Degrees of Confidence: Speaking about Knowledge without Absolute Claims

Epigraph

A wise man, therefore, proportions his belief to the evidence.

David Hume, An Enquiry Concerning Human Understanding X.87. Exact English sentence within his argument about experience; not a modern probability calculation.

What this article is about

“The parcel has arrived,” “It will probably arrive today,” and “I really hope to receive it today” convey different things. One reports an event, another predicts the future, and another expresses a wish. Conversation easily blurs them, especially when reassurance or an appearance of confidence is wanted. This article considers how to describe knowledge proportionately to its grounds while retaining the ability to act and revise what was said. Hume, Locke, and Peirce discuss belief, probability, and settled opinion differently; bounded passages are considered without merging them into a modern theory. Book, parcel, and shared-arrangement examples are invented. No scientific confidence scale or calculation for every event is offered. We distinguish felt conviction, grounds, a claim’s scope, and decision costs. Cautious speech can be precise about what is established, what is inferred, and what remains missing for a further conclusion.

I. Claim, confidence, and desire concern different things

“I am certain” may describe a feeling, the quality of a check, or readiness to make a decision. Clarify which is meant. Strong conviction alone does not say how information was obtained. A well-supported report may be spoken quietly and anxiously; perfectly confident delivery can accompany error. This distinction offers no way of judging people by tone. It directs attention toward a claim and its grounds. If the question is whether a parcel arrived, establish what was observed and where it is confirmed rather than judging whether the answer sounded sufficiently convincing. The information needed concerns the event, not a performance of confidence.

In Sections III–IV of The Fixation of Belief, Peirce distinguishes doubt, belief, and passage toward settled opinion [1]. Satisfaction with a belief does not itself make the belief true. This is philosophical reasoning, rather than a clinical anxiety model or a contemporary reliability test. Separate two outcomes: someone has stopped doubting, and a statement has obtained appropriate support. One may accompany the other without replacing it. “Say it confidently” may therefore fail to improve a report. Another request may be more useful: explain what has been established and how you learned it.

Desire can be expressed directly. “I would like the book to be here today” need not become “It will definitely be here today.” Interest and hope remain without adding knowledge of an unobserved event. When somebody schedules their time around our words, the distinction has practical significance. Say that delivery is expected but receipt has not been confirmed. This is neither coldness nor a ban on supporting another person. Support may consist of clear information, an available next step, or willingness to wait together. It does not require promising an outcome beyond what we know and can do, even when such a promise would sound comforting.

The distinction does not require doubting every word. If a book is before us and the question concerns that accessible copy, we can report seeing it. This does not establish that somebody else received their copy or that the contents were verified. Confidence has a subject. Naming it precisely allows a clear answer without extending it to everything connected with the object. “The copy is on the table now” can be sufficiently definite. “Every issue with the book is resolved” needs other support. A modest claim’s scope often makes speech more accurate than adding “absolutely” to a broad formulation whose separate parts have never been examined.

II. Grounds determine the strength of a report

In Enquiry X.87, Hume proposes proportioning belief to evidence and discusses differences between uniform and variable experience [2]. The passage belongs to his argument about miracles; its entire religious polemic and historical language of infallible experience are not adopted here. Retain the bounded question: are two expectations equally supported? A checked report about this delivery and a habit drawn from previous weeks supply different grounds. A familiar course of events can support a prediction without establishing what happened today. The distinction concerns the support available for the particular event, rather than a general judgment that one person is always more dependable.

Naming grounds is more useful than increasing confidence’s volume. “I saw the received box” describes one access route. “The sender wrote that it was dispatched” describes another. The second can be correct and important, but dispatch differs from receipt. Neither automatically establishes the contents. There is no need for a ladder placing sight above records or testimony in every case. Compare grounds with the subject. A record may date an event more accurately than memory; observation may establish an object’s present location better than an older entry. Confidence should follow that substantive relation, rather than the prestige of a preferred kind of information or a blanket ranking of its sources.

