Multi-level science

This is a stub. Here I will outline the limitations of hierarchical reductionism and argue for a multi-level approach.

We have a natural need to try to explain large-scale phenomena (LSP) in terms of smaller-scale phenomena (SSP). However, LSP cannot be said to "arise from" SSP unless the LSP is wholly explainable in terms of the SSP. Anything less is corner-cutting.
Explain high-level phenomena and lower-level phenomena (HLP and LLP)
It's often believed that the derivation of the HLP from the LLP will "come later". However, as we've seen, new LLP often emerge that take people totally by surprise, such as relativity. These cause paradigm shifts.
Truth and knowledge: discuss the probability thing etc. That science is a search for a model rather than a search for truth, but it's quite handy that scientists believe that it's a search for truth because they are often more motivated. Discuss frequentist probability and Bayes' theorem.
The LLP are often considered axiomatic, and scientists rarely wonder at the mystery of it all: why does gravity exist, and who put it there? They are just as valid questions to ask as "is there a god?".
Consciousness as an LLP: consciousness is not explainable in terms of anything else, although there are factors that lead to observations of consciousness (such as the existence of a brain: consciousness has not been observed in non-sentient beings).
Dismissal of observed HLP that do not tally with predictions from LLP (e.g. the Mpemba effect)