For sure it is a wicked problem which education would help immensely with.
In my own experience as a systems Engineer, we don't distinguish systems as complicated, always complex. In reality, most physical systems are complex, rather than complicated (In the way defined in your refs).
It states there that complicated systems can be constructed by designing the parts individually, and putting them together to form the desired system.
That is seldom true in actual practical systems, though that approach is often tried, it almost always ends in tears, especially if the system is expected to fly.
More usually, we will create a monolithic system model beginning with a single block of the overall system with all inputs, outputs, stakeholders, and stakeholder constraints defined.
We then progressively refine the monoblock towards the subsystems into which the overall system might conveniently be decomposed (The parts).
So the specification for each part comes out of the System Design Process, rather than the other way round, as described in the definition of a complicated system.
The difference we see there might be a classical difference between the way traditional academic establishments describe systems theory, and the way it is described in industry (I've worked in both).
Simulation is used throughout the industrial model based system design process, to confirm the system at every stage in development performs as expected for the stage of development.
The reason for this is that the dynamics at the inputs and outputs of any system effectively feed both ways through every element of the system, with each element having a characteristic response, which is completely different from the response in isolation.
So every practical system fits the definition of a complex system, as far as I can see.
Even something as simple as a multiple link pendulum, we need to know how each part performs as part of the functioning system, rather than in isolation, to be able to specify the dimensions of each part in such a way that will meet the behavious as a system, required by the stakeholders.
I would argue this also applies to social systems.
We each respond to what we sense in our environment, but when the stimulus is something sensed by our group, our response might be completely different than expected from our individual characteristics.
A classic example is the phenomenon of flexible bridges oscillating when more than a critical number of people or animals are walking on it, at some point the group response hits a resonant frequency of the bridge, and the whole thing starts to swing, sometimes with a period of several seconds, and there is not a thing any individual can do about it. No matter what we do, if we are on such a bridge under those conditions, our individual response keeps adding to the group response powering the swing.
(I experienced this personally as one of a crowd on the lower suspenced bridge across the river Duro in Porto, Portugal during a new year celebration there, pretty hilarious, but it was actuall really dangerous and scary)
Simulation at every opportunity is our way of virtually testing the system before it is actually physically built from fully specified physical components and tested.
This whole process is formally known as Model Based Systems Engineering ("MBSE"), which is my speciality as an Engineering Consultant.
Personally I think we can educate folk with the only solution, a solar powered infinite donation based economy, by providing access to the System model under design, with enough supporting information for all to understand how it works, and for all to perhaps participate in the System Design, since all are stakeholders of the system.
Sooner or later, that design will appear in the public domain, produced by some party or another, perhaps multiple, where all will be able to evaluate, before committing to a final implementation, I believe.
Perhaps some views of it might look something like this:
