As a final year engineering student I was developing models on an analogue computer to solve non-linear differential equations, reading amongst others a Jay Forrester book. Seems obscure even now. What stuck in my head were ideas of non-linearity, systems thinking, feedback loops - some of the basics of complexity and chaos. This was 55 years ago! For me the world has always been more non-linear than linear, complex rather than simple with emergent properties. However I’ve learned that for most people this is far from intuitive and deeply challenging. Even ‘bright’ people - especially accountants and economists - with rigid mental models. Meteorologists have understood this for decades as have environmentalists. A few economists get it with great work at Santa Fe. Failure to understand systems and complexity is a huge barrier to progress, be it environmental, economic or social. Maybe we should be teaching the basics in high school. It’s a whole different way of seeing the world.
"BUT, in any complex adaptive system, we will always need to keep listening to the local parts...Because, they will be the first to see (feel) the fire / enemy / storm (variation)" -- this reminds me of Hayek's "The Use of Knowledge in Society" for ex:
"If we can agree that the economic problem of society is mainly one of rapid adaptation to changes in the particular circumstances of time and place, it would seem to follow that the ultimate decisions must be left to the people who are familiar with these circumstances, who know directly of the relevant changes and of the resources immediately available to meet them. We cannot expect that this problem will be solved by first communicating all this knowledge to a central board which, after integrating all knowledge, issues its orders. We must solve it by some form of decentralization. But this answers only part of our problem. We need decentralization because only thus can we insure that the knowledge of the particular circumstances of time and place will be promptly used. But the “man on the spot” cannot decide solely on the basis of his limited but intimate knowledge of the facts of his immediate surroundings. There still remains the problem of communicating to him such further information as he needs to fit his decisions into the whole pattern of changes of the larger economic system."
However Hayek claims that "in a system in which the knowledge of the relevant facts is dispersed among many people, prices can act to coordinate the separate actions of different people", and that in an increasingly complex system, the incentive structures created by reaction to changes in prices are a more effective mechanism than the ill-fated attempt to communicate information to a technocratic center node.
Of course I think this errs in its assumption of efficiency-maximization of resource allocation as the "only" guiding principle of societal organization, something which is captured in Polyani's concept of the "Double Movement" i.e. the reliable, endogenous societal reaction function we can expect against over-subjection to market-based organization. I suppose this reaction function in turn could be interpreted from a cybernetic point of view, i.e. dispersed participants transmitting information back to the central node of policy makers that effectively conveys "I don't like being subject to market forces." Dan Davies captured this well and succinctly here I think: https://backofmind.substack.com/p/the-only-message-the-channel-can
Thanks - great comment. Yes, I'm going to write about Hayek at some point in the future when I try and explain all this in relation to economies. I think the flaw in Hayek's approach was that he was talking about 'closed systems' in which all the parts are recycled within the system and (therefore) can keep creating larger than sum of parts outcomes (emergence in nature). We have created leaks in our economic system so that parts can escape and be hoarded outside the system. Anyway, I plan to explore those ideas in later posts but I'm not there yet. I also plan to talk about social economies vs material / money based economies. Social economies are really good at organising resources. Prices are just one form of feedback. But conformity / reciprocity / shame / gossip do a pretty good job too. I only started to understand this when I read anthropologists (Graeber, Harvey Whitehouse) and psychologists (Cialdini, Bandura etc;). Polanyi counterparts today would be all the heterodox economists (Steve Keen, Michael Hudson etc;) and the eco writers (especially the brilliant Nate Hagens). I'm now going to read the Dan Davies post!
"leaks in our economic system so that parts can escape and be hoarded outside the system" --- perhaps some connection there with current dynamics of structural global current account imbalances, whereby exogenously determined over-saving/under-consumption preferences (foremost by China) stymie the ability of trade accounts to equilibrate; rather than household import consumption serving as an equilibrating force via the trade account, we see savings "siphoned off" / deliberately "leaked" into the capital account and asset markets, thereby preventing current account equilibration (discussed at length by the likes of Michael Pettis and Matt Klein, and even going back to Keynes' ultimately ignored proposals at Bretton Woods in the 1940s).
