A value is an enduring evaluative principle about what an individual, group, institution or society regards as important, desirable, worthy, permissible or worth protecting.
Values shape what counts as a good outcome, which trade-offs are acceptable, what deserves priority, which actions are ruled out and how success is interpreted. Integrity, safety, fairness, autonomy, achievement, efficiency, compassion, privacy, reliability and tradition can all function as values, but they do not all operate in the same way.
Good classification therefore separates values from preferences, goals, priorities, rules and standards. A preference favours one option; a goal identifies a desired state; a priority allocates limited attention; a rule constrains behaviour; a value supplies part of the evaluative reason why any of those structures matter.
Quick answer: how should values be categorised?
- Object: what does the value concern—people, outcomes, methods, relationships, institutions, knowledge or environment?
- Role: intrinsic, instrumental, procedural, protective or expressive?
- Source: personal, familial, professional, cultural, ethical, legal, religious, organisational or civic?
- Strength: weak preference, strong principle, default priority or non-negotiable boundary?
- Hierarchy: which values take precedence when they conflict?
- Scope: universal, role-specific, institutional, local or context-dependent?
- Evidence: stated principle, repeated action, sacrifice, policy, resource allocation or institutional practice?
- Conflict: what other values pull in a different direction?
- Stability: enduring, evolving, situationally activated or contested?
- Implementation: how does the value become rules, incentives, choices, controls or habits?
- Review: when should interpretation of the value be reconsidered?
This article complements How to Categorise Preferences, How to Categorise Priorities, How to Categorise Goals and How to Categorise Rules.
Value is not preference
A preference is comparative and local: tea over coffee, morning over evening, option A over option B. A value is usually broader and more durable: health, learning, privacy, fairness. Preferences can express values, but they can also reflect convenience, habit or temporary state.
Value is not goal
A goal can be completed; a value usually continues. “Reach 70 marks” is a goal. “Take learning seriously” is a value that may generate many goals across time.
Value is not priority
A priority determines what receives scarce attention now. Values influence priorities but do not map to them mechanically. Safety may be a permanent value while a current operational priority is restoring service.
Value is not rule
A rule says what must, may or must not happen. A value helps explain why the rule exists. The same value can generate different rules in different contexts.
Intrinsic values are valued for their own sake
Friendship, dignity, beauty, knowledge or wellbeing may be treated as worth pursuing independently of what else they produce. Intrinsic classification does not require universal agreement; it records the role the value plays in the actor’s evaluative structure.
Instrumental values matter because they produce something else
Efficiency, punctuality, discipline or standardisation can be valued because they support reliability, learning, cost control or trust. Instrumental values should remain linked to the ends they serve so the means do not become detached from purpose.
Procedural values concern how decisions are made
Transparency, due process, participation, consistency and independent review can matter even when several outcomes would be acceptable. Procedural values protect the legitimacy of the route, not only the destination.
Protective values defend boundaries
Safety, privacy, dignity and rights can operate as constraints on optimisation. A system may refuse a profitable or efficient option because it violates a protected value.
Expressive values communicate identity
Tradition, craftsmanship, sustainability, openness or independence can signal what an individual or institution understands itself to stand for. Expressive values can shape choices even where direct material benefit is small.
Personal values belong to individuals
Personal values can guide life choices, relationships, learning and work. They should not automatically be generalised to family, team or institution.
Professional values attach to roles and disciplines
Accuracy, confidentiality, competence, impartiality and duty of care can arise from professional practice. These values may create responsibilities that exceed personal preference.
Organisational values require institutional evidence
A poster saying “quality” is weak evidence. Hiring, promotion, budgets, product decisions, review processes and responses to failure reveal whether a stated organisational value is actually institutionalised.
Civic values govern shared life
Rule of law, public trust, equality, participation, order and pluralism can shape institutions and collective decisions. Civic values often conflict, requiring explicit balancing rather than pretending one value determines every answer.