A weak ground for one claim does not strengthen its opposite automatically. Tracking without receipt confirmation may reflect an update delay, another handover route, or unfinished delivery. Emotional convenience cannot select among them. Describe the information available and a next way to clarify it. “Receipt is not yet confirmed in the accessible record” differs from “The parcel is lost.” The further a conclusion extends beyond the material, the more connecting steps it needs. Caution matters here because it preserves the particular unestablished part, rather than because tentative phrasing is inherently superior. Later information may clarify that part without cancelling everything already known about dispatch.

Grounds may be strong enough for a bounded answer and insufficient for a general recommendation. Several timely deliveries constitute known experience when dates and records are checked. They do not promise that outcome for every person under all conditions. Conversely, one delay need not establish universal unreliability. Clarify the scope: which cases were examined, what is known about them, and why they bear on the present expectation. No statistical calculation from these everyday illustrations is being supplied. The subject is the language of inference. Stating limits preserves experience without granting it knowledge of every possible case, including circumstances nobody has observed or appropriately compared.

III. Probable differs from both known and arbitrary

In Essay Book IV, Chapter XV, Locke distinguishes knowledge from accepting a probable proposition; XVI’s opening connects assent’s degree with its grounds [3]. These terms belong to a particular philosophy of knowledge, rather than a ready modern statistical scale. They allow a middle position: grounds may support a conclusion without final certainty. It need not be declared a fact or equated with arbitrary fantasy. Explain why the account is considered and what a stronger report would require. This permits comparison among grounds rather than making all uncertain answers equally suitable merely because none has received a claim of absolute certainty.

Suppose the book usually stays in the common room, but nobody checked today. Its presence there may be reasonably expected if conditions have not changed and past experience supports that expectation. Someone might have borrowed it. If collection is needed now, clarify its location rather than debate whether “probably” is an acceptable word. The word gains meaning through grounds and subject. Without explanation, it may mark a supported account, a speaker’s caution, or concealed lack of information. “It usually stays there; today’s location is unchecked” therefore tells a reader more than an isolated qualifier whose practical meaning they are left to infer.

An account acquires substance when supporting grounds and significant obstacles are named. The usual book location is known, but the room was rearranged yesterday. That does not establish that the book moved; it weakens simple reliance on the former pattern. “It may be in the room, but the former shelf is unconfirmed” makes the revision intelligible. We are not calculating a number from two everyday facts. We show why confidence has changed. This helps someone decide whether to search there or first ask a person involved in rearranging. Qualified language serves their choice rather than displaying the speaker’s sophistication in being cautious about everything.

Unknown can also be a precise answer state. Without suitable information about the book’s present location, say “I do not know where it is now.” This preserves knowledge of ownership or an earlier sighting. Completeness does not require creating a convenient explanation. Offer to check an accessible place or ask the owner, while retaining their right to decline unrelated private questions. A precisely named unknown makes a substantive next step easier to identify. “Nothing is clear” may conceal established parts. An honest answer keeps those parts beside the gap instead of allowing one uncertainty to erase every available ground or every useful distinction about the situation.

IV. Confidence words need explanation

“Established,” “inferred,” “possible,” and “unchecked” can help, but labels do not perform the check. “Established” needs a subject and appropriate support. “Possible” does not make an account likely merely because it is imaginable. “Unchecked” describes work with information, rather than necessarily a claim’s falsity. These words assist when connected with content. Instead of a compulsory universal scale, offer a brief explanation: “I see the copy now,” “The sender reported dispatch,” “This extends prior experience,” or “Receipt information is absent.” Readers receive both the speaker’s confidence and grounds they can evaluate themselves, rather than a label that asks them to accept somebody else’s certainty without knowing its source.

Do not mechanically weaken every statement with “perhaps.” If an accessible circumstance is established, continual softening can obscure it. “The notice apparently says eleven” is odd when the text is clear and the question concerns only the written number. “The meeting will apparently begin at eleven” concerns a future event and has different scope. A qualifier’s appropriateness depends on its subject. Speech can be definite about an observed entry and cautious about execution. There is no need to choose between unconditional severity and universal fog. One answer can contain different degrees of support for different parts when their distinction is clearly explained to the person receiving it.

