Expert disagreement: finding your way without your own specialism

Epigraph

We will, if we can.

Socrates replies to Theaetetus’s expectation that a mistake will be corrected; Plato, Theaetetus, opening discussion, Benjamin Jowett translation. The action is supplied by the preceding reply, not an independent maxim.

What this article is about

Two specialists give different answers, and you lack the training to check their calculations. It is tempting to choose the most confident, famous, or agreeable voice. Resisting that temptation still leaves the question of how to act. This article proposes a limited approach: identify the disagreement, examine each conclusion's grounds, and distinguish knowledge from acceptable consequences. You need not master a profession overnight. Consulting specialists remains necessary; the question is what you entrust to them. An opening episode in Plato's Theaetetus, Hume's discussion of conflicting testimony, and a teaching in Proverbs help pose different questions while retaining their philosophical and religious settings. The practical task is to prepare the next conversation and identify what a decision still needs. This is an original reasoning procedure, not a validated expert rating, a guarantee of correct choice, or advice for a particular medical, legal, or financial case.

I. Establish a shared question first

To assess disagreement, establish whether people answer the same question. An invented library is considering moving its reader catalogue to a new system. One specialist says records can be transferred without losing their main fields. Another says existing links will no longer reach the right entries. These reports need not conflict: preserving fields and preserving links are different outcomes. Describing this as a dispute over whether the new system is good creates a false choice between people. An accurate description creates another task: check both outcomes and identify which the library needs. Agreement begins by defining the subject, not by attempting to reconcile participants at any cost. A shared question can leave different answers visible.

Write the question in ordinary language, then ask each specialist to clarify it. What will be transferred? Which version of the original system is involved? Will readers retain access during the change? Where does the outcome to be checked end? This list does not replace a technical specification. It reveals promises concerning different objects, periods, or conditions. A reader can request an explanation relevant to their decision without knowing how to perform the calculation. An unfamiliar term is not necessarily evasive. Ask the specialist to connect it with the task: which outcome does it describe, and what changes in library work if its condition fails? That connection matters more than simply replacing every term with an everyday word.

An actual contradiction remains if, under identical conditions, one says a particular link will survive and the other says it will fail. The dispute now has an address. There is no immediate need to decide which person is generally more reliable. Ask which version and check support the opposing reports. Perhaps they saw different builds; perhaps they define a preserved link differently. This may resolve the disagreement or reveal its depth. The written question need not contain an answer. Its purpose is to hold the subject steady so that clarification reduces uncertainty rather than adding impressions from two persuasive performances. Understanding the disagreement is already useful work, even when its technical explanation remains unavailable to you.

II. Competence concerns a particular subject

At the beginning of Plato's Theaetetus, Socrates discusses whom to trust about similarly tuned lyres and similar faces. He asks about a musician and a painter, distinguishing these activities from Theodorus's geometry. We consulted this episode in Jowett's translation, not the entire dialogue. It raises a bounded question about the scope of skill, not a contemporary qualification procedure. Excellence in geometry does not answer every comparison by itself. Nor should we infer that a non-specialist cannot notice a resemblance. The article uses a question about relevant preparation while retaining the hypothetical movement of Socrates's conversation. Its historical setting does not establish our modern selection rules.

In the library, someone might be good at moving databases but rarely work with reader search. Another might understand search but not be responsible for transforming records. Neither is thereby an impostor. Connect preparation with a particular part of the decision. Which similar task has the person performed? What did they check themselves? Which outcome are they responsible for now? Their answer can show useful experience, but past success does not automatically confirm a new version. Experience warrants attending to a particular argument; checking today's outcome remains separate work. Related competence can also be named without presenting it as complete or dismissing its contribution. A team may need several kinds of knowledge that no single member wholly possesses.