Ethical values evaluate right and wrong
Honesty, fairness, compassion, non-harm and respect for autonomy can operate as moral reasons. Ethical classification should record the principle or framework rather than presenting normative judgement as a measurement.
Values can be universal in aspiration yet local in interpretation
Two institutions may both claim fairness while operationalising it differently: equal treatment, equitable support, procedural neutrality or outcome correction. Label agreement does not prove semantic agreement.
Value definitions need behavioural meaning
“We value integrity” becomes operational only when people know what behaviour follows: accurate representation, conflict disclosure, refusal to falsify evidence, correction of mistakes and consistency under pressure.
Values can conflict
Privacy can conflict with transparency, autonomy with safety, efficiency with resilience, loyalty with impartiality, compassion with consistency. Conflict does not mean one of the values is unreal; it means a decision requires hierarchy, trade-off or redesign.
Value hierarchy should be explicit where stakes are high
An institution may decide that safety outranks speed, integrity outranks short-term performance, or rights outrank popularity. Explicit hierarchy makes decisions more consistent and auditable.
Some values are lexicographic
A lexicographic value must be satisfied before lower-level benefits are considered. For example, a safety threshold may need to pass before cost optimisation begins.
Some values are tradeable
Convenience, speed and aesthetic preference can often be exchanged against cost or flexibility. The acceptable trade-off rate is itself part of the value structure.
Non-negotiable values create hard boundaries
When a value is declared non-negotiable, the system should identify what actions are excluded and who has authority to define exceptions, if any. Otherwise “non-negotiable” remains rhetorical.
Stated values and revealed values can diverge
People and organisations may state one principle while repeatedly sacrificing it under pressure. Repeated choices, resource allocation and tolerated behaviour provide evidence about operative values.
Behaviour is evidence of values, not a perfect readout
Constraints, fear, limited information and conflicting duties can cause actions that do not fully reflect values. Classification should avoid turning every observed choice into a moral diagnosis.
Sacrifice reveals value strength
A value appears stronger when an actor accepts real cost to preserve it: money, time, convenience, popularity or opportunity. Costly consistency is stronger evidence than effortless agreement.
Values can be aspirational
An actor may sincerely value something while behaviour still falls short. Aspirational status should not be confused with established practice, but aspiration can guide improvement.
Values can become institutionalised
A value becomes more durable when embedded in rules, role design, incentives, review gates, training, budgets and correction processes rather than depending on one leader’s personality.
Institutionalisation can distort a value
A metric created to represent quality can become a target that displaces the deeper value. Operationalisation should therefore be reviewed to ensure the proxy still serves the principle.
Values influence metrics but should not collapse into them
Trust, learning and fairness can be measured imperfectly through indicators, but no single metric fully owns the underlying value. Good systems preserve qualitative judgement and multiple evidence sources.
Values influence incentives
If collaboration is valued but only individual output is rewarded, the incentive structure contradicts the stated value. Alignment requires that rewards, sanctions and attention do not systematically punish desired behaviour.
Values influence policy
Privacy values can generate access rules; fairness values can shape allocation policy; safety values can create hard review gates. Policy should retain traceability to the values it is meant to implement.
Values influence plans
Two plans pursuing the same goal can differ because one prioritises speed and another resilience. Values help explain why the chosen route is preferred.
Value drift can occur
Institutions can slowly change what they reward and tolerate while retaining the same public language. Periodic review should compare stated values with real decisions and practices.
Value interpretation can evolve without abandoning the value
A society can continue valuing fairness while changing its understanding of what fairness requires. Version interpretation and implementation, not merely the label.
Values can be contested
People can disagree about which values matter, what they mean, how strongly they bind or which takes precedence. A classification system should preserve the nature of the dispute rather than forcing artificial consensus.
Group values require aggregation or authority
“Our community values…” can mean majority opinion, constitutional principle, leadership statement, inherited tradition or negotiated consensus. The source of collective value claims should be explicit.
Minority values may remain protected
Collective decisions should not automatically erase minority rights or deeply held principles. Majority preference and legitimate value boundaries are different objects.