Yes, very much my way of thinking. I read all Pettis’s stuff but if I ever write it, I won’t be letting the $$ off the hook either. Printing $$s as the unit of exchange for global trade has caused so many problems. As Keynes predicted it would.
Indeed....I think a lot of interpretation is inherently simple cause-and-effect / chicken-or-the egg based; either "the US is deciding to consume beyond its means and borrowing via the capital account to fund its trade deficit", or "China is deciding to consume less than it should and is foisting its savings upon the US." Rather than seeing it as a dynamic two-way relationship that requires a US propensity to leverage (either via household sector deficits and pseudo-safe asset creation pre-GFC, or public sector deficits and Treasury creation post-GFC).
As to establishing the USD as the foundation of the global system - obviously this stems from the push for such by Harry Dexter White etc at Bretton Woods (and pushback against Keynes' bancor model). Afterwards though its arguable that much of the perpetuation of that system was led by offshore actors (albeit with implicit support of the US), foremost in the proliferation of the offshore eurodollar market as covered by Josh Younger / Lev Menand as well as Perry Mehrling.
Btw Leverage Points by Meadows was my favorite thing I read in all of 2023, and I’ve linked to it several times in my posts. I love her quote:
“The world’s leaders are correctly fixated on economic growth as the answer to virtually all problems, but they’re pushing with all their might in the wrong direction.”
Thanks! Leverage Points is a stunning essay. I read it years ago and didn't really understand it. It was only when I got into systems and went back to it that I grasped how important it was.
Intriguing post Claire. I leveraged some of Meadows’s thinking early on in the development of LEED, which I believe contributed to its successful integration of sustainability into the built environment. Your extensions of systems thinking has opened some new visions for me in the political domain, which is very exciting! TY
I am a great opponent of the word "noise" (even if Kahneman is a genius). It seems innocuous, but it is similar to the word "weed" in that it makes a strong value claim without stating it. If I only value a tree's board-feet of timber, then the branches are "refuse", and the plants that surround the tree in the forest ecosystem become "weeds." Weeds have a negative connotation, growing where they are not wanted, and our desire is to eradicate them, even though the plants that intermingle with the trees are usually important components of the forest system dynamics (to keep with your paradigm).
Noise is a statistical term, and beloved of reductionists (like Nate Silver) who are simultaneously genius and incredibly dumb for the reasons you outline above. Signal is what matters to the outcome they are trying to hit, and noise is everything else. In complex systems, there is no noise, there is only signal that is either not useful or not comprehensible to the reduced model. Calling it noise implies that it is worthless, but as you articulated above, information outside the model can rapidly destroy a model.
An example: My team founded a randomized testing company in the early 2000s called Offermatica. We were one of the first organizations to run structured, randomized tests with millions of participants. In one such test, for Walmart.com, we randomly assigned participants into cohorts to see two different versions of the Walmart home page, and measured which page resulted in the highest conversion rate. Within about 25 minutes, a T-test (the best we had at the time) indicated a winner with 96% confidence. Not perfect, but far better than most studies dream of achieving. 45 minutes later the other branch was winning with 25% confidence.
We hired hedge fund quants and even consulted with a member of Princeton's faculty about how to improve our confidence calculations. What they concluded is that there was "noise" in the system that was occluding the results. What I concluded is that the noise was actually unmodeled signal, values more important than those we had isolated.
I no longer use the term noise because in a complex adaptive system, the concept is nonsense. Any information that can change the system outcome is signal, and the nature of complexity is unmodelable sensitivity.
You are absolutely right. A better word would be variation - as in it better conveys the potentiality of it all. And we totally need (latent) variation in complex systems because - as you point out - that’s what drives the adaptation. It’s also why the cult of optimisation is so misconceived. Thanks for your response.