Qualifications, position, and reputation can convey relevant information, but their relation to the question still needs explaining. Avoid inventing a ranking by counting honours without a common criterion. A professional title sometimes covers several activities. Clarifying the subject can help you seek relevant experience or an additional opinion on a particular part. This does not demand personally verifying every document and biographical detail. The depth of checking depends on consequences and available conditions. If the relevance of someone's training cannot be established, record the gap. Confidently using a professional word does not close it; an unpleasant manner does not establish the absence of appropriate knowledge. Both require attention to what the person can actually explain about this task.

III. Identify the level of the conclusion

A specialist can report an observation, explain it, forecast an outcome, or recommend an action. Distinguish these levels. “Selected entries opened in the test copy” reports a check. “The transformation preserved relationships” explains its result. “Every future request will work” extends the conclusion. “We can move the catalogue tomorrow” adds a decision. The first statement can be accurate while the others need further grounds. Someone can accept an observation and reject a recommendation. If you heard only the final sentence, reconstruct the transitions. You need not investigate the program yourself: the specialist can explain which checks support the proposal and where confidence already concerns an assumption. The distinction keeps their answer proportionate to the work performed.

“Why do you think that?” can be too broad. Ask instead what result was obtained, what was inferred from it, and which untested condition the next step needs. This invites the structure of an answer, not an examination administered by a layperson. A technical proof may remain difficult, but you can learn whether it concerns the relevant version and outcome. If the argument requires specialist assessment, an independent specialist can receive that particular argument. Their task is then to consider the identified transition rather than repeat a general impression of the project. Such a referral needs the original materials and the question's boundary. Otherwise the additional opinion can become another recommendation that cannot be compared with the first.

An honest answer can contain different degrees of certainty: particular fields have been checked, existing links partly checked, and future search usability remains unevaluated. Treating this as evidence of insufficient confidence would miss the boundary of completed work. Equally, do not announce “the specialists agreed” when they agreed only at the first level and still disputed the decision. A separated record prevents their words from silently expanding. Otherwise a limited technical result can acquire the force of permission for the whole transition, although nobody gave that permission. A conclusion should retain its subject and conditions when it is passed to a manager or the library's readers. A short account can still preserve those essential distinctions.

IV. Request grounds that can be discussed

An accessible account helps orientation but cannot alone replace grounds. In our example, ask for the version checked, the selected entries, and the result of checking links. This is a hypothetical illustration, not a universal migration standard; another task needs different materials. The relevant connection is what the material allows someone to check about the disputed assertion. A hundred pages on system advantages might say nothing about old addresses. A brief check of several addresses does not establish the whole catalogue either. A package's size does not determine its sufficiency without the question it must answer. Name inadequate support specifically while preserving the useful part already obtained. Missing coverage and an entirely worthless result are different assessments.

Accessibility does not mean every professional activity becomes simple for a non-professional. Grounds may contain formulas, specialised measurements, or protected information that cannot be freely distributed. Request an account of the procedure and its limits appropriate to the decision-maker. Where expert examination is needed, agree to transfer material to an authorised reviewer. Lack of open access is not proof of deception, but affects what you actually established. Distinguish “the specialist reports that material exists,” “the reviewer can access it,” and “its contents have been checked.” A similarly reassuring voice does not merge these into one confirmed result. A clear explanation of access limits can itself help decide which further inquiry is possible and who should perform it.

Requests should be proportionate. A reversible trial on a copy may permit a limited check. Irreversibly removing the previous catalogue calls for a different degree of readiness and separate permission. This distinguishes consequences; it is not a technical storage instruction. We do not know a particular library's arrangements. Ask specialists what enables a return to the previous state and what remains untested. If no answer is available, do not quietly substitute confidence. Postponing that step can be a substantive conversational outcome while preparation continues where the missing answer does not impede authorised work. A decision can therefore narrow its scope without treating either the entire project or every professional contribution as a failure.

V. Compare the circumstances of conflicting reports

In section X of An Enquiry Concerning Human Understanding, Hume considers circumstances of testimony and contrary reports. The passage belongs to Of Miracles and a philosophical argument, not a modern test of professional forecasts. Our limited question concerns the grounds of assurance when reports conflict. We neither adopt all his religious conclusions nor use appearance as a validated lie detector. The procedure for questioning specialists is our proposal, open to revision.

