What is the difference between a theory, a hypothesis and a law?
In everyday speech "theory" means a guess. In science it means close to the opposite, and this single ambiguity causes a great deal of confused argument.
A hypothesis is a specific, testable proposed explanation for a limited observation. It must make predictions that could turn out false — falsifiability is the requirement. "This fertiliser increases yield" is a hypothesis. Hypotheses are the working units of research, and most are discarded.
A scientific theory is a well-substantiated explanation of some aspect of the natural world, supported by a large body of evidence, repeatedly tested, and capable of explaining many observations and generating new predictions.
A theory is not a hypothesis that got promoted by accumulating enough support. It is a different category of thing: an explanatory framework rather than a single claim. Evolution by natural selection, plate tectonics, germ theory and general relativity are theories. They are the most robust structures science produces, not tentative ones.
A scientific law describes what happens, usually as a mathematical relationship, without explaining why. Newton's law of universal gravitation describes the force between masses precisely; it does not say what gravity is. General relativity — a theory — provides the explanation.
This is the point most often misunderstood: a theory does not "graduate" into a law. They answer different questions. Laws describe; theories explain. A theory sits above a law in explanatory scope, not below it.
Laws can also be superseded or bounded. Newton's law is not wrong, but it is an approximation that breaks down at high velocities and strong gravitational fields.
What follows practically:
"It's just a theory" is a category error when applied to evolution or climate science.
Theories can be revised, and are — but revision requires evidence that the framework fails to accommodate, not merely disagreement.
A model is a third thing: a simplified representation used to make predictions, which may be deliberately incomplete.