Confirming Examples and the Exceptions We Do Not See

Epigraph

the negative instance is the most powerful.

Francis Bacon, Novum Organum, Book I, aphorism XLVI, closing clause; English edition edited by Joseph Devey, 1902. Bacon is discussing the establishment of an axiom.

What this article is about

Seven successful examples can be entirely truthful while the conclusion drawn from them reaches too far. We see finished work, appreciative replies and meetings that took place; cases without a photograph or a storyteller remain outside the account. The problem need not lie in the examples themselves. It can arise when “this happened in these cases” becomes “this always happens.” This article examines that transition: who was included, which cases could disappear, and what a discovered exception actually changes. Bacon and Mill help frame questions about the grounds for generalisation. The Gospel image of the beam in one’s own eye adds an ethical demand on the person examining another’s claim. These different concerns are not presented as one theory of human behaviour. The contemporary examples are invented to make the reasoning visible, without measuring how often people err or promising a cure. Checking a bounded claim can preserve respect for people and their freedom to decline participation.

I. Where “everyone” enters the account

Imagine an open workshop where nine visitors made paper greeting cards. The organiser kept seven photographs of finished cards and wrote, “Everyone left with a completed card.” The photographs are genuine, the work exists, and its makers are pleased with what they made. Yet the caption says more than the pictures show. The pictures show seven completed pieces. The caption additionally speaks about every participant and what happened before each person left. Those additional parts require support, even if the gathering was enjoyable and the organiser’s wish to share its pleasures is understandable. Throughout this example, the workshop and visitors are invented; the numbers are part of the reasoning, rather than observations from an actual event.

Compare three possible statements: “The pictures show seven finished cards,” “Most visitors completed a card,” and “Every visitor completed a card during the session.” The first describes what is shown. The second requires the group’s size and distinct participants’ outcomes; knowing every remaining outcome is unnecessary for a majority. The third can be contradicted by one established case within the named group. It is more than an enthusiastic version of the first statement. The claim’s structure changes: several observations become an assertion about an entire class of cases. Words such as “always,” “nobody,” “any” and “without exception” make similar moves in other conversations. A writer can notice these words and ask whether the available material supports the additional reach they give the sentence.

Qualifying the claim does not diminish the seven cards. Acknowledging those achievements does not require assigning the same achievement to everyone else. We can retain the pleasure, describe what was actually seen, and leave open what remains unknown. An unnecessary generalisation may seem minor until it supports a further promise: that any new visitor will certainly finish a card, for example. An account of the past then begins to shape someone else’s expectations. The more decisions depend on a broad statement, the more consequential its limits become. We examine the move from observation to a further verbal commitment; enjoyment need not depend on it.

II. A case can disappear before it is assessed

Return to the two missing photographs. One visitor left early; another did not want to be photographed. Neither fact alone establishes whether the person finished a card. Early departure is compatible with quick completion, and declining a picture is compatible with any outcome. If we immediately classify both visitors as failures, we repeat the selection error in the opposite direction. We need available information relevant to the claim. Where it is absent, a reasonable statement describes the known pieces and acknowledges that the other outcomes are unestablished. It does not fill the gap with an invented negative story or impose a duty on the visitors to reveal what they made. Their absence from the pictures remains a limit on what the pictures tell us.

The reason a case enters an account can depend on the very outcome the account is supposed to establish. Finished cards are convenient to photograph for a post about completion; unfinished work may never enter the selection. That selection can honestly display successful results. Its design alone does not establish the exact share of successful results among all visitors. We need to distinguish an exhibition of examples from an examination of a general claim. The two can serve different purposes, and neither purpose should be quietly presented as the other. Naming the purpose in advance helps identify which missing information actually obstructs the proposed conclusion, and which information simply falls outside what the account sets out to describe.

An unseen exception need not have been deliberately concealed. It may have gone unrecorded, disappeared from correspondence, or fallen outside the convenient observation period. Discovering an incomplete selection therefore does not establish that the organiser deceived anyone. It establishes a limit to the available grounds. “How was this selection assembled?” is generally a more precise question than “Why did you hide the failures?” The first can be discussed through actions and records. The second adds a claim about motive that requires its own evidence. A speaker’s good faith and the completeness of the material are separate questions. Respect for the speaker does not remove the need to examine the material when a conclusion depends on its coverage.

