A conviction is most useful when its owner can say what would change it. On confirmation, what Popper can and cannot teach a manager, uncertainty versus indecision, and a three-sentence exercise in intellectual accountability.
There is a question that can make an apparently straightforward conversation unexpectedly difficult: what evidence would change your mind?
It sounds almost rude. It suggests that a position, however carefully argued, should remain open to correction. Yet the question can also be an expression of respect. It treats an opinion as something substantial enough to be examined rather than a badge that must be protected.
We usually ask people what they believe and why. We ask for supporting examples, credentials, principles and experiences. Less often do we ask what observation would make them abandon or revise the belief. Without that second question, the difference between reasoned confidence and an emotionally defended preference may remain invisible.
A conviction can be sincerely held and poorly tested. Sincerity does not settle the matter.
The asymmetry of looking for confirmation
Suppose a manager believes a team is missing deadlines because employees are insufficiently disciplined. Every late delivery reinforces the view. An employee asks for more clarity about priorities, and the request is taken as a lack of initiative. Another employee finishes early, and the result is credited to exceptional personal effort. Both outcomes are interpreted inside the same original theory.
This is a hypothetical example, but its reasoning trap is familiar: evidence is not merely collected; it is classified through the explanation already in place.
What would change the manager’s mind? Perhaps a careful analysis showing that delays cluster around client approvals and cross-team handoffs rather than individual performance. Perhaps a process redesign improves reliability without changes in staffing. Perhaps the opposite: improvements fail until skill gaps are addressed. The point is to identify observations that distinguish competing explanations.
A useful belief should risk being wrong in a way that can be noticed.
What Popper can and cannot teach a manager
Karl Popper’s philosophy of science is often reduced to a slogan: try to falsify your theory. His actual argument was more careful. The Stanford Encyclopedia of Philosophy explains his emphasis on testable, risky predictions, while also noting that real-world methodological decisions are complicated by measurement error and the difficulty of accepting a counterexample as decisive.1
A management hypothesis is not automatically a scientific universal law. A prediction about a market can fail because the execution was poor, the environment changed or the sample was small. One failed experiment may not settle a contested organisational question. Still, the underlying discipline is valuable: identify what would count against a claim before seeing the result, rather than explaining away every uncomfortable outcome afterward.
The right lesson is not that every belief must collapse at the first anomaly. It is that no belief should become immune to anomalies by definition.
The difference between being uncertain and being indecisive
A person who names uncertainties can appear less decisive than someone who speaks in absolutes. But uncertainty and hesitation are different states.
Consider a decision to enter a new market. One leader says the opportunity is compelling because the market is large. Another says the opportunity is promising if a specific customer segment has a costly, unmet problem, if distribution is accessible at an acceptable cost, and if early purchasers value the differentiation. The second position contains more conditions. It also offers more ways to learn.
The first sounds confident because it cannot yet be inspected. The second is more useful because its assumptions can be questioned, measured and revised.
Decision-making under uncertainty does not require perfect knowledge. It requires making explicit which facts matter, what evidence is available and what risks will be accepted. Sometimes the rational decision is to act with incomplete information while establishing an honest review date. Sometimes it is to wait because the missing fact could reverse the choice.
The point is to separate the strength of a commitment from the infallibility of its supporting beliefs.
A conversation with evidence should have a stopping rule
AI tools make it easy to generate arguments for and against almost any position. That abundance is valuable for exploration, but it can create an illusion of independent corroboration. An answer from a second system is not necessarily a second source of evidence; models may rely on overlapping information or replicate assumptions embedded in the question.
The previously drafted AI Opinion Spiral manuscript identifies this as a risk of accumulating fluent explanations without a named decision standard. Its CLEAR protocol proposes clarifying the decision, defining evidence rules, actively seeking disagreement, assigning accountability and recording a revisit trigger.
That protocol is a useful reminder, but the deeper issue is broader than AI. We can keep consulting colleagues, articles and spreadsheets until we find a description that allows us to avoid responsibility. The purpose of evidence gathering is not to eliminate uncertainty forever. It is to reduce uncertainty enough to act or to state why acting would be premature.
Before seeking one more opinion, it is worth asking what new kind of observation could actually change the decision.
Some beliefs are not laboratory hypotheses
The question has limits. A moral commitment cannot always be revised by finding a new percentage in a survey. A preference for a certain kind of writing is not false because most readers prefer another style. The decision to respect another person’s privacy does not become indefensible merely because publicity would be more profitable.
Our lives include empirical beliefs, interpretive judgments, values and tastes. They require different forms of reflection. It would be a category mistake to demand an experiment for every preference or to pretend that a scientific result alone resolves a dispute about human dignity.
Yet even here, the discipline of intellectual openness has a role. A moral position may remain firm while our understanding of its consequences changes. An aesthetic judgment can deepen through new encounters. An interpretation of another person’s action can soften when context appears.
Changing one’s mind is not always changing one’s values. Sometimes it is learning how to apply them more responsibly.
A small exercise in intellectual accountability
THE THREE-SENTENCE EXERCISE
For any consequential opinion, write three sentences before entering a debate. First: I currently believe … because … Second: The strongest serious alternative explanation is … Third: I would revise this view if … The third sentence should name an observation, reliable source, threshold or pattern that could matter, rather than a demand that other people simply agree with the author.
For example, a team might believe its slow delivery is mainly a capacity problem. The alternative is that rework caused by ambiguous briefs consumes capacity. The revision condition might be a measured decrease in turnaround time after the brief-and-approval process is redesigned, while staffing remains stable. The point is not to invent a universally correct causal test. It is to design a comparison that can teach the team something.
A further question improves the exercise: what evidence would make the opposing view weaker? Fair thinking is not the ritual of doubting ourselves more harshly than others. It is the willingness to apply comparable standards to competing explanations.
Keep the right to remain unfinished
People sometimes defend old beliefs because revising them feels like disowning the person who once held them. But continuity of character does not require continuity of every conclusion. A life devoted to learning ought to leave room for ideas that become more precise, less certain or finally untenable.
There is no shame in saying that a claim was the best explanation available at the time and that the available evidence has changed. There is also no obligation to perform public doubt about every serious conviction. Intellectual humility is a practice of proportion, not a costume.
The question I return to is not whether I can imagine being persuaded by anything. It is whether I have left enough room for reality to tell me something I did not expect.
The answer will not always arrive as a dramatic reversal. More often it may be a quiet edit: a deleted certainty, a narrower claim, a new condition in a sentence that once felt complete.
Perhaps that is what it means to take one’s own thinking seriously.
ABOUT THIS ESSAY
Published 20 November 2026.
Citations & references
- Stephen Thornton, ‘Karl Popper’, Stanford Encyclopedia of Philosophy (substantive revision 31 July 2026). https://plato.stanford.edu/entries/popper/
Connections
THOUGHT ATLAS
Philosophy