Like if you want to mathematically model what happens in a pool table hall when somebody strikes a ball with a great deal of force... by the time you get to the six or seventh bounce you are going to have to start to take into account the position and movement of people standing around the table watching it. The airflow, the vibration of them moving, the relative gravitational forces, etc. It all matters all the time.
And the problem only gets worse the larger the scale and longer the timelines.
Like if you want to manage a economy.
It is tempting to want to look at "things from a high level" and imagine that all the details just kinda average themselves out. So it isn't necessary to figure out the behavior of each individual in a national economy. It should be possible to simply plot out the results of their decision making over time and extrapolate that into the future to come up with meaningful policy decisions and 5 year plans.
The problem is that that doesn't work. Because all the details matter all the time.
Also the very act of making policies causes changes in the behavior economy in wildly unpredictable manners. Every individual actor involved is going to change their behavior and decision making based on your decision making, which then changes the behavior and decision making of every other individual, etc etc. In a endless fractal involving billions of actors, since your national economy is not isolated from the forces of every other economy and visa versa.
Also trying to make targets out of measurements and indicators tends to destroy the value of the measurements and indicators. Meaning that by setting policies you are destroying the information you are basing your decisions off of.
So you can't collect enough information to make good decisions. The information you receive is already obsolete by the time you get it. And the act of trying to create rules and policies based on the information you do have tends to destroy what little value it has left.