III. Bacon: what will not fit the accepted picture

In Book I, aphorism XLVI of Novum Organum, Bacon describes an understanding that, once it has accepted a proposition, draws subsequent material into its support. Contrary instances can be overlooked, dismissed, or disposed of through a distinction. His argument includes a story about temple tablets commemorating people saved from shipwreck: a questioner asks about those who perished. We are considering Bacon’s construction of this example, without claiming to have independently established the historical event. The pertinent question concerns cases that cannot supply the same account of success. Increasing the number of already selected commemorations does not, by itself, answer that question about the cases missing from the display.

We can apply the question to the invented workshop as our own analogy. Ten more photographs of finished cards from another session will not establish the outcomes of the two unpictured visitors at the first session. The new photographs may document further successes, but they do not automatically close that particular gap. We need to ask which part of the claim they test. If the caption concerns the first group, information about another group has a different scope. If the caption promises a universal outcome for any session, its claim reaches further still. An account’s overall persuasiveness can diverge from each item’s precise relation to its conclusion, even when every item is authentic.

Bacon’s negative instance concerns the testing of an axiom. Here, we do not treat every unwelcome story as a decisive refutation. The form of the claim matters. One unfinished card within the group contradicts “everyone finished,” but remains compatible with “seven people finished.” It does not, by itself, calculate the success rate of future workshops either. The historical aphorism helps direct attention; the logical force of a particular case still needs to be determined separately. We infer neither a measured law about every person nor a reader’s infallible ability to recognise bias in others. Checking begins with accessible content, and its limits should remain visible even when the contrary instance makes the original claim look doubtful.

IV. An exception must concern the same claim

Suppose a visitor later volunteers that they left with an unfinished card. If the original caption really included every visitor to that session, the report concerns its subject. Before changing the statement, however, it is worth establishing which piece and which time the person means, whether a retelling alters their words, and whether the report provides sufficient grounds for the proposed correction. We cannot first narrow the group to the seven photographed visitors and then maintain that the original broad caption was true all along. The new group can define a meaningful scope for a new statement. Without an acknowledged correction, it does not retrospectively alter what the earlier universal statement said about the whole group.

Another story may sound similar while being logically irrelevant. A visitor might recall leaving a drawing unfinished at a different workshop elsewhere. That does not refute a claim about the nine people at our invented session. It might suggest a question for future inquiry, but it cannot substitute for a relevant counterexample. Keep the object, group, condition and time together: which card, made by whom, what counted as completion, and by what point? This precision protects against the organiser’s excessive confidence. It also protects against a checker’s premature satisfaction at finally finding something negative. A welcome contrary story cannot settle an issue outside its scope.

Outcome and evaluation also differ. An unfinished card need not make the gathering itself a failure. A person might have preferred experimentation, conversation or slower work. Those possibilities do not turn the piece into a completed card. We can recognise the value of participation while correcting the description of its result. If “success” was undefined, its meaning genuinely needs discussion. If the caption spoke specifically about finished work, shifting to the gathering’s general value changes the question. A generous reinterpretation helps when it is identified as a reinterpretation. It cannot preserve the original conclusion at any cost while concealing the fact that a different criterion has taken its place.

V. Mill: would we have seen an exception if one existed?

In Book III, Chapter III of A System of Logic, Mill examines the grounds of induction. In §2, he asks for reason to believe that a contrary instance would have become known before an absence of known exceptions supports a universal conclusion. This asks more than “Have I encountered an exception?” The same passage retains provisional reliance on enumeration and allows that, in special cases, it can practically amount to proof. We do not turn this selected passage into a prohibition on inference from experience. The limited distinction we use is between having no known contrary case and having grounds for believing the available way of observing would disclose one if it existed.

At our workshop, photographs were taken only after a card was finished and its maker agreed to a picture. Unfinished work could therefore remain unseen under the very rules used to assemble the selection. Its absence from that selection is not evidence that the work did not exist. A different claim would require different accessible material: voluntarily supplied information about each participant’s outcome, for instance. Proposing such information still gives nobody a right to demand a report. Where suitable support is unavailable, the caption can be narrowed. We need not resolve every uncertainty by collecting more personal information. Sometimes accurately limiting the inference meets the task while preserving the boundaries of people whose outcomes are not known to us.