Great post. I wonder about conflation of Meadow's "Places to Intervene..." and complex adaptive systems. My understanding is that, in complexity, one can't identify a hierarchical set of causal relationships. In other words, how do you know that "mindset" is not the outcome of the system instead of the root cause? Or both? Her model is comforting because it gives us things to look for and ways to theorize about change (e.g., if we could only change minds...), but the iceberg itself is a theoretical construct posing as a map, no?
Thanks. Good question. You are absolutely right and I have had to think about my answer for a while. But I think the answer is that although complex systems are inherently unpredictable, that doesn't mean we should simply react to them instead of trying to hypothesise and test how they will behave. Daniel Dennett has this taxonomy to describe how species interact with their environment: the lowest level is Darwinian (random trial & error learning), the next level is reinforcement learning (getting rewarded for success) the next level is Popperian (developing a hypothesis to test an intuition) and the next level up is thinking tools (what he calls Gregorian evolution, I can't remember who inspired this). He frames this as "what models of the world can I use to advance my possibility of gaining an advantage?". So I'm going to put Donella Meadows leverage points into that box. They can inform a hypothesis even if the hypothesis turns out to be wrong. Otherwise we're back to random, brute force experimentation to respond to complexity when we may be able to shortcut some of that with well worked out cognitive strategies.
Yes, this is something I have been thinking about in my own work. There are times when a method like soft systems mapping can be useful in creating hypotheses, which in turn suggest theories of change and potential interventions. This is particularly true in collaborative sense-making and the technique is easy to learn. The risk is that people begin to confuse the map with the territory, which is where I see the iceberg misapplied. Are you familiar with Dave Snowden's work? If not you may be interested in his Cynefin framework and Estuarine Mapping method, the latter being a sense-making technique for complexity.
Thanks, yes I really like cynefin. I haven’t heard of his estuarine mapping though? Sounds interesting. I put together my own checklist on sensemaking in the chaotic domain. Really just a distillation of when it’s done well (preventing meltdown in one of the Fukushima reactors) & when it’s done badly (literally every corporate scandal). If you have a link, I’d appreciate it.
Woah. I've been noticing this trend but I originally thought we were much more far off from actively celebrating INFJ personality types. Thanks for this, words cannot describe how giddy this made me feel.
Glad it made you feel that way! I'd love to say that there is more understanding about this but we need only look at what is happening in the US to see that top down control, the hero / genius myth, over-confidence etc; are still alive and well. Overturning a paradigm that's so associated with 'exceptionalism' is tough.
Very possibly I am getting confused, but— I don’t think it’s true that emergent systems must become more complex, is it? If there’s a simple way to do something and a massive evolutionary pressure towards it, you’d expect more complex processes to get less common
I think it’s a good question. My take would be this - there’s a difference between a simpler organism (one that’s been through fewer doors due to selective pressures) and a complex organism that has condensed those doors into a simpler phenotype or a more energy efficient design. Thinking about industrial design, a hammer is a pretty simple entity that doesn’t benefit much from further adaptation whereas an iPhone is an incredibly complex system that has collapsed down 50 years of learning into a palm sized device. The device is simplified because the complexity has been compressed (into technology). You could imagine the same with language. A haiku is sparse but complex. It carries considerably more meaning & condensed knowledge / learning than a sheet of A4 narrative text describing the same thing. I think I read a paper that said the human brain has fewer neuronal connections than other mammalian species - which seems counter-intuitive. Except it only has this more energy efficient design because it’s passed through more doors driven by selection pressures. If you see evolution as doors then an organism can’t go backwards. It can reorganise into something simpler and go forward. But all the prior knowledge will be condensed somewhere, available for re-use should it be needed.
