Scientific Experimentation Cycle¶
Procedure — instantiates Iterative Refinement Loop
Implements refinement through hypothesis, test, evidence interpretation, and revised hypothesis or design.
A Scientific Experimentation Cycle refines understanding — a hypothesis about how something works — by stating what you expect in advance, running a controlled test designed to disconfirm it, interpreting the evidence against a pre-declared criterion, and revising the hypothesis for the next round. Its defining move is the hypothesis committed before the test: the loop's integrity depends on predicting the outcome and the interpretation rule up front, so that the data can actually falsify the belief rather than merely decorate it. Unlike a pilot that deploys to decide a rollout, this cycle runs under controlled conditions to learn what is true, holding variables fixed so the result attributes cleanly to the manipulated cause. The output is not a decision to scale but a sharpened model of the world that shapes the next experiment.
Example¶
A plant-biology lab is trying to raise the yield of a bacterium that produces a useful enzyme, and growth has stalled. The team frames a hypothesis before touching a flask: nitrogen is the limiting nutrient; raising the nitrogen source will increase yield up to a plateau. The evaluation criterion is declared in advance — a statistically meaningful yield increase across three replicates versus the current medium, with contamination controls — so the result can't be reinterpreted to fit whatever grows.
They run the controlled test: identical flasks, one variable changed, everything else held fixed. Yield rises, then falls at the highest nitrogen level — not the clean plateau predicted. That mismatch is the gap diagnosis: the drop suggests nitrogen isn't simply limiting but becomes inhibitory past a point, or shifts the pH. The revision rule turns that reading into the next hypothesis (test pH-buffered media across a narrower nitrogen band) rather than declaring victory on the middle data points. Each cycle leaves the team with a truer model of what limits the organism, and the next experiment is designed against that sharpened understanding.
How it works¶
What distinguishes the scientific cycle from generic iteration is predict-to-falsify under controlled conditions:
- State the hypothesis first. Commit to a specific, falsifiable prediction and the interpretation rule before running the test, so the evidence can genuinely disconfirm it.
- Control the test. Hold everything fixed except the manipulated variable, so a result attributes cleanly to a cause rather than to a confound.
- Interpret against the pre-set criterion. Judge the outcome by the declared standard (effect size, significance, controls), not by whichever slice of data looks encouraging.
- Diagnose the mismatch, then revise. Where the result departs from the prediction, read why, and let that reading — not a fresh guess — generate the next hypothesis or design.
Tuning parameters¶
- Control tightness — how strictly confounds are held fixed. Tight control gives clean attribution but narrows how far the result generalizes; loose control is more realistic but muddier to interpret.
- Replication count — how many repeats before believing a result. More replicates beat noise and self-deception but cost time and materials; too few mistake a fluke for a finding.
- Hypothesis scope — how broad a claim each experiment tests. Narrow hypotheses are decisively testable but slow to build a big picture; broad ones move faster but are harder to falsify cleanly.
- Effect-size bar — how large a result must be to count. A demanding bar avoids chasing noise; a lenient one detects subtle effects but invites false positives.
When it helps, and when it misleads¶
Its strength is learning that survives scrutiny: a prediction committed before the test, judged against a pre-set criterion under controlled conditions, is the machinery that separates what is true from what is merely hoped. Diagnosing the gap between prediction and result — rather than celebrating a partial match — is what turns a surprising outcome into a better model.
Its failure mode is HARKing — hypothesizing after the results are known, then presenting the post-hoc story as if it had been predicted, which quietly converts an exploratory fishing trip into a fake confirmation.[n1] The classic misuse is running many comparisons and reporting only the ones that "worked," or moving the interpretation goalposts after seeing the data. The guarding discipline is to fix the hypothesis and the criterion before the test, keep the record of what was predicted, and treat a surprising result as a new question to test rather than a finding to bank.
How it implements the components¶
experiment_hypothesis— the pre-committed falsifiable prediction is the object each cycle tests and revises.evaluation_criterion— the interpretation rule declared before the test (effect size, significance, controls) disciplines what the evidence is allowed to mean.gap_diagnosis— reading why the result departed from the prediction converts a mismatch into a specific, testable next question.revision_rule— the diagnosed gap generates the next hypothesis or design, rather than a fresh untethered guess.
Scientific Experimentation Cycle tests to learn but does not run a bounded live deployment on the real population, read its measured effects, or resolve a scale/stop/modify gate under a protected floor (working_artifact_or_behavior, feedback_source, stopping_or_pivot_condition, protected_invariant) — that's Policy Pilot Cycle, which deploys to decide a rollout rather than to establish a truth.
Related¶
- Instantiates: Iterative Refinement Loop — the cycle supplies the loop's hypothesis, criterion, and evidence-driven revision.
- Sibling mechanisms: Policy Pilot Cycle · Plan-Do-Check-Act Cycle · Agile Sprint · Coaching Session · Design Iteration · Draft Review Cycle · Model Tuning Loop · Retrospective Action-Item Loop
Editorial Notes¶
Form Classification¶
Form family: Experiment, Test & Rehearsal
Rationale: Scientific Experimentation Cycle operates as an active test, trial, simulation, drill, or rehearsal that generates evidence through a deliberate attempt or perturbation because it implements refinement through hypothesis, test, evidence interpretation, and revised hypothesis or design.
Independent corroboration: The frozen evidence defines Scientific Experimentation Cycle as 'Implements refinement through hypothesis, test, evidence interpretation, and revised hypothesis or design', so its operative form is Experiment, Test & Rehearsal.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Statistics & Experimental Design
Origin pattern: Convergent development
Present-day reach: Universal
Rationale: Hypothesis, test, evidence interpretation, and revision are the canonical experimental method.
Related originating lineages:
- Data Science & Analytics — Data science, analytics, and operational monitoring supplies a parallel or contributing lineage for the mechanism's defining operation: implements refinement through hypothesis, test, evidence interpretation, and revised hypothesis or design.
- Mathematics — Mathematical modeling, proof, and abstract-structure practice supplies a parallel or contributing lineage for the mechanism's defining operation: implements refinement through hypothesis, test, evidence interpretation, and revised hypothesis or design.
- Philosophy — Philosophy of science materially articulates the logic of hypothesis revision.
Review resolution: Both blind reviewers agree that statistics_experimental_design is the primary historical origin. Explicit reconciliation of alternate_origin_disagreement, origin_mode_disagreement, domain_reach_disagreement starts from reviewer_a's mechanism-specific evidence: Hypothesis, test, evidence interpretation, and revision are the canonical experimental method. Reviewer A proposed alternates=philosophy, origin_mode=convergent, domain_reach=universal, and encyclopedia_synthesis=false; reviewer B proposed alternates=data_science, mathematics, origin_mode=single_lineage, domain_reach=multi_domain, and encyclopedia_synthesis=false. The final record retains every independently supported alternate from either review (philosophy, data_science, mathematics) without an arbitrary cap, selects origin_mode=convergent to represent the combined lineage evidence, and records domain_reach=universal and encyclopedia_synthesis=false. Present-day transfer is recorded as reach and is not treated as proof of historical origin.
Review outcome: Reconciled after independent review; high confidence.
Notes¶
[n1] HARKing — "Hypothesizing After the Results are Known," presenting a post-hoc hypothesis as though it had been predicted in advance. It inflates false positives and is the reason a pre-committed hypothesis and criterion are the cycle's integrity guarantee. ↩