stop high-capability organizations from developing.
The scale-free regulative development view on this is that economic entities embody multiscale competence architectures (MCAs) a.k.a. Quantum Reference Frames (QRFs)[2]. The more complicated and capable in predicting their environments QRFs are, the more compressed messages they can communicate across the organisational boundary, but to do this effectively they need sophisticated languages or protocols, i.e., institutions.
Also, clearly, advanced organisations need the environment of other organisations that are at least comparably advanced to cooperate and compete with via the institutional mediums.
The morphological space (morphospace) of the socioeconomy is thus the space of organisation of economic and social activity, embodied by entities, sub-entities (agents, humans, and AIs), and the patterns of their interaction: institutions, horizontal networks, corporate, national, and trade boundaries.
Khan describes how subsides work to develop the Korean industry because after the WWII the high-level networks between organisations and political figures (i.e., QRFs that are high in the hierarchy) have been decimated because they were previously “implemented with” Japanese who were all gone[1]:
[…] the real difference was the nature of Japanese colonialism in breaking up many horizontal political networks in Korea. And that was because Japanese colonialism was a very aggressive and oppressive form of colonialism. It didn’t rule through intermediate classes in its colonies, it ruled directly, and it ruled through great force and great viciousness, which is why you will not find any Korean today or any Chinese today who has a good word to say about Japanese colonialism, because it was very rough.
One consequence of that was that those horizontal networks — which businesses have with politics and other groups and unions and so on — were actually decimated in South Korea. So, the business groups that emerged in the post-Japanese period did not have the networks to protect their rents, did not have the connections with politics. So, now in the 1960s Park Chung-hee comes on and he starts trying things which are, in a sense, quite obvious. We can’t produce these things, so why don’t we give some export subsidies? Why don’t we give some protection? Why don’t we give them some low-cost loans from the publicly owned banks? Things which every developing country has tried. It’s not rocket science. It’s obvious, you can’t produce these things, your productivity is low, let’s help our businesses.
The difference was not that the South Koreans had innovated something called industrial policy. Everybody and their dog was trying it at that time. In fact, the South Koreans learned a lot from Pakistan, which also had a military government at that time and was doing exactly the same things: export subsidies, import protection, low-cost loans to large business houses, et cetera. […] Why don’t we share these rents and prevent anyone from taking it away? The South Koreans couldn’t do that, because these companies were not connected to the banks, to the politicians, and so on. And therefore, when the state gave these subsidies and they said to them, “You have to achieve these export targets,” there was no way they could protect their rents if they failed to achieve the targets.
These companies were quite happy to give kickbacks, by the way, to Park Chung-hee, to the top leaders. We know this now because there’s a lot of evidence about the corruption in the system at that time, just as we know the corruption in the Chinese system in the 1980s, when it was growing rapidly. The difference is this, if you’re Park Chung-hee and you know this company is not meeting its export target, but is willing to give me a kickback from my subsidy, do I want that, or do I want to give this subsidy to a company which will meet the export target and therefore will make lots of profits and therefore will be able to give me a kickback which is much bigger? Again, it’s not rocket science. If you’re Park Chung-hee you will say, “This is a failed company, it’s just giving me back some of my own money as a kickback. Why should I take that? I’ll close it down and I’ll shift that subsidy, that protection, to some other company.”
This phenomena looks exactly like metamorphosis. To build new advanced structures (high-capability organisation), it’s sometimes not enough just to provide the energy (subsidies): some of the existing advanced structures also need to be destroyed. This is completely obvious when we talk about “spatial” (“substance”) re-structuring (which we may think what mostly is going on in standard biological metamorphosis). The important insight here is that information exchange networks should be thought of just as physical[9] as biological cell, tissue, and organ connections, and changing them requires expending free energy.
Khan doesn’t say that his political settlement framework, or institutional economics more generally answer all interesting questions about the economy. Thus, he is against a totalising view that any respectable economic theory should be a “theory of everything”. Instead, he endorses picking up pragmatically any economic theory that already exists and seems best equipped to answer this or that question that the economist has, very much in the “all models are wrong, but some are useful” style[1]:
I don’t think it’s actually even feasible to have a political economy textbook, because it would be too big. There are too many different questions. Political economy is basically about how the world works. You can’t put that in a textbook. There are going to be different positions. Those different positions are not a problem. They are a good thing. Because nobody knows the truth. Nobody knows the right answer. But we each have our own angle, our own take. We can explain lots of things, but others can explain things in a slightly different way. And so I would say a good policymaker needs to have an awareness of different schools of thought, different methods, and so on. For some problems, the neoclassical approach might be actually quite good. For other approaches, it might be not only useless but dangerous. It might actually make things worse.
[…] I think each framework has some basic questions that they’re asking, and then they’re answering them in some ways. If you look at the political settlements framework, it is asking a number of questions about policy implementation, and the interface between organizational power and institutions, and then asking you to locate your policy in a way that will make incremental changes effective. Then everything else follows from that. You need lots of different building blocks to make sense of it. As you rightly say, you can draw on neoclassical economics, you can draw on other political economy frameworks.
