What would make you change your mind?

Imagine that you help run a neighborhood workshop. A friend sends you a chart showing that a new evening program doubled attendance. You argued for that program, spent weekends preparing it, and persuaded two skeptical volunteers to help. The chart is satisfying. You forward it with a sentence: “We now know this works.”

Then someone asks what the chart counts. Visits or individual people? The same number of weeks before and after? Were the opening celebration and its guests included? Did another program close during the comparison period? You do not yet know. Nothing in these questions proves the program failed. But your sentence claimed more than you had established.

This is a hypothetical case, and it contains the central problem of the course. Good judgment requires more than being able to defend a belief once it is yours. It requires a fair relationship between the belief, your reasons for holding it, and your willingness to revise it. The test becomes difficult when a conclusion flatters your efforts or protects someone you like.

The aim is not to become the person who says “technically” until everyone leaves. You still have a workshop to run. People need decisions, explanations and commitments. We will learn to make those decisions while keeping separate what is known, what is inferred, what matters and what remains uncertain.

Begin with a claim that can be examined

“The program works” might mean that more people attend, that participants learn useful skills, that the cost is reasonable, or that the workshop reaches people previously excluded. These are related possibilities, but they are not interchangeable. Attendance could rise while learning falls. A popular program could be too expensive to sustain. A small program could serve an important purpose well.

Before searching for more evidence, write a sentence specific enough to be wrong. For example: “During its first eight ordinary sessions, the evening program attracted more distinct participants per session than the previous afternoon program.” That sentence identifies a period, an outcome and a comparison. It still leaves questions about measurement, but those questions now have a place to attach.

Notice the word “ordinary.” If you exclude an unusually successful launch event after seeing the data, you should explain why. If you include it, explain why it represents the program you intend to continue. Definitions can become a quiet way of selecting the result you prefer. A fair definition follows the question you are trying to answer, rather than whatever produces a favorable number.

Specificity also makes disagreement less personal. Someone can doubt your attendance comparison without opposing the program or disliking your volunteers. A vague claim invites a vague challenge; both parties may hear a judgment of their character. A precise claim gives them something smaller and more useful to investigate together.

A belief has more than one dimension

Consider three questions. Is the claim true? How strong are your reasons? How confident are you? Truth, justification and confidence can come apart. A lucky guess can be true. A carefully supported conclusion can turn out false when new evidence arrives. A person can be intensely confident for poor reasons.

Suppose the program really did improve attendance. That does not retroactively make your original inference careful. You forwarded the chart before checking what it represented. Conversely, if you examine the available records responsibly and later discover a hidden counting error, the error does not prove that you were dishonest. We need to evaluate both the world described by a claim and the process through which someone reached it.

This distinction matters when learning from experience. If every favorable result counts as proof of good reasoning, lucky errors will receive applause. If every unfavorable result counts as proof of negligence, careful decisions under uncertainty will be punished. A useful review asks what information was available, which alternatives were considered, and whether the conclusion was proportionate to the evidence at the time.

Confidence should communicate that proportion. “The records strongly suggest an increase” is different from “I have not checked the records, but the room felt busier.” Neither phrase must conceal its speaker's practical recommendation. You can favor continuing the program while being candid about the limits of the attendance evidence.

Knowing something does not mean knowing everything nearby

In Plato's Apology, Socrates describes examining people reputed to be wise. His encounter with skilled artisans is especially instructive: he acknowledges knowledge they possess, then criticizes the move from competence in their craft to confidence about other important matters. The passage distinguishes real expertise from its unwarranted extension. It is Plato's presentation of Socrates' defense, not a recording that gives direct access to every historical exchange. Apology, the examination of artisans.

Your friend may be excellent at organizing workshops. That gives you reasons to take their practical experience seriously. It does not establish that their chart measures distinct participants correctly or identifies why attendance changed. Respect for a person and scrutiny of a particular claim can coexist.

The same limit applies to you. Knowing how much work went into a program does not make you the best judge of its effects. Yet being involved does not automatically disqualify you either. You may know where the records are incomplete and what happened on an unusual evening. The task is to make relevant knowledge available while making the inference inspectable.

A good question is therefore narrower than “Are you an expert?” Ask what this person is well placed to know, how they know it, and whether the present claim stays within that competence. Someone may be an excellent witness to an event and an unreliable interpreter of its causes. Later chapters will develop justified trust without requiring you to become an expert in everything yourself.

State what could count against you

Before collecting more favorable examples, identify a finding that would weaken your conclusion. If attendance was counted differently in the two periods, the comparison might fail. If the increase came entirely from one launch event, it might not support a claim about ordinary sessions. If participants attended because another workshop closed, the new schedule might not explain the change.

