[arkades]

"I'm sorry, but no. Observation comes first, then hypotheses, then prediction, then verification. Explanations go from proposed to confirmed, but they are certainly not the genesis of scientific knowledge.

Not really. I mean, I appreciate the old "Scientific Method" card, that doesn't actually describe how it happens, pretty much ever. All observations happen schematically: by the time you're in grade-school, you've already gotten a basic grounder in Science: The Lies to Children Edition, and all future observations and learning are elaborations on the groundwork. Observations without theory happen at somewhere around, I don't know, age 2. By the time you can speak, you've already started the process of sense-making. In reality, observation always happens against the background of pre-existing theory. It's turtles, all the way down. (This is hardly ground-breaking: Thomas Kuhn was discussing this half a century ago).

"The phenomenon itself must come first, otherwise all you have is the fitting of facts to theory."

Which is precisely what we do. We collect facts, interpret them (that is, use them as a representation of underlying trends or relationships), and then collect more facts to see if our generalization holds true beyond the initial dataset that we used to generate our ideas. Science is absolutely about fitting facts to theory, and then collecting more facts to test that theory, and then elaborating that theory based on those new facts. It's entirely circular (for at least the last few centuries): no modern scientist sets out gathering observations without any pre-existing theory in his head.

[pradocchia]

Right, so you are describing what Kuhn termed the period of "normal science": theory in hand, collect new facts, fit facts to theory, repeat.

HOWEVER, that theory-in-hand rests on an earlier paradigm shift, where previously anomalous observations were re-integrated via some new conceptual framework that better accounted for all observations, not merely the conformed ones that scientists had focused on in the prior period of normal science.

So yeah, perhaps in the day-in, day-out existence of professional scientists, most work is fitting facts to theory. However, that theory exists because facts come first, because at some point in the past, mounting factual evidence overwhelmed the theoretical biases of an earlier generation.

Do I really need to spell this out for the HN crowd? For you, who reference Kuhn?

[lutusp]

> no modern scientist sets out gathering observations without any pre-existing theory in his head.

Or without the intent to shape a new theory, a new, testable inductive generalization, as when Einstein shaped special relativity. That's my favorite example because it happened in a theory vacuum, no pun intended. There must be a theory to inform the research, or the research must lead to a testable theory, or both.