“As far as I know” helps when it states an access limit rather than quietly protecting the speaker from every responsibility. Add the place or time checked. “Today’s accessible messages show no time change” gives readers an intelligible scope. If another known channel might contain an update, mention it. A limitation should not conceal significant information already available. When relaying the report, preserve its conditions rather than forwarding only a strong conclusion. Otherwise a carefully bounded initial statement becomes someone else’s confident claim with an evidential scope much wider than what was actually examined. A short condition may be essential to the report’s meaning, not merely optional caution.

Caution does not require listing every conceivable exception. For an ordinary question, retain limits that change evaluation or the next step. A visible book does not require discussion of every philosophical possibility of perceptual error. A delivery prediction does need to preserve the absence of receipt confirmation. Proportion makes qualification useful rather than endless. Ask what a reader might mistakenly regard as established if the condition disappeared. If a significant part of the decision changes, keep it. If the qualification merely complicates speech without changing the subject, shorten it. Clarity comes from selecting the relevant limit, rather than maximizing the number of words that express doubt about what is being reported.

V. A number does not create a measurement

A report sometimes sounds more persuasive after a percentage is added. “Nearly certain” becomes a precise number without new information. That percentage might describe a personal estimate, a particular model’s output, or a proportion of observed cases. These subjects should not be blended without explanation. When giving a number, state what it represents and how it was obtained. Without a calculation, do not imply one. A qualitative estimate can identify grounds and limits. This does not reject numbers: quantitative reporting has value when its meaning is clear. A notation’s apparent precision does not itself establish the accuracy of what the notation is supposed to represent.

Imagine a record in which nine of ten known dispatches arrived on the expected day. That proportion can be described if the data were checked. It does not automatically provide today’s delivery probability. Ask whether conditions are comparable, why that set was selected, and what was excluded. No forecast is calculated here; the numbers are invented to distinguish a set’s description from a claim about a new event. Correct arithmetic can accompany an overextended conclusion. Preserve what was counted and which further inference is proposed. If the connection’s grounds are unchecked, a precise proportion should not hide uncertainty about extending it to a different circumstance.

An estimate of effort may also look precise without adequate grounds. Somebody promises to find the book within minutes despite not knowing its present location or room access. State an expectation as an expectation: its assumptions, conditions, and what would change the estimate. Experience of comparable searches may support a bounded forecast, but wishing to finish quickly does not reduce uncertainty. The other person should understand what they can reasonably plan around. A range can be more useful than one number, yet two endpoints do not make it justified automatically. Explain their origin or say that a preliminary estimate has not yet been made.

Locke’s historical “probability” and Hume’s reasoning about experience supply no formula for these everyday percentages. They should not certify invented precision with a philosophical signature. Specialist quantitative reporting needs the relevant method and explanations. Here the distinction between counting, inference, and personal conviction suffices. When readers ask “Why that number?”, answer substantively or acknowledge a limit. “This is my preliminary estimate under these conditions” is clearer than a number resembling an established law. Information becomes more useful for shared decisions when linguistic precision matches the available grounds instead of decorating them with accuracy the check never produced.

VI. Revision can answer new information

Confidence can change without dishonesty. Someone saw the book on the table in the morning and later learned it was moved. The observation does not become false because its present location differs. The answer about now has changed. If the morning sighting concerned another copy, the earlier claim needs correction. Distinguish changed circumstances from discovered error. Both allow a new report, but the report means something different in each case. Preserve enough history for others to understand why today’s answer differs and which earlier information remains usable when discussing the past. Revision need not either erase a valid observation or protect an invalid identification.

New testimony does not automatically take priority merely because it is new. A person may relay an older report later, or a newly found entry may carry an incorrect date. Examine content and origin as in the initial check. Stronger relevant support can change the conclusion; a significant unresolved doubt can lower confidence temporarily. Irrelevant information need not alter a bounded answer. Revision differs from continual wavering through a named reason. Readers benefit from knowing whether new information appeared, wording became more precise, or the original answer exceeded its grounds and that overstatement has now been acknowledged. These are different changes despite all producing a revised sentence.

Revision sometimes means narrowing scope. “Books always return to this shelf” may have grown from a few familiar cases. After an exception, say that this was usual within the observed arrangement and the present location needs checking. Past experience need not be abandoned; remove the word promising more than it established. This changes content, rather than merely replacing “always” with “usually.” Identify the new limit. If it remains unclear, describe the known cases directly. Speech gains grounds in what was actually examined without maintaining a broad formula simply to preserve the speaker’s appearance of consistency. A narrower accurate account can retain more useful information than either total defence or total rejection.

