Make it simple, but not simpler than that – Einstein

The human mind is made of concepts or models. These are the sum of how things appear to us not how they actually are. This distinction has already been made by Platon in his allegory of the cave and is an important one to remember when operating in the world.
Any kind of model inevitably falls short of being a completely accurate depiction of reality – it is just a model after all. Total accuracy is impossible. But models are nonetheless extremely useful because they allow us to think coherently and efficiently. In fact models are how we think.
Some of these concepts are more simple, some are more complex – all of them try to be a useful depiction of reality – not necessarily an accurate one. Given the limited mental capacity of the human mind, eg. only being able to hold 7+/-2 units of information in the mind at once, it is highly critical to point out that the best model is not necessarily the most accurate or complex. The best model is in most cases not the most simple either of course but the most useful.
Usefulness in this context relates to degree of accuracy needed to remain actionable and not to miss out on the the significant aspects of the matter at hand.

Metaphorically speaking, one could compare the necessary level of accuracy of a model to be useful to the graph of a rational function.

Complex models can always get more complex, however they ultimately and inevitably fall short of a perfect description of reality, just like the graph of a rational function never quite reaches the asyndetode.
Equally every model already is a description of some sorts of reality – after all concepts are based on reality. Therefore there is always a tiny degree of accuracy given.
Usefulness now cannot be found at the extremes obviously, rather it has to be found in the “bend” of the graph so to say – where complexity meets usefulness.

Working with this model then elicits the question of “how far should we go up the curve to have all the relevant aspects covered?”

Naturally, the answers will vary from issue to issue. Climate change requires much more complex models than does making spaghetti, for example.
Obvious as it may sound, it might be argued that failing to keep this principle of the usefulness of concepts in mind is responsible for many major planning fallacies and false action taking. Missing out on the possibility of different scenarios for example has proven disastrous on the geopolitical level for many countries. Likewise are some modern tools such as computers on the verge of overcomplexity in regards to their usefulness, especially of average consumers.

Conclusively, this meta-model seems to be essential in handling our complex world and especially in handling it with the right degree of simplicity and/or complexity. Keeping in mind that a model is only as right as it is useful. The metaphor of the rational function might prove useful in simplifying this (meta-)model in itself.