Mill’s question can also be asked of a claim about agreement. If a discussion record includes only those willing to speak publicly, the absence of an objection does not establish agreement by every invited person. We may need to understand how responses were obtained. But searching for exceptions does not entitle us to treat silence as disagreement either. We know the record is incomplete with respect to unanimous agreement; that is enough to remove an unsupported “everyone.” This contemporary application is our own reasoning, rather than an experiment reported by Mill. It separates the accessible material from what the method of obtaining it leaves outside view, without deciding the unrecorded participants’ views for them.

VI. A negative report needs care as well

Looking for an exception can become a fresh form of one-sidedness. We stop examining pleasing examples but accept the first negative report because it undermines someone else’s confidence. The desired outcome has changed while the handling of evidence remains the same. A report about an unfinished card can also be misheard. Perhaps the person had not addressed the envelope while the card itself was complete; perhaps they meant a second piece they intended to make. We need not invent alternatives to defend the organiser indefinitely. We should examine a distinction when actual wording or circumstances make it relevant to the stated outcome. Otherwise, multiplying possible reinterpretations can merely postpone a correction that the evidence already warrants.

Precision does not require equal effort for every possible doubt. We can name the detail at issue, identify an available way to clarify it, and then acknowledge the limit. If someone declines to discuss their work, we do not gain a final checking result; we can nevertheless stop attributing an established outcome to them. That is already a substantive change. Retaining uncertainty can be more honest than declaring the universal caption true or false without sufficient information. Any new accusation needs its own grounds. Removing an excessive positive inference does not automatically license an equally excessive negative inference about the session, its organiser or every visitor. Each proposed conclusion must still answer to the material that actually concerns it.

Not every confirmation is independent of the others. Seven posts of the same photograph provide seven places to find it, but one pictured outcome. A review reproduced by several people does not establish several observations. This is reasoning about the composition of the evidence, rather than a statistical model of dependence. In an ordinary conversation, we can ask what is being repeated: an event, an account of the event, or a reference to an earlier account. The same distinction applies to an exception. Several retellings of one unfinished card do not make it several separate failures. What we count must correspond to the subject of the claim, rather than the volume with which a report circulates or the number of occasions on which we encounter it.

VII. The checker’s view belongs in the examination

In Matthew 7:3–5, the mote in another’s eye is contrasted with the beam in one’s own. The selected passage addresses the person judging; its continuation concerns seeing clearly and helping the other. This is an ethical and religious appeal, rather than a historical instruction on induction or a modern diagnosis of attention. Its comparison with our topic is limited to the checker’s responsibility for their own view. It does not establish that every factual objection must wait until the objector achieves personal perfection. A statement can require correction even when the person noticing its error has mistakes and limitations of their own. Those limitations should be acknowledged without deciding the statement’s truth by the objector’s character.

A checker can ask why a contrary instance seems particularly desirable to them. Yet identifying an emotion does not settle the value of the instance itself. An irritated person can point to a genuine exception; a benevolent person can overlook a consequential detail. Self-examination therefore has two distinct movements: acknowledging one’s stake, and separately examining the grounds for the conclusion. It must not become a way of excluding unwelcome testimony by judging the witness’s state of mind. Consider the unfinished-card report for its content and scope; the speaker’s warmth towards the organiser cannot replace that inquiry.

A correct objection, meanwhile, does not justify humiliation. We can change the caption, acknowledge the unsupported part, and retain a respectful description of the work that was completed. Nobody must turn every participant into an exhibit for public analysis. Factual precision and voluntary participation deserve attention together, although they are different concerns. Information voluntarily given for a private conversation may help the writer remove a generalisation without identifying its source publicly. This is our practical proposal, rather than a conclusion that the parties will certainly agree. It identifies an available correction and its limits while leaving decisions about disclosure with the people concerned. Correcting an assertion should not silently expand the checker’s authority over their lives.

VIII. When a qualification clarifies, and when it rescues a thesis

An exception may be met with “That is a special case.” The distinguishing feature can matter. Perhaps the original statement expressly concerned only people who stayed to the end, and an early departure was outside that scope. We then need to look at the original wording and its stated condition, rather than inventing a condition after encountering difficulty. If the initial claim concerned every visitor, the new condition changes it. Such a change is permissible and can improve accuracy when the writer says so plainly: the earlier caption was too broad; we now describe a narrower group and make no claim about the other outcomes. Acknowledging the change is part of what makes it a clarification rather than an unnoticed escape.