As a final year engineering student I was developing models on an analogue computer to solve non-linear differential equations, reading amongst others a Jay Forrester book. Seems obscure even now. What stuck in my head were ideas of non-linearity, systems thinking, feedback loops - some of the basics of complexity and chaos. This was 55 years ago! For me the world has always been more non-linear than linear, complex rather than simple with emergent properties. However I’ve learned that for most people this is far from intuitive and deeply challenging. Even ‘bright’ people - especially accountants and economists - with rigid mental models. Meteorologists have understood this for decades as have environmentalists. A few economists get it with great work at Santa Fe. Failure to understand systems and complexity is a huge barrier to progress, be it environmental, economic or social. Maybe we should be teaching the basics in high school. It’s a whole different way of seeing the world.
I see it as a critical discipline that can help us frame so many other social & evolutionary sciences.
❤️❤️❤️❤️❤️
"BUT, in any complex adaptive system, we will always need to keep listening to the local parts...Because, they will be the first to see (feel) the fire / enemy / storm (variation)" -- this reminds me of Hayek's "The Use of Knowledge in Society" for ex:
"If we can agree that the economic problem of society is mainly one of rapid adaptation to changes in the particular circumstances of time and place, it would seem to follow that the ultimate decisions must be left to the people who are familiar with these circumstances, who know directly of the relevant changes and of the resources immediately available to meet them. We cannot expect that this problem will be solved by first communicating all this knowledge to a central board which, after integrating all knowledge, issues its orders. We must solve it by some form of decentralization. But this answers only part of our problem. We need decentralization because only thus can we insure that the knowledge of the particular circumstances of time and place will be promptly used. But the “man on the spot” cannot decide solely on the basis of his limited but intimate knowledge of the facts of his immediate surroundings. There still remains the problem of communicating to him such further information as he needs to fit his decisions into the whole pattern of changes of the larger economic system."
However Hayek claims that "in a system in which the knowledge of the relevant facts is dispersed among many people, prices can act to coordinate the separate actions of different people", and that in an increasingly complex system, the incentive structures created by reaction to changes in prices are a more effective mechanism than the ill-fated attempt to communicate information to a technocratic center node.
Of course I think this errs in its assumption of efficiency-maximization of resource allocation as the "only" guiding principle of societal organization, something which is captured in Polyani's concept of the "Double Movement" i.e. the reliable, endogenous societal reaction function we can expect against over-subjection to market-based organization. I suppose this reaction function in turn could be interpreted from a cybernetic point of view, i.e. dispersed participants transmitting information back to the central node of policy makers that effectively conveys "I don't like being subject to market forces." Dan Davies captured this well and succinctly here I think: https://backofmind.substack.com/p/the-only-message-the-channel-can
Thanks - great comment. Yes, I'm going to write about Hayek at some point in the future when I try and explain all this in relation to economies. I think the flaw in Hayek's approach was that he was talking about 'closed systems' in which all the parts are recycled within the system and (therefore) can keep creating larger than sum of parts outcomes (emergence in nature). We have created leaks in our economic system so that parts can escape and be hoarded outside the system. Anyway, I plan to explore those ideas in later posts but I'm not there yet. I also plan to talk about social economies vs material / money based economies. Social economies are really good at organising resources. Prices are just one form of feedback. But conformity / reciprocity / shame / gossip do a pretty good job too. I only started to understand this when I read anthropologists (Graeber, Harvey Whitehouse) and psychologists (Cialdini, Bandura etc;). Polanyi counterparts today would be all the heterodox economists (Steve Keen, Michael Hudson etc;) and the eco writers (especially the brilliant Nate Hagens). I'm now going to read the Dan Davies post!
"leaks in our economic system so that parts can escape and be hoarded outside the system" --- perhaps some connection there with current dynamics of structural global current account imbalances, whereby exogenously determined over-saving/under-consumption preferences (foremost by China) stymie the ability of trade accounts to equilibrate; rather than household import consumption serving as an equilibrating force via the trade account, we see savings "siphoned off" / deliberately "leaked" into the capital account and asset markets, thereby preventing current account equilibration (discussed at length by the likes of Michael Pettis and Matt Klein, and even going back to Keynes' ultimately ignored proposals at Bretton Woods in the 1940s).