Educational values shape teaching beyond marks
Curiosity, responsibility, disciplined thinking, empathy and intellectual honesty influence how learning is designed, not merely what score is targeted. A system that optimises marks while damaging these values may achieve a metric while weakening the broader educational mission.
Values can support diagnostic decisions
If independent thinking is valued, a teaching plan should not solve every problem for the learner. If accuracy is valued, feedback should reward checking rather than speed alone. Values therefore influence which interventions are considered acceptable.
AI systems can encode values indirectly
Objective functions, ranking rules, safety policies, refusal criteria, evaluation rubrics and product defaults all embed evaluative choices. Calling them “technical” does not remove their value content.
Value alignment requires operational specificity
“Be helpful” or “be fair” is too broad for reliable automated behaviour. Systems need explicit constraints, examples, exceptions, escalation routes and evaluation cases that translate values into observable conduct without pretending the translation is perfect.
AI should not infer personal values too confidently
One purchase, click or statement may reflect circumstance rather than enduring principle. Consequential value inference should rely on explicit user expression or strong contextual evidence, and uncertainty should remain visible.
A practical value record
- value ID and name;
- holder: person, group or institution;
- definition;
- object or domain;
- intrinsic, instrumental, procedural, protective or expressive role;
- source and authority;
- strength or non-negotiable status;
- scope;
- higher and lower values in hierarchy;
- known conflicts and trade-offs;
- behavioural examples;
- evidence from decisions and resource allocation;
- linked policies, rules and incentives;
- interpretation version;
- review trigger;
- status: stated, aspirational, institutionalised, contested or superseded.
Worked example: valuing integrity
An organisation says it values integrity. Operational meaning might include accurate claims, visible uncertainty, conflict disclosure, evidence traceability, correction of mistakes and refusal to misrepresent a result even when doing so would improve short-term metrics.
The classification becomes stronger when those expectations appear in review criteria, incentives and correction behaviour. Integrity is then more than a slogan; it is connected to actual decisions and costs.
Worked example: efficiency versus resilience
A system may value both efficiency and resilience. Eliminating spare capacity improves efficiency but weakens ability to absorb shocks. The correct classification does not ask which value is “real”; it records the conflict and the conditions under which one takes precedence.
Questions to ask before accepting a value classification
- What does the value mean in this context?
- Who holds or claims the value?
- Is it intrinsic, instrumental, procedural or protective?
- How strong is it?
- What evidence shows it influences real choices?
- What cost is the actor willing to accept to preserve it?
- What other values conflict with it?
- What takes precedence when conflict occurs?
- How is the value translated into rules, plans or incentives?
- Is the value stated, aspirational or institutionalised?
- Has its interpretation changed over time?
- What would count as evidence of value drift?
Common classification mistakes
- Confusing values with preferences or goals.
- Treating a value label as self-defining.
- Assuming stated values equal revealed practice.
- Ignoring conflicts among genuine values.
- Turning every value into one metric.
- Calling a value non-negotiable without defining the boundary.
- Generalising one leader’s values to an entire institution.
- Ignoring value drift behind stable public language.
- Assuming majority preference settles rights-based questions.
- Inferring personal values from sparse behavioural data.
The deeper idea
Values are the evaluative grammar underneath many other categories. They help determine what counts as success, what is worth protecting, which sacrifices are acceptable and which boundaries should not be crossed.
The challenge is not merely naming values. It is connecting them to evidence and behaviour without reducing them to slogans or metrics. A mature system preserves definition, hierarchy, conflict, implementation and change so values can guide action consistently while remaining open to reflection.
To categorise a value well is to preserve what is regarded as worth pursuing or protecting, why it matters, how strongly it binds, what it conflicts with, and how that principle becomes visible in real decisions and behaviour.
Final answer
Categorise values by object, role, source, strength, hierarchy, scope, evidence, conflicts, implementation and stability. Keep values separate from preferences, goals, priorities and rules, and judge institutional values by the decisions, incentives and sacrifices they actually shape rather than by stated language alone.