A meaningful qualification has content that can be applied to the next case regardless of whether its outcome is convenient. “Participation until the announced end” identifies a discussable boundary. “Real participants are those who succeeded” excludes in advance everything that could contradict success. It may define a collection of completed pieces, but it cannot establish the outcome for everyone who arrived. Ask whether the rule would remain the same when a new case supports the thesis rather than inconveniences it. This comparison reveals the cost of the alteration: are we retaining the subject under examination, or replacing it with a group whose very definition already contains the desired conclusion? The difference concerns the argument’s reach, without requiring an accusation about the writer’s intentions.

Time can also be qualified. “Finished during the session” differs from “finished later at home.” The latter can be a valuable and pleasing result, but it does not remove an exception to the former. Both can be described separately. Their distinction reduces the need to fight over one overall verdict of success or failure. Some claims genuinely require conditions and limits. A responsible correction makes them visible instead of pretending they were always obvious to the listener. Learning from new information is welcome. An openly revised statement allows substantive discussion to continue without obliging its author to defend the old wording to the very end. It also tells readers precisely which conclusion is now being offered for their consideration.

IX. A correction should reach the place where the inference arose

If others have already read the broad caption, private acknowledgment may not be enough to correct the expectation it created. The writer can change the caption where it is visible and briefly identify what it now says. That need not involve a detailed public examination of participants: “These are seven completed pieces; we have not established the outcomes for every visitor.” Where an unfinished outcome is separately confirmed, “everyone finished” is removed. Identifying the individual is not necessarily part of that correction. The place and extent of a correction relate to those of the earlier assertion. In this invented situation, we establish neither a legal duty to publish nor the rules of any particular platform.

Making a correction does not ensure that every reader of the earlier version will see it. We should not therefore declare their expectations automatically resolved. A bounded action can still be described precisely: the text has changed, the new account is available, and the previous promise is no longer repeated. If a particular invitation depended on it, the writer can clarify separately what a new visitor may expect from the session. We cannot know in advance whether the person will accept the conditions or how they will receive the change. Within the writer’s responsibility, the purpose is to make the message proportionate to its grounds and allow another person to decide whether the proposed participation suits them, without guaranteeing their result.

Not every old exaggeration calls for an endless review of every conversation. We can begin with a statement currently being used and an accessible decision depending on it. Examine its grounds and correct the consequential part. Where relevant new information appears, inquiry follows that information, rather than a general feeling that everything must always be doubted. Such a boundary does not excuse a known error. It helps choose a concrete action instead of an indefinite demand never to make another mistake. Responsibility becomes assessable through the revised wording and an account of its grounds. It does not depend on promising the writer’s perfection or complete control over the conclusions other people draw from what the writer has said.

X. Confirmation remains evidence when its limits are named

Searching for exceptions should not erase the meaning of successful cases. Seven finished cards remain seven finished cards. They can establish that completion was possible for those people in those circumstances. That is substantive knowledge even though it does not support a universal promise. A contrary case likewise retains its limited force: it helps remove “everyone finished,” without establishing that creative work is useless or completion impossible. Giving positive and negative evidence their respective scopes reduces the pressure to choose between unconditional admiration and total disappointment. Often the needed result is a more accurate account of what happened. It preserves particular achievements without a sweeping slogan.

Bacon, Mill and the Gospel passage address different questions. We compared attention to contrary instances, grounds for an inductive transition, and a judge’s responsibility for their own view. They cannot be named as authors of one modern method. This article’s practical proposals have not been subjected to measured efficacy testing either. Readers can nevertheless assess their reasoning: is the statement identified, does the evidence concern its scope, could the way the material was obtained hide an exception, and have both directions of inference received care? These questions leave room for disagreement. They do not require assessing a speaker’s whole person through one disputed account, or treating the checker’s preferred conclusion as the one towards which the evidence must move.

