I've been reading "Playing with movement - Hargrove 2019". This book told me about "complexity science", which is a field which supposedly studies complex systems and it does it not by reductionism, but by looking at the system as a whole. It seems that some key concepts used in complexity science are: complex system, emergence, adaptivity, nonlinearity, self-organization, constraints, attractors, feedback loops. This book pitched an idea that with many complex adaptive systems, if you want the system to achieve a certain goal, it's bad to specify a specific plan for it and instead it's better to specify or build constraints under which the system, being adaptive and all, will achieve the goal on its own, supposedly in a more optimal way. Examples:

  • System: my body. Goal: get healthy by physical exercise. Specific plan: do strength training by doing specified sets of reps, using specific instructions on how to position my body. Alternative (constraints): live in such an environment that a lot of physical activity will happen naturally.
  • System: economics of a country. Goal: make it productive. Specific plan: central planning of what to produce, in what quantities and how. Alternative: free market economy, the organizations will behave however they want, and hopefully the whole system will act in an optimal way.
  • System: my body. Goal: eat healthily. Specific plan: write down a schedule of specific meals for the next year and eat them. A potential problem is I might miss something and that might lead to a lack of some nutrients. If instead I eat whatever I want, I will probably crave those other nutrients and hence I'll go and eat them. Obviously, this has downsides.

This idea seems plausible and very important. However, I've never heard of complexity science before. I've been following all kinds of links about complexity science trying to figure out what it is, what fields it's related to and what subareas it has. Also, I want to know how much of it is correct and how to apply it in the real world. Please help me figure out these questions. So, I guess I want a primer. Except I know that if I find a primer written by a complexity scientist, it'll claim that complexity science is the greatest invention ever. Instead, I want a sceptical primer which specifies the domain of applicability of complexity science and the domain of applicability of theory of complex adaptive systems. Below, I list some more specific questions.

  • It seems complexity science is also sometimes called complexity theory (different from computational complexity theory) and is somehow related to systems theory and complex systems science. Systems theory is supposedly a synonym for cybernetics. And maybe systems theory is related to systems thinking. How are all these areas related and to which does the study of complex adaptive systems belong?
  • Complexity science is somehow related to a bunch of math areas: chaos theory, differential equations, fractals, and cellular automata. But at the same time, it seems related to sociology, the study of human organizations, and has some applications to medicine and biology.
  • I want to read or study something to better understand the idea described earlier. I want to understand when that idea is true and when it's false. Any recommendations or thoughts? Is this idea also studied in other scientific fields?
  • In general, how awesome is complexity science? Is it almost entirely correct like math or physics? Or is it full of incorrect information like psychology and other social sciences? It is at least not 100% crackpottery, since some books are published by Princeton university press and Oxford university press.
  • Complexity science seems useful for rationality. Why isn't it popular on Lesswrong?

And here are some links and sources I found about complexity science:

New to LessWrong?

New Answer
New Comment

3 Answers sorted by


Sep 26, 2020


I've studied the subject a bit - one book, a set of lectures, and generally keeping an eye on things from SFI for a few years (they're a reasonably-well-known institute specifically devoted to complex systems).

General summary:

  • Complex systems theory is mostly applied math, and the people doing it are mostly physicists and applied mathematicians. The results are almost always technically correct, although not necessarily interesting - it's rarely obvious how well the models generalize.
  • As of today, "complex systems theory" is really just a bunch of mostly-unrelated math topics (especially chaos, dynamical systems, networks, and the like) which seem to involve some qualitatively-similar behavior. The central question of the field is characterizing that behavior, and figuring out why it pops up in so many systems.
  • The "qualitatively-similar behavior" in question is basically: when systems are large (i.e. many interacting parts) and have no special structure (nonlinear, random network structure, etc), they tend to be chaotic but have some "simple" large-scale statistical behavior. What complex systems theorists would really like is a fully general theory of what large-scale behavior results from what systems.
  • That theory doesn't exist yet. It may not exist at all. So for now, the field looks like a bunch of people studying mostly-unrelated models, without much in the way of unifying results. They do at least share some common tools, though - statistical mechanics, scaling laws, and phase changes are major themes.
  • (Also, there's a bunch of popsci books/articles which are largely unrelated to the actual math, as one usually expects from such things. Just ignore those.)

In short: it's a field where a lot of people think there's something interesting to be found, but it hasn't been found yet.

How useful is their vocabulary and their set of ideas to understand the real world, not as a professional researcher, but just as a rationalist?