[…] In the same way, if you look at Acemoglu, Johnson, and Robinson, their framing question is, “Are your institutions extractive or inclusive?” And then everything else is circling around that. Or is it a limited access order or an open access order? What are the different types of limited access orders, and what are the doorstep conditions?
[…] In neoclassical economics, the organizing principle is that people negotiate their own agreements at an individual level. The market is nothing but a set of contracts between people which are voluntarily made. You can explain quite a lot in terms of the voluntary contracts that people make, and the reasons why they can’t make those voluntary contracts. But at a deeper level, you’re not asking, “Why do people behave like that? Which contracts are enforced? Which contracts aren’t enforced?” As soon as you start asking that, you can’t just look at individuals. You’re looking at power structures. You’re looking at society. You’re looking at history.
In that sense, neoclassical economics is at the bottom of this food chain in asking about the individual contracting, which is really important and useful, but, actually, most of the interesting questions are about why those contracts aren’t enforced. That is how Douglass North began the journey of institutional economics in saying property rights and contracts may exist but they’re often not enforced.
Enforcement, I think, is the key. Enforcement takes you to all of those political economy questions which different people are cutting in different ways. But if you look at all the different political economy frameworks, just asking the same basic question, “How is the organization of society and of the collective organization affecting how individuals behave?” Whereas the neoclassical is starting from the other end, saying, “Let’s take preferences as given. Let’s take the constraints as given. This is how individuals behave.” And political economy is saying, “That’s trivial. You forgot the most important parts of the story of how we got there. How they finally make the contract is the most trivial part of the story.”
I think all of these things are connected. But I think that the really important questions about how social organization/social power affect how individuals behave, contract, the belief systems they have, how they enforce rules on each other, how they punish each other, these are historically specific questions to which you cannot have a general theory. […]
Morphological intelligence is the ability of an organism to solve problems in the morphospace, such as metamorphosis and regeneration. This is an important concept in Levin’s framework. It sounds like it would be good for our socioeconomies to be highly morphologically intelligent and be able to reinvent themselves without revolutions or special circumstances, such as a withdrawal of an oppressive colonial regime, such as Japanese regime in Korea discussed above. Note that morphological intelligence is separate from metacognition, that seems to correspond to governance and political processes in socioeconomies.
As far as I understand, it’s not yet clear what permits organisms of some species, such as frogs[10] and salamanders[11] to be more morphologically intelligent than others, such as humans. However, if it is confirmed to be in some sense strongly correlated with the complexity of the signals that are processed and exchanged by the constituent elements of morphologically intelligent organisms, this could be bad news for the complexity of human intelligence and consciousness. Some food for analogical thought in this direction: colonies of eusocial insects can regenerate large parts of their structure and “headcount”, i.e., are morphologically intelligent, although bees are sometimes considered as highly intelligent among insects, unlike ants. Though, of course, insects are vastly dumber than humans.
Note that this would not contradict the experience of great collectivist, development, organisational projects of the 20th century, such as the industrialisation in the Soviet Union which are commonly believed to be bottlenecked (or made very inefficient) by poor education of the people: the ability to organise (development) is not the same as the ability to *re-*organise (morphological intelligence)! Nor would this potential negative finding (of requiring relatively “dumb” units for achieving high morphological intelligence on the collective level) repeat the common idea in dystopian sci-fi literature that people need to be “dumbed down” to ensure control, stability, and the “common good”: actually, a morphologically intelligent socioeconomy would be agile and resilient to external challenges, rather than be extremely static and brittle, as commonly depicted in dystopias. Still, there are definitely dystopian overtones to this idea and I think it deserves serious thought. Cf. the last paragraph here[12] and Beren Millidge’s “BCIs and the ecosystem of modular minds”.
The morphological intelligence of socioeconomic structures may also depend on the scale: at least some companies, such as Amazon, Microsoft, and Adobe have historically demonstrated impressive agility and capacity to reorganise in response to changing environment, and this depended on their highly intelligent employees! So, low morphological intelligence of states may tell us more about these constructs than about the future of human intelligence and consciousness.
Hyper-developmental biology has been proposed by Levin[13] to describe the collective intelligence of a group of embryos developing together and sharing boundaries. This must be closely related to Khan’s emphasis on the distribution of organisational capabilities as the key to economic development and prosperity itself[1]:
The aim is to help the development of capabilities and organizations with capabilities that eventually results in higher levels of welfare for people, better lives.
[…] it comes back to what your normative goals of development are. And I’ve given you my normative goal. My normative goal is, I want to see a broader spread of organizational capabilities. Because I think that underpins almost everything else. Even if you wanted to fight poverty, I would ask how do I get poor people to have the capabilities to produce things for themselves and engage in activities for themselves, rather than just give handouts.