Anyways, looking forward to that future post!
Yes, very much my way of thinking. I read all Pettis’s stuff but if I ever write it, I won’t be letting the $$ off the hook either. Printing $$s as the unit of exchange for global trade has caused so many problems. As Keynes predicted it would.
Indeed....I think a lot of interpretation is inherently simple cause-and-effect / chicken-or-the egg based; either "the US is deciding to consume beyond its means and borrowing via the capital account to fund its trade deficit", or "China is deciding to consume less than it should and is foisting its savings upon the US." Rather than seeing it as a dynamic two-way relationship that requires a US propensity to leverage (either via household sector deficits and pseudo-safe asset creation pre-GFC, or public sector deficits and Treasury creation post-GFC).
As to establishing the USD as the foundation of the global system - obviously this stems from the push for such by Harry Dexter White etc at Bretton Woods (and pushback against Keynes' bancor model). Afterwards though its arguable that much of the perpetuation of that system was led by offshore actors (albeit with implicit support of the US), foremost in the proliferation of the offshore eurodollar market as covered by Josh Younger / Lev Menand as well as Perry Mehrling.
This is great stuff, thank you 🙏🏻
Btw Leverage Points by Meadows was my favorite thing I read in all of 2023, and I’ve linked to it several times in my posts. I love her quote:
“The world’s leaders are correctly fixated on economic growth as the answer to virtually all problems, but they’re pushing with all their might in the wrong direction.”
Thanks! Leverage Points is a stunning essay. I read it years ago and didn't really understand it. It was only when I got into systems and went back to it that I grasped how important it was.
Intriguing post Claire. I leveraged some of Meadows’s thinking early on in the development of LEED, which I believe contributed to its successful integration of sustainability into the built environment. Your extensions of systems thinking has opened some new visions for me in the political domain, which is very exciting! TY
Good - glad it was useful!
Great survey!
I am a great opponent of the word "noise" (even if Kahneman is a genius). It seems innocuous, but it is similar to the word "weed" in that it makes a strong value claim without stating it. If I only value a tree's board-feet of timber, then the branches are "refuse", and the plants that surround the tree in the forest ecosystem become "weeds." Weeds have a negative connotation, growing where they are not wanted, and our desire is to eradicate them, even though the plants that intermingle with the trees are usually important components of the forest system dynamics (to keep with your paradigm).
Noise is a statistical term, and beloved of reductionists (like Nate Silver) who are simultaneously genius and incredibly dumb for the reasons you outline above. Signal is what matters to the outcome they are trying to hit, and noise is everything else. In complex systems, there is no noise, there is only signal that is either not useful or not comprehensible to the reduced model. Calling it noise implies that it is worthless, but as you articulated above, information outside the model can rapidly destroy a model.
An example: My team founded a randomized testing company in the early 2000s called Offermatica. We were one of the first organizations to run structured, randomized tests with millions of participants. In one such test, for Walmart.com, we randomly assigned participants into cohorts to see two different versions of the Walmart home page, and measured which page resulted in the highest conversion rate. Within about 25 minutes, a T-test (the best we had at the time) indicated a winner with 96% confidence. Not perfect, but far better than most studies dream of achieving. 45 minutes later the other branch was winning with 25% confidence.
We hired hedge fund quants and even consulted with a member of Princeton's faculty about how to improve our confidence calculations. What they concluded is that there was "noise" in the system that was occluding the results. What I concluded is that the noise was actually unmodeled signal, values more important than those we had isolated.
I no longer use the term noise because in a complex adaptive system, the concept is nonsense. Any information that can change the system outcome is signal, and the nature of complexity is unmodelable sensitivity.
