Falsifiability
A claim is only worth something if you know what observation would refute it.
Falsifiability, borrowed from philosophy of science, is the property of a claim that some possible observation would show it false. A claim no observation could contradict carries no information, however confident it sounds.
In engineering the idea has an unusually concrete form: a test is worthless until it has been seen to fail. A test written after a fix, added to a green suite, proves only that the suite is still green. Revert the fix, watch the test go red, restore the fix — now the test is a real net. Skipping that step is how a suite accumulates assertions that were never wired to anything, which is the same species of problem as Vacuous Truth.
The rule generalises to any guard. Before trusting a lint rule, a schema check, or a CI gate, delete the thing it protects and confirm the gate reds. Guards that cannot fail read as guarantees, and a guarantee that is not one is worse than a known gap — see Silent Failure.
It also disciplines prose. "This cast is required because the checker demands it" is a falsifiable claim: delete the cast and typecheck. If you have not run that experiment, the honest sentence is that the cause is not established. Writing the confident version instead is the trap described in Plausible Mechanism.
See also5
Hand-picked in the note itself — the neighbours worth reading next.
Root Cause Analysis
Escalating past the visible symptom until you find the layer that actually produced it.
Method21 connections
Assertion
The statement in a test that must hold, and the only part that can actually fail.
Testing & Verification9 connections
Coverage Gate
A build check that fails when coverage falls below a threshold, and what that pressure produces.
Testing & Verification14 connections
Ground Truth
The measurement that arbitrates, as opposed to the artifact that merely reports.
Method30 connections
Exhaustive Claim
A statement of the form "the last one" or "nothing else does this" — load-bearing, and only as good as the search behind it.
Method13 connections
Related2
Nearby in the graph rather than deliberately chosen. Looser, sometimes surprising.
Linked from16
Notes elsewhere in the wiki that reach for this one.
- Adversarial ReviewAgents & Language Models
Reviewing work with an explicit mandate to refute it, from a perspective that did not produce it.
- AssertionTesting & Verification
The statement in a test that must hold, and the only part that can actually fail.
- Character EncodingSystems & Tooling
The mapping between characters and bytes, and the source of a specific family of invisible bugs.
- Coverage GateTesting & Verification
A build check that fails when coverage falls below a threshold, and what that pressure produces.
- Exhaustive ClaimMethod
A statement of the form "the last one" or "nothing else does this" — load-bearing, and only as good as the search behind it.
- GlobSystems & Tooling
Wildcard pattern matching over filenames, similar enough to regular expressions to mislead.
- Ground TruthMethod
The measurement that arbitrates, as opposed to the artifact that merely reports.
- HallucinationAgents & Language Models
Fluent, confident output that is not grounded in anything real.
- Plausible MechanismMethod
A causal explanation that was inferred rather than tested, and reads as more rigorous for being specific.
- RegressionMethod
A previously working behaviour that a change has broken.
- Reproducible CaseMethod
The smallest set of steps that reliably produces a fault, and the unit of real debugging.
- Root Cause AnalysisMethod
Escalating past the visible symptom until you find the layer that actually produced it.
- Silent FailureMethod
A failure that produces no signal, so the absence of an error reads as success.
- Test-Driven DevelopmentTesting & Verification
Writing a failing test first, making it pass, then improving the code with the test as a net.
- Truncation BiasMethod
Reading a truncated result as if it were the whole result, so every counterexample is invisible.
- Vacuous TruthMethod
A statement that holds only because its subject set is empty — a green result that measured nothing.