School teaches the scientific method as a rigid list of steps. It isn't. In practice it's a fluid, continual loop of observing, guessing, and testing. Michael Shermer lays out the cycle cleanly: induction (form a hypothesis from the data you have), deduction (make specific predictions from it), observation (go gather the data the hypothesis tells you to look for), and verification (test the predictions against new observations to confirm or kill the idea).
The vocabulary matters more than people think. A hypothesis is just a testable statement that accounts for some observations. A theory is a hypothesis that's been well-supported and well-tested. A fact is a conclusion confirmed so thoroughly that it's reasonable to give it provisional agreement — note provisional; nothing is locked. A law is the subset of facts so solid they can be written as a tidy equation or two. But the real load-bearing piece is the hypothesis, and it earns its keep with two properties: it has to be falsifiable, and among the survivors, the one with the most predictive power wins.
For any of that to work, science has to aim at being objective — claims that other people, anywhere, can independently check. That's exactly where mysticism fails: it rests on personal experience no one else can validate. And it's why science's real superpower isn't any single discovery but its ability to self-correct. Follow the core methods and false claims get overturned with time, generation by generation. It behaves like an immune system — or like evolution, with useful ideas kept and bad ones pruned.
Could the whole enterprise go rotten — could a pet theory harden into dogma the way it does in religion? Only if science had a central authority to enforce it, and it doesn't. It's a sprawling, global, competitive community, and the private ambitions of thousands of independent scientists are precisely what keep it honest. The only way to truly corrupt it would be total control, which seems about the least likely outcome of all.