If the world was such that the past was not a reliable guide to the future, then past and future would not be treated equally by the patterns of substantive knowledge in physics. This might be a lot of fun, but predicting outcomes from interventions would be near impossible.
If the world were such that underlying physical regularities varied depending on where you are, like the laws of the countries, then you'd have to significantly adapt how you set up interventions to achieve identical outcomes depending on where you are.
"When faced with a difference in a process look for changes in the environment rather than changes in the patterns"
This has served as a useful heuristic, with wide-ranging experience showing that the patterns do seem to be unchanged, or invariant, not depending on where or when you use them. That is, they are unchanged if you are translated in time or space. This a translational symmetry – something is unchanged when you move about.
For example Hooke’s law does not change over the years, nor is it unreliable if tested in another country.
It's the patterns that are physics. We're looking for things that do not change to serve as reliable guides to action.