This exercise is not a promise to abandon the program at the first complication. Evidence can weaken one claim while leaving another intact. Discovering an alternative cause might reduce confidence that the schedule caused the increase without reducing confidence that the program now serves more people. Revision can be local and exact.

Nor does every imaginable objection deserve equal attention. “Perhaps every attendance sheet was secretly forged” is possible in the thin sense that we can utter it. Without some reason to suspect forgery, it need not receive the same investigation as an obvious change in counting practice. Good judgment is responsive to relevant uncertainty, not paralyzed by the unlimited supply of imaginable doubts.

The exercise works only if a contrary result can actually matter. Suppose you say that high attendance proves success and low attendance proves the program is reaching a select group with unusual needs. Either interpretation might be defensible with evidence. But if you switch between them solely to prevent failure, you have protected the program from evaluation rather than evaluated it.

The same standard should travel with the evidence

Imagine a second volunteer proposes a morning program you opposed. Their chart shows the same apparent increase, with the same missing definitions. Would you forward it immediately? If not, what changed? The evidential problem may be identical while your practical sympathies differ.

You cannot eliminate every preference before reasoning. You can ask whether the preference has altered the standard. Requiring a rival proposal to survive every hypothetical objection while accepting your own on a vivid anecdote is unfair. It is also unhelpful: the workshop ends up comparing your enthusiasm with your suspicion instead of comparing two programs.

Symmetry does not mean treating different evidence as if it were the same. If one program has six months of reliable records and the other has a launch-night estimate, confidence should differ. The principle is that a relevant difference should explain the difference in treatment. You should be able to state that difference without referring to which side you wanted to win.

This is a practical form of intellectual honesty. It includes acknowledging favorable evidence for a position you dislike and unfavorable evidence for one you support. It does not require pretending that your commitments disappear. It requires making them answerable to reasons beyond their appeal to you.

Facts do not choose the purpose of the workshop

Suppose better records establish that the evening program attracts fewer people but reaches participants who cannot attend at other times. Whether that is a success depends partly on the workshop's purpose. Maximizing total attendance and improving access can support different choices.

No attendance chart can decide, by itself, how much weight to give access. That is a value judgment: a judgment about what matters and what people are owed. Calling it a value judgment does not make it arbitrary. You can argue that an organization promising public access should make room for people excluded by its usual schedule. Someone else can raise a resource constraint or a conflicting obligation. The reasons are ethical and practical rather than merely numerical.

Confusion begins when a speaker disguises this disagreement as a dispute about facts. “The data demand closure” may really mean “I prioritize attendance per volunteer-hour over extending access.” Making that priority visible gives others a fair chance to evaluate it. The data remain relevant; they no longer pretend to contain the whole decision.

This course will therefore resist a tempting fantasy: that excellent reasoning consists in feeding facts into a method that chooses your life for you. Methods can expose errors, improve comparisons and clarify consequences. You remain responsible for the purposes you pursue and the treatment of people affected by them.

Correcting the sentence you already sent

Return to your forwarded message. You can now write a correction that does useful work: “I overstated what this chart establishes. It appears to show more visits, but I have not checked unique participants or comparable weeks. I still favor continuing the program while we examine those records, partly because it serves people who cannot attend in the afternoon.”

The correction identifies the original overstatement, supplies the present state of knowledge and distinguishes evidence from recommendation. It does not ask recipients to reassure you that everyone makes mistakes. Nor does it bury the correction in a long explanation of your good intentions. People need to know which information they can rely on now.

There can be a real cost to correcting yourself. Someone may remember your original confidence. That possibility is one reason to be careful before making a claim, but it is a poor reason to preserve an error afterward. A reputation worth having should be connected to reliable conduct, including visible correction when reliability has failed.

The first achievement of good judgment is modest and demanding: know what you are claiming, expose the reasons, and allow relevant evidence to change the claim. Next we will ask how much evidence should change your confidence, using numbers simple enough to inspect rather than impressive enough to intimidate.

Application

A volunteer says, “Our new program must be successful because we worked harder on it, and anyone who doubts that should prove it failed.” Identify three problems, then rewrite the claim so it can be evaluated fairly.

An explained answer: Effort does not establish the desired effect. “Successful” lacks a specified outcome and comparison. The demand shifts the burden away from the person making the claim. A fair revision might ask whether comparable attendance records show more distinct participants, while separately examining learning, access and cost. Doubting the original inference does not establish failure; it identifies a reason to investigate.

Practice: Choose one claim you recently repeated. Write what it says, your strongest reason, your present confidence, and one plausible finding that would weaken it. If the claim matters to a decision, add the value judgment that connects the evidence to your recommendation.