Reading the OP, I think science still works by the same principles it always did. And people still attempt to apply it to reality in the same way they always did.
It's not like people could solve every problem before quantum mechanics or relativity came along. Most of the formulae are okay in labs to figure out fundamentals and work in some systems outside the lab, but many systems are too chaotic to predict even if you know the formulae describing the parts. For a simple example, ask a scientist to predict what one of those swings with chains supporting a seat would be doing by the fifth swing if I gave it a push, or if it will rain on this day next month. Ask them to predict turbulance. And best of all, ask them to predict a system where the components think and choose what they will do.
I think social sciences make sense, but in these cases you need to be even more aware of the uncertainties and assumptions in what you are predicting. You are more or less making a more educated guess than you would if you had nothing to go by.
But that's all science. It's just rules to predict things. If you imagine that things you have never seen and could never see look certain ways or that the universe is your mathematical formulae, you've lost sight of what you are doing. You know these things by the evidence you have of them, and anything more that you see is a metaphoric projection of this evidence into the world with which are are familiar, usually to make thinking about them more natural.