It’s not about good systems being simple and bad systems being complex. The real world is complex. With great cost and energy we can temporarily induce a simple state to facilitate mass production. It’s a fragile state. Entropy will return, usually by tumbling into chaos.
Complexity is the only “real” state. All the others are a state of mind in response to a situation. All the world is one networked supersystem. We model bits of it in our heads only so we can cope. You can be in any domain/state: it’s a relativistic state, relative to the observer. Dave Snowden says it is the default state, but one could also say its the underlying state. After all, reality is a Complex Adaptive System that we stir into chaos, or we wrangle (temporarily) into known complicatedness or clarity. The underlying system is always a complex adaptive system in the real world (except for technologically induced bubbles of simplicity with no humans in them, e.g. computer networks, or approximations of Simple where the system behaves long enough or often enough for us to pretend, e.g. most clerical processes). If we aren’t investing energy to simplify our interaction with the system, to move up into Complicated or Clear (or Chaotic), then we are interacting with the natural system, hence complex.
All reality is complex, as in a Complex Adaptive System. Local states of low entropy can be induced by lifeforms, by the energy capture of life. Humans take that to the extreme with highly bounded, protected bubbles, local systems, that are temporarily able to be worked with as Complicated or Clear. It’s a state of mind of the observer in the situation, a way of interacting with the system which is forced to behave itself for a while. That Complex domain is when we are closest to the complex reality, not seeing the world in another domain, not constraining it to behave as another domain, not investing energy to create a bubble of low entropy.
This might seem too abstract and theoretical. Certainly there’s a community out there who wrestle over the philosophy in impenetrable discussions. I’m not really interested in the theory, either way. Cynefin is the best known model and that’s all I care about. It’s a means to communicate not a hill to die on. The posturing, from all sides, bores me. The people I work with don’t GAF, but the picture helps them grasp reality, and sense-make.
All work is networked. Only by zooming in can you find anything like a nice tidy stream. And only locally can you constrain the system to simple Clear behaviour.
it must be defined and repeatable to get up to that highest state of being automated. It’s a subset of Clear. Bosses like manual work to be that way too, so that we can industrialise it, dehumanize it, Taylorism. Any defined repeatable work depends on a stable “production line”, value stream. We have to literally or virtually bolt it down, constrain it to work predictably. That takes effort and money to build a factory or a process. We are moving up that diagram, reducing entropy by expending energy, to create a simple predictable state so we can do standardized work. All of that work, that time and money, is an investment in temporarily reducing entropy within a highly bounded local area. We need those islands of Clear to do repeated work, but they’re not a natural state and we shouldn’t try to make the whole world like that.
It costs money and takes effort to create and enforce controls, to build fail safes, increases reliability, build security etc. You are creating a bubble of simple in a complex world. You have to build strong bounds to contain it and protect it. It’s a fragile thing, vulnerable to external events or internal failures. it’s not normal. We have to do it so that we can have repeatable defined – i.e. cheap – work, but we shouldn’t ever assume we can keep it that way nor that we can make the rest of the world equally contained and simplified. It might be an actual factory, or clerical work on a computer, or automated software deploys. They’re all artificial conditions forcing the world to behave itself… for a while. Sooner or later something breaks, entropy wins. For the most spectacular examples of that, search “steel mill accident” on YouTube. They go from Clear to Chaotic in fractions of a second.
Seems like nobody died in this one. I prefer those.
I’ve pondered whether AI can extend automation beyond the Clear domain to work that is not well defined. I don’t have the expertise to answer it. Since generative AI regurgitates human knowledge it could hypothetically play the expert role in the Complicated domain. So long as we stay within what is Known, AI can possibly deal with what is Unfamiliar. It seems likely. I’m waiting to see it.
It’s easy to surrender to the futility of our existence. I’m watching my young adult sons wrestle with that. Then we get on with creating bubbles of low entropy because there is value – and calm – in them for as long as we can maintain them, until the inevitable disruption. The struggle against entropy is real.
Read more about that diagram here