Not sure about this, and if it does then to a much lesser extent. The properties of matter are self-organizing. Life started out (relatively) simple and got more complex over time.
True, but that's not related to my point.
All science does is assume that reality (or part of it) is explainable.
I think you may be confusing description with explanation.
"How reliable is experience?"
Truth:= Whatever model of interrelated definitions accurately predicts the outcomes of experiments (or happenings in general).
This view is called model-dependent realism. Though I just made up that definition, so it could be better but should work for our purposes. Notice how it fits for mathematics as well as science in general.
In this view, your question becomes meaningless. Science is realiable because it produces results, not because it has some claim to an obscure absolute truth beyond our experience.
An explanation
is a claim to obscure absolute truth beyond experience. Not a few scientists (e.g. Penrose) are enthusiastic Platonists who believe that mathematical truth somehow inhabits a higher plane of experience and science is a way to get glimpses of that plane.
That's an entirely mystical point of view.
Accepting that science is descriptive - and only descriptive of experience, not of reality - isn't.
There is absolutely nothing in science that is remotely explanatory, because the concept of "explanation" doesn't apply - except as narrative and mythology.
Maybe two models, relying on different definitions, might do equally well for certain experiments. That's the wave/particle 'problem' in QM. The thing is, there doesn't need to be one absolute truth, it's beyond our experience anyway. So why not be pragmatic and accept both models if they produce consistent results?
Because you end up with models that contradict each other and are clearly incomplete, at best.
When people first discovered non-Euclidean gemoetry, they thought of it as a cool mathematical trick. Only later, when Einstein used it for general relativity, did it turn out to have bearings on nature. It's remarkable how armchair logic comes to find out things about the universe. It shows that there must be something right in our methology.
No, all it proves is that if you throw enough random math at the wall, some it sticks.
This is
the point here. There's nothing miraculous about the process. Grad courses are math mills producing more or less random models that are sort of consistent, maybe, perhaps, creating random predictions that are likewise.
The predictions are tested. Occasionally, some of them match experience.
But this is just a random walk through a theoretical landscape. Genius-levels of insight happen occasionally (q.v. Einstein) but that's not how most science is done.
And you're still missing the key point which is that science assumes that intrinsic human perceptual and analytical biasses don't matter, and it's possible be truly objective in spite of them.
In fact it's assumed, for completely arbitrary reasons that cannot be justified logically or empirically, that a professional consensus of description and experience is identical to
objective understanding.
In reality we have no idea how wide experience can be. It's possible we don't even know how wide human experience can be.
So assuming that human experience and logic are somehow magically pre-calibrated for objective truth, and can do more than make predictions about human experience and logic, is a position that can't be justified.