Distinguish wanting doubt to end from obtaining grounds for ending it. Peirce describes belief’s stability and attachment to it [1]; he does not thereby prescribe endless doubt or any comforting account. In our conversation, define an adequate stopping point: appropriate confirmation was obtained, or accessible inquiry’s limit was stated. Unchanged checks need not then be repeated to display diligence. A substantive new discrepancy still deserves attention. An answer’s stability comes from subject and grounds rather than a promise never to revise an opinion. It can be used now while retaining an intelligible condition under which future clarification would become necessary.

VII. Decisions need not pretend complete certainty

Action is sometimes needed before a final answer. People may agree to search the common room while the exact book location remains unconfirmed. The decision can be appropriate, permitted, and manageable without making the assumption an established fact. Say both parts: “The location is unknown; we will begin with the accessible room.” Participants then understand the available knowledge and proposed action. Artificially increasing confidence is unnecessary to permit a small reversible step. Yet being able to act does not remove consequences. More consequential decisions need suitable grounds and competence; this ordinary illustration does not substitute for the requirements of those decisions or establish a universal rule for acting under uncertainty.

People can accept different costs of uncertainty. One may check the room while passing; another would travel specially and needs confirmation. The first is not automatically more reasonable for being willing to try. Decision conditions differ. Clarify information, offer another step, or allow refusal. Discussion becomes more precise when grounds’ quality is separated from each person’s action cost. That cost includes time, access, effort, and possible inconvenience to others, rather than only a chance of success. Participants who agree on what is known may still decide differently about whether it suffices for their own participation in the proposed next step today.

A promise concerns another subject: what someone undertakes. “I will check the accessible room and report” differs from “I will certainly find the book.” One can lie within a chosen action; the other depends on an unknown location. Do not weaken a promise until its obligation becomes unintelligible, or extend it into a guaranteed outcome. Name a reporting time, access conditions, and the response to an obstacle. This preserves responsibility for the work without claiming control over the entire event. If conditions change, inform anyone relying on the promise and offer a clear correction rather than silently disappearing from the arrangement.

Confidence in a decision does not entitle someone to treat another person’s doubt as a vice. They may see a missed question or decline the proposed commitment. Discuss grounds and conditions instead of demanding complete trust. A shared resource also involves other participants’ rights; personal conviction does not replace agreement. Without a common task, differing opinions need not require joint action or a final dispute. “We assess this differently” can preserve a boundary. It does not make every statement equally true. It recognizes that knowledge, trust, and the right to choose one’s own undertaking concern different questions whose connections should be explained rather than collapsed into a demand for obedience.

VIII. Precise speech leaves room for correction

A confidence report helps when another person can understand its structure. What is claimed, what supports it, which limit remains, and what action is proposed? These questions require no long report in every conversation. One sentence may suffice: “Dispatch is confirmed by a message; receipt is unchecked; I will review the accessible record this evening.” Knowledge, uncertainty, and a promise connect without blending. The reply can be calm and definite. Avoiding an absolute word need not weaken the position. Definiteness comes from a precise subject and intelligible grounds rather than volume, a confident persona, or a demand that questions immediately cease.

If an earlier statement exceeded knowledge, correct it substantively. “I called delivery complete, but had checked only dispatch” identifies the error. A fortunate eventual outcome does not justify the overextension: accidental correctness does not establish the original grounds’ quality. Correction need not become a confession of being unable to know anything. Name the specific mistake, retain established information, and clarify the current state. If another person already relied on the earlier wording, communicate the relevant correction. Responsibility concerns how information is reported and amended, rather than a promise never to face uncertainty again or an attempt to preserve an unblemished image of confidence.

Precise reporting can develop without conversation becoming a contest in caution. “Possible” and “I do not know” do not automatically make someone attentive; “I know” and “confirmed” do not automatically make them arrogant. Examine subject and grounds. Clarity sometimes requires firmly naming a checked detail and sometimes stopping an overextended conclusion. An answer’s strength lies in fit rather than manner. Participants retain an adult position: they may check, question, decide, or decline. Neither endless doubt nor unconditional faith in the speaker is imposed. Information is offered with intelligible boundaries instead of becoming a test of loyalty or an opportunity to rank people by their style of expressing confidence.