The observation that advanced organisations need (or “want”) other advanced organisations to be their suppliers, partners, and consumers that I noted in the section “Advanced organisations and institutions need each other” is just the simplest one. But I think something more insightful could be learned from biology on this topic. Or, vice versa, something could be learned from economics and applied to biology? For example, could the surprising result that cross-embryo morphogenetic assistance is more effective when all embryos are exposed to a chemical that disturbs normal embryo development than when only half of the embryos are exposed[14] be interpreted in economic terms, such as that “self-interested” embryos, like companies, are usually not eager to help their competitors when the latter are suddenly affected by some calamity?
People have started to discuss the economic analogy in the comments to Levin’s post already!
I’m primarily interested in the regulative development framework that I connected with institutional economics above insofar as what it tells us about how the AI transition of the socioeconomy, and most importantly, how to design policies that will achieve the intended developmental goals, which is the main focus of Khan’s work.
Khan, M. & Wiblin, R. (2021). Mushtaq Khan on using institutional economics to predict effective government reforms
Fields, C., & Levin, M. (2022). Regulative development as a model for origin of life and artificial life studies [Preprint]. PsyArXiv. https://doi.org/10.31234/osf.io/rdt7f
Levin, M. (2023). Collective Intelligence of Morphogenesis as a Teleonomic Process (pp. 175–198). https://doi.org/10.7551/mitpress/14642.003.0013
Levin, M. (2022). Technological Approach to Mind Everywhere: An Experimentally-Grounded Framework for Understanding Diverse Bodies and Minds. Frontiers in Systems Neuroscience, 16, 768201. https://doi.org/10.3389/fnsys.2022.768201
Friston, K. J., Ramstead, M. J. D., Kiefer, A. B., Tschantz, A., Buckley, C. L., Albarracin, M., Pitliya, R. J., Heins, C., Klein, B., Millidge, B., Sakthivadivel, D. A. R., Smithe, T. S. C., Koudahl, M., Tremblay, S. E., Petersen, C., Fung, K., Fox, J. G., Swanson, S., Mapes, D., & René, G. (2022). Designing Ecosystems of Intelligence from First Principles (arXiv:2212.01354). arXiv. http://arxiv.org/abs/2212.01354
This approach is also called methodological holism. However, Khan also recognises individualist explanations of the socioeconomic phenomena: see the “Methodological pluralism and pragmatism” section below.
Fields, C., Friston, K., Glazebrook, J. F., Levin, M., & Marcianò, A. (2022). The free energy principle induces neuromorphic development. Neuromorphic Computing and Engineering, 2(4), 042002. https://doi.org/10.1088/2634-4386/aca7de
Fields, C., Fabrocini, F., Friston, K., Glazebrook, J. F., Hazan, H., Levin, M., & Marcianò, A. (2023). Control flow in active inference systems. OSF Preprints. https://doi.org/10.31219/osf.io/8e4ra
Fields, C., Glazebrook, J. F., & Marciano, A. (2022). The Physical Meaning of the Holographic Principle. Quanta, 11(1), 72–96. https://doi.org/10.12743/quanta.v11i1.206
Vandenberg, L. N., Adams, D. S., & Levin, M. (2012). Normalized shape and location of perturbed craniofacial structures in the Xenopus tadpole reveal an innate ability to achieve correct morphology. Developmental Dynamics: An Official Publication of the American Association of Anatomists, 241(5), 863–878. https://doi.org/10.1002/dvdy.23770
Arenas Gómez, C. M., & Echeverri, K. (2021). Salamanders: The molecular basis of tissue regeneration and its relevance to human disease. Current Topics in Developmental Biology, 145, 235–275. https://doi.org/10.1016/bs.ctdb.2020.11.009
Kaufmann, R., & Leventov, R. (2023). Gaia Network: A practical, incremental pathway to Open Agency Architecture. https://www.lesswrong.com/posts/AKBkDNeFLZxaMqjQG/gaia-network-a-practical-incremental-pathway-to-open-agency
Levin, M. (2024, January 17). What groups of embryos know: Toward a hyper-developmental biology. Forms of Life, Forms of Mind. https://thoughtforms.life/what-groups-of-embryos-know-toward-a-hyper-developmental-biology/
Tung, A., Sperry, M. M., Clawson, W., Pavuluri, A., Bulatao, S., Yue, M., Flores, R. M., Pai, V. P., McMillen, P., Kuchling, F., & Levin, M. (2024). Embryos assist morphogenesis of others through calcium and ATP signaling mechanisms in collective teratogen resistance. Nature Communications, 15(1), Article 1. https://doi.org/10.1038/s41467-023-44522-2
Leventov, R. (2022). Properties of current AIs and some predictions of the evolution of AI from the perspective of scale-free theories of agency and regulative development. https://www.lesswrong.com/posts/oSPhmfnMGgGrpe7ib/properties-of-current-ais-and-some-predictions-of-the