If specialist A checked entries at their new addresses and B checked old reader links, their results do not yet compete on the same ground. If they checked the same addresses on different dates, investigate version changes. Identical starting conditions would require examining differing actions or results. You can prepare that comparison without diagnosing the technical cause yourself. Do not announce a cause after hearing the first plausible explanation. “Perhaps they used different copies” remains a hypothesis until those copies are compared. The record separates work on the discrepancy from a reassuring story that merely conceals its inconvenience. A plausible cause can direct investigation; plausibility alone cannot declare the original contradiction resolved or justify expanding a limited conclusion.

A dispute can persist after materials are exchanged. Ask each person which transition in the other's argument appears mistaken. “They have less experience” may inform selection but does not explain different results. “Their check excludes old addresses” identifies further work. If the difference cannot be made accessible, commission a bounded comparison from a third specialist. This does not guarantee agreement: they may find missing data or propose another position. Its value depends on whether it clarifies the decision's grounds. The number of people consulted does not measure checking depth or turn an open question into an answer. Additional expertise helps when its task is clear enough to reveal what its contribution has actually changed.

VI. Agreement and independent reports

Five identical answers may derive from one report. In the invented library, staff repeat the system author's conclusion, and the manager counts their agreement as five confirmations. They really agree, but independent checks do not follow. Ask who saw the result, who relied on one report, and who performed separate work. Repetition is not necessarily useless: it can show how a rule is understood or how widely a recommendation is known. Yet it should not acquire the name of verification if nobody performed a new check. This is a logical distinction in our situation, not an empirical theory of how professional agreement arises in every field. Its value is in describing what these particular reports contribute.

Professional agreement also needs a subject. Supporting a general approach differs from confirming a particular version. Specialists may agree that catalogues can be moved under specified conditions and disagree over whether those conditions hold here. General opinion can guide questions without cancelling local checking. The opposite error is deciding that one dissent destroys accumulated knowledge. An opinion may concern a narrow detail, new material, or a different cost. Identify the disagreement's extent: what is shared and what disputed? Dissent then ceases to be a universal reason to treat all positions as equally supported or every decision as impossible. Neither a count of voices nor the mere presence of disagreement supplies the full basis for this decision.

If assessing professional discussion is difficult, ask for a relevant review and an explanation of its relation to the task. That review becomes a separate source requiring appropriate reading; an institution's name cannot replace its content. Retain its date, domain, and question. An older conclusion can supply a useful historical basis, but unknown changes cannot be announced as checked. This article supplies no measured consensus figure for a contemporary profession. Our question concerns handling a reported agreement: who describes it, what it covers, and how it supports the present decision. Answering may require further research rather than an immediate vote. The request is for usable grounds, not a performance of unanimity by everyone available to ask.

VII. Facts, forecasts, and acceptable costs

Specialists sometimes agree about technical feasibility but recommend different actions. One accepts a short access interruption to complete the move; another favours slower preparation so readers retain access. Both the interruption forecast and judgments of acceptable consequences may differ. Discuss them separately. Asking only “who is right?” risks transferring a decision about others' inconvenience along with a technological assessment. A specialist can explain consequences; authority to choose their distribution is determined separately. In a library, that choice may belong to a role or agreed procedure. The article establishes neither a real institution's organisation nor its rules. It asks how professional knowledge enters a decision whose scope includes people who are not making the technical assessment.

Continue “what do you recommend and why?” by asking which goals make the recommendation appropriate and which consequences it treats as acceptable. Both options may suit different priorities. This does not make facts relative: interruption length remains checkable, while acceptability requires responsible people's decision. Nor should conflicting interests of readers and employees become a purely technical mistake. An improvement can create work that needs agreement. When the specialist separates these questions, their answer is easier to use accurately. When you preserve the distinction in retelling, a limited recommendation does not become everyone's consent. Understanding the technology can clarify the available options while leaving a real choice about obligations and priorities unresolved.