Confidence can survive a qualification by acquiring a more modest object. We can be confident that the pictured pieces exist without knowing the other outcomes. We can consider an exception compelling against a particular thesis without drawing a conclusion about every future workshop. Being able to express these different degrees of knowledge matters. A confirming example then becomes evidence with a stated limit rather than decoration for an answer accepted in advance. An unseen exception remains a question to investigate or honestly leave open. Its absence from our collection is insufficient reason to declare that it does not exist, just as the possibility of its existence is insufficient reason to declare that it occurred. A bounded account can acknowledge both limits at once.

What this means in practice

Choose one of your statements that is actually being used now. Write it down without strengthening or softening it, then mark words that extend its conclusion across a group: “everyone,” “always,” “usually,” or “in the cases shown.” Name the group, time and outcome concerned. Beside it, list only the examples you possess and how they reached you. This examination does not require searching every earlier conversation for a mistake. Its subject is one accessible statement. You can stop when its scope is clear, or leave the question open where the material is insufficient. An accurate account of that limit is already a possible result of the exercise.

Ask whether a different outcome would have appeared in the selection. If it would not, do not treat its invisibility as proof of absence. Try naming a relevant exception while checking its precise subject as carefully as the positive examples. Do not demand a personal account or evidence from people who have not agreed to participate. Sometimes replacing a universal sentence with a description of the known cases is sufficient. If a new fact actually changes the statement, acknowledge the change directly. Preserve the difference between correcting a thesis and making an unsupported accusation against its author, even when the initial claim has proved too broad.

Finish with wording you can answer for: what was observed, what the information does not establish, and what consequential new evidence would warrant returning to the question. If the earlier conclusion was public, consider an accessible correction in the same place without revealing others’ personal information. These actions are offered for voluntary reflection. They carry no promise to improve attention, resolve disagreement or prevent a future mistake. Their proposed value here concerns the precision of the selected message. Decisions about participation and the boundaries of inquiry remain with you and the other people involved; uncertainty does not itself give anyone a claim on another person’s story.

Key quotation

first cast out the beam out of thine own eye

Matthew 7:5, clause, King James Version. The verse continues with seeing clearly to help another; its context is ethical and religious.

Connections between traditions

Comparison	Connection and limit
Bacon and the invented workshop	I.XLVI directs attention to contrary instances; the photographic selection is a separate application, without historical verification or psychological measurement.
Bacon and Mill	Both selected passages raise the question of an exception. Mill III.III.§2 separately considers its possible discoverability and allows bounded use of enumeration. No shared complete theory is claimed.
Mill and silence	Known cases versus what observation could disclose helps examine an incomplete record of agreement. Silence itself is declared neither assent nor objection. The example is authorial.
Matthew and the checker’s responsibility	The beam image is compared with examining one’s own view; a religious appeal becomes neither a statistical procedure nor a ban on criticism.
Logical correction and respect	Removing an excessive universal claim is compatible with acknowledging finished work and voluntary disclosure of outcomes. This is an original practical proposal, without guaranteed agreement.

Primary sources

Francis Bacon, Novum Organum, Book I, XLV–XLVII: English edition edited by Joseph Devey, 1902. The selected aphorisms were read completely; the epigraph is XLVI’s closing clause. This edition’s English translator has not been established, and no Latin collation was performed. The tablets story is considered as part of Bacon’s argument rather than an independently verified event.

John Stuart Mill, A System of Logic, Book III, Chapter III, §2: eighth edition, Harper & Brothers, 1882. Complete §2, neighbouring §1 and the opening of §3 were read. The limited argument about discovering a contrary instance retains Mill’s allowance for provisional enumeration. Neither the entire work nor the complete theory of induction is claimed read. Historical descriptions of peoples and sciences are not adopted as present-day facts.

Matthew 7:1–6: King James Version. The complete selected passage was read; the key quotation is a clause of 7:5, with an original RU rendering through this English version. The ethical and religious address to the person judging is retained. No Greek collation or general prohibition on factual criticism is inferred.

About the authors

Andrey Zarinsh is the author and founder of United Wisdom. He explores inner transformation, the unity of traditions, intercultural dialogue, and contemporary spiritual discipline. Instagram: @andrey.zarinsh.

Maria Evilkut is a co-author of United Wisdom. Alongside her research, she seeks to test these ideas in practice and observe how principles of unity and mutual understanding work in real life between people and cultures. Instagram: @ilkutki.