Most students learn "the scientific method" as a tidy eight-step flowchart — question, hypothesis, procedure, conclusion — that you start at the top and finish at the bottom, usually writing the conclusion the teacher already put on the board. That list is not exactly wrong. It is just not how science actually works, and the gap between the poster and the real thing is where most of the learning lives.
This page is the short version of the reasoning that sits underneath every lab in the pack. The full essay, with the diagram of the cycle, is on the blog: Follow the data, not the conclusion.
A loop, not a line
Real science runs a cycle — the Cycle of Scientific Enterprise — that practicing scientists run, and re-run, sometimes for a career, with no fixed endpoint:
| Stage | What happens |
|---|---|
| Theory | The current best explanation — everything we think we know so far. |
| Hypothesis | A specific, testable prediction the theory makes. |
| Experiment | An honest test that could come out either way. |
| Analysis | Does the data agree with the prediction — yes or no? |
The "Yes" arrow loops the result back to strengthen the theory and adds a new scientific fact. Most students get this far. The part they miss is the other arrow.
The "No" branch is the entire point
When the data doesn't agree with the prediction — the comparison you were sure of comes back inconclusive, the test result doesn't line up with what the evidence seemed to promise — an honest analyst doesn't quietly massage the notes or rewrite the conclusion to match the theory. They take the “No” arrow and ask three uncomfortable questions, in order:
- Did I run the analysis well — uncontaminated samples, correct technique, careful comparison?
- Was my hypothesis well-formed?
- And — if both of those check out — is the theory itself wrong?
That last question is what makes science self-correcting rather than a belief system. The cleanest one-line definition we have, from Karl Popper, is that a claim is scientific only if it could in principle be shown to be false. A theory that survives an honest experiment that could have killed it is stronger than it was yesterday. A claim that no possible observation could disprove isn't a theory at all.
A real scientist designs experiments that could prove them wrong — and then runs them anyway.
Don't fool yourself
Richard Feynman put the whole habit into one sentence: "The first principle is that you must not fool yourself — and you are the easiest person to fool." His name for doing the surface motions of science without the underlying honesty — "cargo-cult science" — describes exactly the lab report where the conclusion is decided in advance and the numbers are nudged to produce it. That is the opposite of what we train.
Record what you observe, not what you expected
We don't teach this as philosophy. We teach it as habits at the bench. The most important one: students record the actual observation — the real measurement, the print that only shows a few clear points, the result that came back inconclusive — not the outcome they were hoping for. The expected outcome is the theory; what the evidence actually shows is the data. When they disagree, the data wins on the page, and we ask why. That single discipline — the honest record over the expected one — is the working heart of the lab notebook, and it is why predictions go in before results.
Why this is the point of a real lab
You cannot outsource any of this. An AI can hand a student the textbook answer; it cannot make them honestly record what the evidence in front of them actually shows, defend why their result disagrees with what they expected, or take the "No" arrow when it would be easier not to. That is exactly why the course is built around demonstrated work at the bench (see AI-proof by design) — the habits of honest science are the one thing a machine can't perform for you.
The eight-step poster teaches students to reach the conclusion. A real lab teaches them what to do when the data refuses to cooperate — which is the only part that was ever science.