You are absolutely right. A better word would be variation - as in it better conveys the potentiality of it all. And we totally need (latent) variation in complex systems because - as you point out - that’s what drives the adaptation. It’s also why the cult of optimisation is so misconceived. Thanks for your response.
Great post. I wonder about conflation of Meadow's "Places to Intervene..." and complex adaptive systems. My understanding is that, in complexity, one can't identify a hierarchical set of causal relationships. In other words, how do you know that "mindset" is not the outcome of the system instead of the root cause? Or both? Her model is comforting because it gives us things to look for and ways to theorize about change (e.g., if we could only change minds...), but the iceberg itself is a theoretical construct posing as a map, no?
Thanks. Good question. You are absolutely right and I have had to think about my answer for a while. But I think the answer is that although complex systems are inherently unpredictable, that doesn't mean we should simply react to them instead of trying to hypothesise and test how they will behave. Daniel Dennett has this taxonomy to describe how species interact with their environment: the lowest level is Darwinian (random trial & error learning), the next level is reinforcement learning (getting rewarded for success) the next level is Popperian (developing a hypothesis to test an intuition) and the next level up is thinking tools (what he calls Gregorian evolution, I can't remember who inspired this). He frames this as "what models of the world can I use to advance my possibility of gaining an advantage?". So I'm going to put Donella Meadows leverage points into that box. They can inform a hypothesis even if the hypothesis turns out to be wrong. Otherwise we're back to random, brute force experimentation to respond to complexity when we may be able to shortcut some of that with well worked out cognitive strategies.
Yes, this is something I have been thinking about in my own work. There are times when a method like soft systems mapping can be useful in creating hypotheses, which in turn suggest theories of change and potential interventions. This is particularly true in collaborative sense-making and the technique is easy to learn. The risk is that people begin to confuse the map with the territory, which is where I see the iceberg misapplied. Are you familiar with Dave Snowden's work? If not you may be interested in his Cynefin framework and Estuarine Mapping method, the latter being a sense-making technique for complexity.
Thanks, yes I really like cynefin. I haven’t heard of his estuarine mapping though? Sounds interesting. I put together my own checklist on sensemaking in the chaotic domain. Really just a distillation of when it’s done well (preventing meltdown in one of the Fukushima reactors) & when it’s done badly (literally every corporate scandal). If you have a link, I’d appreciate it.
https://cynefin.io/wiki/Estuarine_framework
Woah. I've been noticing this trend but I originally thought we were much more far off from actively celebrating INFJ personality types. Thanks for this, words cannot describe how giddy this made me feel.
Glad it made you feel that way! I'd love to say that there is more understanding about this but we need only look at what is happening in the US to see that top down control, the hero / genius myth, over-confidence etc; are still alive and well. Overturning a paradigm that's so associated with 'exceptionalism' is tough.
Very possibly I am getting confused, but— I don’t think it’s true that emergent systems must become more complex, is it? If there’s a simple way to do something and a massive evolutionary pressure towards it, you’d expect more complex processes to get less common
I think it’s a good question. My take would be this - there’s a difference between a simpler organism (one that’s been through fewer doors due to selective pressures) and a complex organism that has condensed those doors into a simpler phenotype or a more energy efficient design. Thinking about industrial design, a hammer is a pretty simple entity that doesn’t benefit much from further adaptation whereas an iPhone is an incredibly complex system that has collapsed down 50 years of learning into a palm sized device. The device is simplified because the complexity has been compressed (into technology). You could imagine the same with language. A haiku is sparse but complex. It carries considerably more meaning & condensed knowledge / learning than a sheet of A4 narrative text describing the same thing. I think I read a paper that said the human brain has fewer neuronal connections than other mammalian species - which seems counter-intuitive. Except it only has this more energy efficient design because it’s passed through more doors driven by selection pressures. If you see evolution as doors then an organism can’t go backwards. It can reorganise into something simpler and go forward. But all the prior knowledge will be condensed somewhere, available for re-use should it be needed.