Degrees of confidence allow different parts of an answer to have different support. They do not replace subject knowledge or create a final scale for every person and event. Historical philosophers help pose questions; a present conclusion still requires present, appropriate grounds. Conviction can coexist with a recognized limit; a decision with an unestablished future; correction with retained knowledge. Such speech promises neither agreement nor protection from every failure. Its sufficient outcome is a report another person understands substantively: this is what we know, this is why, this is where checking ends, and this is the available next step without claiming absolute certainty over the whole situation.

What this means in practice

Choose one everyday report that mixes an event, expectation, or wish. Avoid a task requiring specialist decisions about health, law, or safety. Separate what happened, what is inferred, and what is desired. Name time and subject so the claim does not quietly expand.

Identify grounds for each significant part: observation, an accessible entry, somebody’s report, or an extension of prior experience. State what was actually checked. Do not increase confidence merely because of a strong feeling, an attractive account, or repeated copies of one source.

Rewrite proportionately to the grounds. A checked detail can be stated firmly while another question stays open. Explain a percentage or range’s origin; without calculation, do not imply measurement. If no quantitative estimate has yet been made, say so.

Choose one permissible next step or stopping point. Separate decision from knowledge and promised work from guaranteed results. Obtain agreement where another participant is needed. Do not demand unrelated private information to confirm a preferred account.

Finish with a brief report: what is established, what is inferred, what would change the answer, and what you undertake. Correct a mistaken earlier sentence specifically. This original exercise is neither a demonstrated psychological method nor a universal probability calculation. A more accurate reply and intelligible boundary suffice; final certainty about the entire event is not required.

Key quotation

Probability is likeliness to be true

John Locke, An Essay Concerning Humane Understanding IV.XV.3. Exact opening clause before a comma; concerns his historical account of knowledge and assent, not a modern numerical model.

Connections between traditions

Question	Primary grounding	Boundary
Is ending doubt sufficient?	Peirce, III–IV [1]	Settled opinion is not identified with truth.
Does conviction’s strength fit its grounds?	Hume, X.87 [2]	Historical experience language supplies no modern forecasting formula.
Is there a position between knowledge and fantasy?	Locke, IV.XV [3]	A probable judgement has grounds without full certainty.
How should assent’s degree be determined?	Locke, IV.XVI.1 [3]	A philosophical question about grounds, not a universal numerical scale.
Can action coexist with an unknown?	Editorial application [1–3]	Decision and promised work are described separately from established facts, not prescribed by the classics.

Different positions remain distinct. Everyday examples and practice are original; psychological efficacy and a unified probability theory are not claimed.

Primary sources

Charles Sanders Peirce, The Fixation of Belief, 1877. English original. Actually read: displayed III–IV and V’s opening, concerning doubt, belief, inquiry’s settlement, and attachment to opinion. This is philosophical reasoning, not a clinical model; its historical behavioural illustrations are not prescriptions for readers.
David Hume, An Enquiry Concerning Human Understanding, 1748. Consulted text: 1777 posthumous English text, edited L.A. Selby-Bigge, second edition 1902, Gutenberg #9662. Actually read: X Part I paragraph 87, with neighbouring 86 and 88–89 context. The epigraph is in X.87, not VI. The full religious polemic and historical notion of infallible experience are not adopted as modern standards.
John Locke, An Essay Concerning Humane Understanding, first published 1690. English text, Gutenberg #10616, electronic transcription marked Based on the 2d Edition, Books III–IV. Actually read: selected IV.XV.1–5 paragraphs concerning grounds and probable assent, XVI.1’s opening, and the displayed title page. Neither whole-book reading nor a precise print date for this electronic copy is claimed. The key quotation is XV.3’s opening clause before a comma. This is historical philosophy, not statistical calculation.

Short Russian quotations are United Wisdom’s translations of the specified English texts. Everyday examples, reporting language, decisions, and practice are original applications; neither philosophical unification nor demonstrated efficacy is claimed.

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.