If time is a true dimension rather than just something we model as a dimension, it's not unreasonable to think that outcomes in the present could be influenced by constraints that exist in the future - if the "universe" function must be valid according to some constraints at all points in the time dimension, "past" states that lead to invalid future states will never occur.
The idea of invalid universe states is of course purely fun conjecture, but this same concept also maps to probable vs improbable universe states as well. If a future state is improbable (in the sense of a Bayesian prior according to some underlying distribution of energy in the universe) that might cause the universe to evolve in a way that seems improbable at the moment by moment level, but is actually the most probable sequence of events when integrated over the entire duration of the universe.
If time is a true dimension rather than just something we model as a dimension, it's not unreasonable to think that outcomes in the present could be influenced by constraints that exist in the future - if the "universe" function must be valid according to some constraints at all points in the time dimension, "past" states that lead to invalid future states will never occur.
The idea of invalid universe states is of course purely fun conjecture, but this same concept also maps to probable vs improbable universe states as well. If a future state is improbable (in the sense of a Bayesian prior according to some underlying distribution of energy in the universe) that might cause the universe to evolve in a way that seems improbable at the moment by moment level, but is actually the most probable sequence of events when integrated over the entire duration of the universe.