I've come to realize that pretty much every general rule we have is only ever really true within a bounded context defined by a certain set of assumptions, and

I've come to realize that pretty much every general rule we have is only ever really true within a bounded context defined by a certain set of assumptions, and models that we think are universally applicable seem so simply because we can't (yet) look beyond their context of applicability. Even scientific models that we regard as describing fundamental laws of nature will stop describing reality accurately once you hit extreme conditions -- Newtonian physics breaks once you're dealing with relativistic contexts, for example. The human mind has tendency to generalize its heuristics, and hold certain rules to be either absolutely true or absolutely untrue, which unfortunately hobbles our ability to use our rationality to its fullest potential. People will often dismiss an entire model as being universally "false" as soon as they see a single data point that it doesn't work for, and then try to formulate a new model that's still trying to encompass everything, everywhere. Maybe we'd do better to interpret falsifying data points as helping to map the outer boundaries of the model's applicability, and formulate multiple models, each optimized for different contexts, and each regarded as valid within its bounds.