It's engineering. What resolution do you need? It's like the idiot light example. You can model the temperature with a gauge with a needle that represents the temperature of the engine coolant. Obviously old school. Anyway, if you don't care about the temperature, and you only care about temperature when it's too high, an idiot light is the right resolution. One is more accurate than the because it models the reality of the coolant temperature with a higher resolution.
And I don't get why abstraction is a part of the conversation. It's not arbitrary. Take code for example. I write web services in higher level languages that abstract the machine code that computers actually run. How long would it take me to code a complex system using just 1's and 0's versus how long it would take to write the same system in a higher level language. The abstraction only hides the complexity. It doesn't dissolve it. Someone had to think out and create each level of abstraction that leads to productivity gains.
I think the discussion of modeling is fascinating and all, but the direction is getting a bit off topic from the what I'm saying about binary thinking. I'm using it to refer to perception of reality. We build a mental model of reality, and maybe you don't like it being referred to as a model, but that's what it is. Perception is indeed a model of what we think is reality. A worldview is a model of our theory of everything. It's obviously reduced to 42. Anyway, when we reduce our perception of complex reality to simply binary outcomes, that's not an accurate model.
Even situational awareness is a sort of mental model of what's happening around us using all our senses as inputs. The concept of OODA loop is a conceptual model of human decision making.