Two answers to this. First, complex systems theory has very little vocabulary/ideas of their own. What few they do have (e.g. "emergence") are not especially useful yet, because there's not much substantive theory backing them up yet. They're basically just pattern matching, but we don't yet know whether the patterns correspond to any common underlying structure. Second, because complex systems theory draws from so many other areas, it is useful as a gateway into a bunch of other fields. In studying the vocabulary/ideas of "complex systems theory", you'll mostly be studying tools from other fields which aren't really specific to complex systems, but those tools are really general and interesting. It's especially useful in that complex systems theory tends to draw on the most general tools from other fields, so you'll end up learning tools with quite wide applicability.


Sep 26, 2020


I can only give a very partial answer, focusing on the negative side. I hope someone more informed on the positive side can add their perspective.

"Complex systems" has always seemed to me to be a non-apple, and many of the words used around it, like "emergence", are synonyms for "magic". Real things are done under the umbrella of the term, but I see no coherence in the area that the umbrella covers. It is, however, a fertile field for generating popsci books.

BTW, "complexity theory" is also the name of a branch of mathematics that studies what resources (usually time and space) are required to solve computational problems, like sorting a list, or finding a 4-colouring of a given map. This complexity theory has nothing to do with the "complexity science" you are asking about. I mention it only to avoid a possible confusion.

Similar here. Reading the title, thinking "explaining how exponential complexity is worse than linear will be a piece of cake". Reading the text, thinking "okay, how is this different from cybernetics?"

Even Wikipedia just says "study of complexity and complex systems", and then points towards computational complexity and systems theory. Wikipedia has its flaws, but...

Even among the resources linked as "some courses/primers/introductions", half of them do not contain words "complexity theory" or "complexity science". Which makes me doubt:

It is at least not

... (read more)
Complexity theory seems to be a rarely used synonym for complexity science. Although, it's used in the title of one of the books. I've mistakenly used "complexity theory" too many times in my question. I've just fixed that. Regarding some courses/primers/introductions, I found them by following links and citations from other complexity science related things and by using connectedpapers.com to find similar books/articles, not just by googling complexity science. (Except for the classcentral courses, but those talk about dynamic systems, chaos, and fractals, so they are probably also on-topic) So they most probably support the idea of complexity science. You can also Ctrl+F "emerg" to find the use of the word emergence in them and see that they talk about complexity science. To be clear, I've checked Understanding complexity by Scott E. Page - the book contains lectures and is published by Princeton university press and Complexity: a guided tour - Mitchell 2011 published by Oxford university press and they definitely talk about emergence, self-organization and contain other vocab associated with complexity science.

Suppose Xs are some small parts of a big thing and Y happens in the big thing due to how Xs work and how they interact together. I think people say "Y is an emergent outcome of Xs doing whatever it is that they do" means "Y is an outcome of Xs doing whatever it is that they do and for human it would be difficult to figure out that Y would happen if they just looked at Xs separately".

Yes, this is a motte of "emergence". The problematic part is when you turn the concept of "despite understanding the rules of all little pieces, it is still difficult for a human to predict some patterns of their interaction" into a noun, and then kinda suggest that it refers to a mysterious thing that many difficult-to-predict patterns have in common, and that there is a way to study this mysterious thing itself, and by doing so gain insight (going beyond "yep, complex things with many parts are often difficult to predict") into all these difficult-to-predict patterns. In other words, if you make it seem as if understanding of e.g. gliders and biological evolution (two examples of "emergence") allows you to better predict stock markets (another example of "emergence"... therefore, they all should have something in common, and you can study that). Quoting Eliezer: (source)


Sep 26, 2020


Hey, I've become interested in this field too recently. I've been listening to the Jim Rutt show which is pretty interesting, but I haven't dived into it in any real depth. I agree that it is something that we should be looking more into.

I won't pretend to be an expert on this topic, but my understanding of the differences is as follow:

  • Systems theory tends to involve attempts to understand the overall system, while complex systems are much more likely to have emergent novel behaviour, so any models used need to be held more lightly/it's more likely that we have macro level trends that are there and we just don't know why
  • Cybernetics is mostly about control systems (classic example is the thermostat). Feedback loops are an important part of systems theory, but they are just one particular tool.
  • Regarding the breadth of applications, we can model dynamics in formal mathematical situations, then try to claim that similar dynamics occur in actual physical systems
2 comments, sorted by Click to highlight new comments since: Today at 6:24 PM

Steven Strogatz mentions:

A few weeks ago, someone was asking about "complex systems". Is there anything to it, or is it just a buzzword? How does it compare to agent-based modeling, chaos theory, systems science, etc.? This concise survey by Mark Newman answers those questions: arxiv.org/abs/1112.1440

My basic and primitive understanding (From Taleb Etc.) is that there are a few ideas that are important; such as ergodicity; at least when trading in a complex market and why you should follow Kelly criterion. Also, fat tails etc. 

But when I did the research like you, it seemed quite sparse.