THE ART OF
THE WRONG
TURN.

The grace of being wrong: falsifiability and explanatory limits

Imagine a theory that predicts that generous people give because they are generous, and selfish people refuse because they are secretly generous in a defensive way. Every possible action confirms it. Such a theory may be vivid literature; it is a poor test of behaviour. An account gains intellectual risk when it says what would count against it.

Karl Popper made falsifiability central to his account of scientific testing, although the philosophy of science has since developed substantial criticisms of using any single criterion to settle what counts as science. The useful lesson for this museum is narrower: when an explanation survives every imaginable counterexample, ask what work it is really doing. A metaphor can suggest a hypothesis. It cannot serve as evidence that the hypothesis is true.

Still, not every worthwhile sentence is an empirical proposition. A poem can change how a city feels without predicting its traffic. A moral argument can identify a value conflict without claiming a laboratory effect. The error arrives when art's evocative license is quietly converted into research's authority.

Curator's test: Say what kind of claim you are making—empirical, interpretive, fictional or normative. Then ask what sort of objection would actually change it. If the answer is “none,” do not dress it in the uniform of a discovery.

Research context: Stanford Encyclopedia of Philosophy, “Karl Popper” ↗ (interpretive source, not an endorsement of this exhibition).

The ethics of public experimentation

Iteration sounds harmless in a studio: publish, observe, improve. Outside the studio, a rough draft can be a public instruction, an exclusionary interface, a damaging joke or a misleading promise. The costs are not distributed evenly. The person who enjoys the experiment may not be the person forced to absorb its failure.

There is a meaningful difference between letting visitors try a strange but clearly fictional sculpture and silently testing a consequential process on people who cannot opt out. Consent, context, reversibility and an honest statement of uncertainty matter. A “beta” label does not cancel the need for judgment, particularly where health, money, safety, employment or reputation are at stake.

The repair is not necessarily to stop inventing. It may be to lower the stakes: use fictional data, offer a nonexperimental path, make the exit visible, seek informed participation, and avoid collecting private stories when the idea can be evaluated without them. Often the ethical imagination is simply the ability to picture the least powerful visitor at the threshold.

Curator's test: Who gets the upside, who is exposed to the downside, and who can refuse? If the experiment fails tomorrow, can the affected person recover today? These questions are a design method, not a moral decoration.

Context: W3C WAI, “Involving Users in Evaluating Web Accessibility” ↗ (guidance on meaningful evaluation); National Academies, Fostering Integrity in Research ↗ (research-integrity framework). Our broader ethical argument is editorial.

Feedback is a social gift, not a command

The creator says, “Nobody understands what I'm attempting.” The visitor says, “I can't find the exit.” Both may be correct. The creator has a model of the intended experience; the visitor has evidence about one experienced reality. Treating the second as an attack forfeits information. Treating the first as an excuse forfeits responsibility.

A useful critique usually names an observable point of friction, the consequence, and—where possible—a proposed alternative. “The turquoise type disappears against the grey on my screen” is something a designer can investigate. “Your entire personality is illegible” is merely a memorable insult. Not every judgment deserves equal weight, and an author need not accept every prescription.

The gift metaphor has its limits. Feedback can also be demanding, extractive, biased or unsafe. Asking someone harmed by an experience to educate its maker, for free, forever, can repeat the burden. The critic is entitled to boundaries. The maker is responsible for sorting signal from noise without inventing malicious motives.

Curator's test: Can you restate the strongest objection so that its author would recognize it? Can you separate what was observed from what was inferred? A generous critic and a receptive artist may still disagree. That is not a broken conversation.

Repair is a verb; apology is a sentence

A statement of regret can be necessary. It can acknowledge that a decision had consequences and that another person mattered. It can also become a performance of sensitivity that protects the very mechanism responsible for the problem. The museum therefore distinguishes the rhetoric of repair from the substance of repair.

In the case of the unreadable interface, the repair is not a moving declaration about inclusion. It is a readable, keyboard-operable and tested interface. For the invented citation, the repair is not “sources are important”; it is withdrawing the unsupported assertion, checking records, and telling readers what changed. For a metric that warped practice, it is restoring professional judgment and listening to those judged by the number.

Repair can cost prestige, speed, elegance, budget, narrative simplicity or a beloved joke. Sometimes the cost is an honest admission that no fix will recover lost trust on demand. Those costs do not make repair irrational; they make its choices visible. If improvement is always described as a free upgrade, the curator has hidden another exhibit inside the label.

Curator's test: What materially changes for the person who encountered the failure? What remains unrepaired? Whose confirmation, if any, would count? The work is not finished just because the author feels transformed.

Design reference: W3C, Web Content Accessibility Guidelines (WCAG) 2.2 ↗. Compliance is a process of evaluation, not a claim established by a well-written apology.