Retain forecasts with their conditions. “The move takes one working day with these data” does not promise that period regardless of the source database. If some conditions are unchecked, ask how this affects the decision. Absolute certainty is not demanded. Identify which remaining uncertainty is accepted and by whom. Available, authorised options may include a reversible step, further checking, or postponement. After accepting a risk, however, do not say the specialist proved the outcome. Responsibility includes naming that difference honestly. It remains even after success: a good result does not rewrite what was known at the time. Keeping the earlier grounds visible also makes later learning more precise than simply remembering that the choice worked.

VIII. Interests, explanation, and a right to ask

A specialist might be a product author, contractor, researcher, or employee expected to implement the recommendation. These relations can matter and should be identified when relevant. An interest does not prove an argument false. The system author may understand it best; an independent consultant may lack the relevant version. Separate a person's relationship to the decision from a report's grounds. Request relevant working conditions and independent examination of the disputed point. Do not infer hidden greed from tone or declare impartiality established because one conversation was unpaid. This article offers no method for reading intentions. It proposes identifying a relationship when that relationship affects the available grounds, responsibility, or scope of a professional answer.

Transparency does not demand someone's entire personal history. Ask about a relation material to the decision: who commissioned the assessment, which result the obligation covers, and which data are accessible. If protected conditions cannot be disclosed, record the limit and consider another possible check. You have a right to understand what supports your action; the specialist can specify the scope of their assignment and answer. These rights do not automatically resolve conflict, but make it discussable. Otherwise explanation becomes a status contest: one demands unconditional trust, the other impossible access to everything, and the disputed assertion disappears from view. A bounded question can protect both the decision's needs and information that has no legitimate bearing on it.

Presentation also has limits. Confidence can accompany accurate knowledge or overgeneralisation; caution can reflect a known boundary or insufficient preparation. Tone alone does not settle which. Return to what the person checked and can explain. Unpleasant communication remains a separate condition of collaboration: you need not endure humiliation to obtain information. Refusing an unsuitable format does not prove the content refuted. Where the task matters, another channel or specialist may be available. Choose conditions that permit understanding while retaining participants' dignity. Manner should not become an independent instrument for measuring truth. A respectful format supports discussion; it does not supply evidence that an otherwise unsupported conclusion is correct or guarantee that everyone will agree.

IX. Hear another account without equalising support

Proverbs 18:17 in the KJV describes the first person presenting a case as apparently right before another examines it. This religious wisdom teaching occurs among sayings about speech and judgment. It establishes neither a modern scientific procedure nor equal support for every opposing side. Our comparison is limited: the first persuasive account does not finish the question. A material objection needs to become visible. The verse does not supply our entire voluntary discussion procedure; contemporary working conditions require their own decisions.

Hearing another specialist does not divide trust equally. If one presents a check of the relevant version and the other a conjecture about another system, their grounds differ. That conjecture might nevertheless reveal an untested condition. Identify its current status: established result, possible cause, question about coverage, or preference for action. This preserves the objection's useful part without inflating it into complete refutation. It also protects the first conclusion from an undeserved victory when another person's weak formulation supposedly proves that no difficulty exists. A decision should consider a material objection rather than feature two opinions merely for visible balance. The difference concerns what the accounts warrant, not which person receives an equal share of conversational attention or personal worth.

This work does not require publicly examining someone who reports harm. The article considers professional assertions about an agreed task, not compulsory debate over another person's experience. Reports do not all permit the same format. If participants cannot freely decline discussion or face punishment for a question, that materially affects the decision. An appeal to hearing both sides cannot conceal it. Separate fact-checking, seeking help, and allocating authority. Scope becomes especially significant with medical, legal, or financial consequences: our general questions replace neither qualified work on a specific case nor permission for independent action. Keeping the decision's boundary clear helps prevent a reasoning exercise from being used to impose an inappropriate burden on the person affected.

X. Deciding when disagreement remains

Two incompatible conclusions can remain after clarification, with no available check yet. State what was established, where the discrepancy arose, and what the next step needs. This is more useful than “all specialists disagree” because it preserves the actual boundary. The library can continue describing fields without treating old-link transfer as confirmed. Work independent of that answer can continue if authorised. When the disputed outcome is necessary to an irreversible action, missing support limits that action. We do not determine the technical safety of a particular move; we show how to preserve knowledge status in ordinary language. A limited open issue can remain limited instead of turning the whole project into an undefined state of doubt.

A responsible choice can name a return date, an appropriate reviewer, and a revision condition. Every conversation need not generate an elaborate protocol. A record sufficient to recover the connection between grounds and decision is enough. Identify changed circumstances separately, retaining the earlier conclusion. A successful new check does not make a previously unchecked result checked in the past. A specialist's mistake does not establish that all their work is useless. Discuss the particular discrepancy, correction, and consequences. This allows responsibility without indiscriminate trust or suspicion, both of which erase differences among questions. A specific record gives later participants something more useful than memories of who sounded most confident or who was blamed when a problem emerged.

Orienting without your own specialism means doing the available work and recognising its limit. You clarify the subject, distinguish conclusion levels, request grounds, and identify decision authority. Specialists perform professional examination. Further research can follow instead of demanding an immediate winner. Plato's question about skill, Hume's account of conflicting testimony, and religious advice about the first account offer different horizons; they form no single theory of expertise. Our comparison suggests an original question: what warrants trusting this person with this conclusion under these conditions, and what will I retain as unknown if a persuasive answer is unavailable? The question keeps consultation necessary while allowing the person who must decide to understand the boundary of what is being entrusted.

What this means in practice

Choose one question you already intend to discuss with specialists. Write it without demanding endorsement of a preferred option. Name the object, desired result, and known conditions. Separate your purpose from the professional assertion that needs checking. If consequences affect others, identify who has authority to decide about them. This original conversation-preparation exercise has no independently established efficacy and replaces neither professional consultation nor an institution's rules.

Record answers separately: what was observed, explained, forecast, and recommended? Ask for material support and a limitation. For one disputed point, establish whether initial conditions match and propose comparison by a competent reviewer if your understanding is insufficient. Do not make this a demand for public justification. A written answer, separate meeting, or another available format may be suitable. You need not disclose irrelevant personal information or perform a dangerous action to check someone else's words.

End with the next step's status: confirmed result, limited authorised trial, further research, or awaiting a material answer. Name what remains unknown and which answer could change the choice. Preserve a revised position as a revision, rather than pretending the specialist always said it. Where agreement remains unavailable, identify the discrepancy's subject instead of assessing all participants. Clear postponement can be an outcome: a particular action waits because its required result is unconfirmed. The record guarantees no correct outcome, but preserves an intelligible basis for choice and the ability to return without reconstructing the dispute from memory.

Key quotation

but his neighbour cometh and searcheth him.

Proverbs18:17, final clause, King James Version: the preceding clause concerns the apparently right person who presents a case first. Religious wisdom, not a rule that every side has equal support.

Connections between traditions

Plato / Hume: scope of skill and circumstances of testimony raise different grounds for trust. Geometrical reputation and a particular report’s conditions are not interchangeable. The ancient dialogue and philosophical discussion of miracles supply no modern expert rating.
Plato / Proverbs18:17: our shared question is whether a first voice still needs consideration. Socrates’s domain question differs from religious advice about a presented case. Contemporary voluntary discussion is not derived from the dialogue’s historical hierarchy.
Hume / Proverbs18:17: conflicting reports can be compared with examination of a first account. Philosophical grounds of assurance and wisdom advice retain different purposes. Hearing another voice does not establish equal support for both.
Plato’s correction reply / our decision: Socrates’s reply to Theaetetus allows correction within what is possible. Recording uncertainties and revision conditions is our proposal, not a quoted Platonic organisational instruction.
Three contexts / the library: competence, conflicting reports and a first account help examine the invented example. Our separation of observation, forecast, agreement and authority is not attributed to all three texts; no measured procedural efficacy is asserted.

Primary sources

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.