Harm

Type: Node / Realized Adverse Effect

Working Definition: Harm is a realized adverse change, condition, impairment, deprivation, or loss borne by a subject with respect to some affected interest, capacity, state, or integrity, within a specified scope and interval. Nothing is harmful simpliciter — the relation always holds to some subject and with respect to some affected interest or capacity, relative to an evaluative basis such as a baseline, a protected interest, a capacity, or a Standard, whether or not any Agent has yet identified or disclosed that basis.

Function in the Grammar: Harm names the realized, subject-borne side of an adverse effect. Three concepts do related but genuinely distinct work, and this entry exists to keep them from collapsing into each other: Danger specifies a prospective pathway to possible harm before the particular Harm token at issue is realized — Danger may equally concern continuing or repeated Harm after an earlier Harm has already occurred; Harm describes the realized adverse condition a subject actually bears; Consequence connects a source to that realized condition as an effect. Once a Harm has been interpreted or represented and carries Meaning, Significance may assign weight to it within a Frame; Significance does not operate directly on a wholly uninterpreted condition. Evidence and Claim support or contest whether a Harm exists and what caused it. Accountability addresses responsibility and response, without being entailed by Harm's existence alone. None of these concepts substitutes for another, and Harm's own admission is justified by the specific gap the surrounding concepts leave: Consequence's own vocabulary explicitly disclaims carrying "harm" as a structured property — its Semantic Constraints state that "assessments of importance, harm, benefit, proportionality, or deservedness must disclose their affected subject, perspective, basis, and scope," a requirement Consequence names but does not itself supply argument slots for.

This entry sits in Division V, alongside Consequence, not Division IV alongside Danger. Danger's own Division IV placement note states the test precisely: Danger belongs there because its entire demonstrated grammatical work is in service of how a prospective relation gets appraised, claimed, recognized, and acted on. Harm does not clear that test the same way. Harm's own demonstrated work is on the realized-effect side: a structured, subject-borne condition that Consequence's own vocabulary needs but does not supply, matching Consequence's own placement rationale — "the general relation needed to say what follows from an occurrence, act, condition, or governed determination... beside Obligation without being limited to normative cases." Harm's own appraisal-independent and recognition-independent existence is real: an injury does not require any Agent to have appraised, claimed, or Recognized it to have occurred. But appraisal-independence alone does not settle placement, since Consequence is equally appraisal-independent — its own text gives the example directly: "a pollution event may harm a community before any forum recognizes the harm" — and Consequence sits in Division V. What is decisive is where each entry's own compositional yield actually falls: Danger's yield is in structuring appraisal; Harm's is in structuring the realized condition that Consequence's effect argument otherwise leaves unstructured.

Harm's relation to a prior Danger relation is optional composition, not a requirement. A Harm may realize a specific Danger relation's pathway, in which case the correspondence is worth stating explicitly; but most realized Harms will have no prior appraised Danger at all, and Harm must remain fully representable in that ordinary case. Treating a Danger relation as though it were Harm's causal source is a category error this entry does not make: in an ordinary case of realized harm, the causal source is the event, act, or condition Consequence's own source argument already names — a component failure, a deployment, an act of violence — not the earlier prospective relation that represented one way the harm might have come about.

Formal Pattern

Harm(h, subject: s, adverse_effect: e, interest: i, kind: k,
     scope: S, occurs_over: t, basis: b)
  is well-formed iff:
  s, e, i, k, S, t, and b are supplied and correctly typed as follows:
  s is a Subject(s): the person, population, system, or thing the change
    is borne by
  e is an AdverseEffect(e): a realized change, condition, impairment,
    deprivation, or loss in s
  i is an AffectedInterestOrCapacity(i): the interest, capacity, state, or
    integrity e affects
  k is Kind(k): the character of the adverse effect (e.g. physical,
    psychological, economic, relational, reputational)
  S is Scope(S): the population, part, domain, distribution, or other
    boundary within which e is asserted to affect s — distinguishing
    harm to one member, subgroup, capacity, or region from harm
    generalized to the entire subject
  t is Interval(t): the period over which e occurs or continues
  b is EvaluativeBasis(b): the basis on which e counts as adverse with
    respect to i — a baseline, a protected interest, a capacity, or a
    Standard — distinct from any further judgment about whether the act
    or source that produced e was wrongful, justified, or net-beneficial
    overall

  A well-formed Harm statement is not thereby a true one: supplying every
  slot produces a well-formed candidate, not a fact about the world.

Harm(h, subject: s, adverse_effect: e, interest: i, kind: k,
     scope: S, occurs_over: t, basis: b)
  holds iff:
  e is realized in or borne by s within S during t
  e adversely affects i
  b supplies the evaluative baseline, criterion, or relation under which
    that effect is adverse
  k accurately characterizes the effect

  This truth condition requires neither Evidence, disclosure, nor any
  Agent's Attention to be satisfied — an unnoticed Harm must remain
  possible. Whether any of this has been disclosed by anyone is a
  separate question, addressed by the Claim-disclosure requirement below.

SEVERITY(h, x)
  -- asserted only where support permits; not a required argument of
     Harm(h), and its absence does not affect whether Harm(h) is
     well-formed or holds

SUBJECT(h, s)
ADVERSE_EFFECT(h, e)
AFFECTED_INTEREST_OR_CAPACITY(h, i)

A Claim asserting Harm(h) must disclose:
  the affected subject
  the affected interest or capacity
  the asserted adverse effect
  the evaluative basis
  the relevant scope and interval

  -- disclosure is a requirement on the Claim, not on Harm(h)'s truth; an
     appraisal-independent Harm may exist and hold before any Agent has
     identified or disclosed any of this

Harm(h) does not entail:
  a prior Danger relation was ever appraised, claimed, recorded, or
    Recognized
  wrongdoing, fault, or Breach by any Agent
  that the source or act producing e was unauthorized, unjustified, or
    net-harmful overall
  Accountability, Enforcement, or Repair
  agreement about k, or about SEVERITY(h, x) where asserted — either may
    remain disputed while Harm(h) still holds

HAS_CONSEQUENCE(source: src, effect: h, under: cb, within: S, over: T)
  -- Consequence connects a source to h as its effect; h fills
     Consequence's existing effect argument, and is not a separate
     primitive Consequence must be extended to hold

REALIZES(harm: h, danger: d, via: c) iff
  Danger(d) holds
  c is HAS_CONSEQUENCE(source: src, effect: h, under: cb, within: S, over: T)
  SUBJECT(h) corresponds to SUBJECT_AT_RISK(d)
  KIND(h) corresponds to KIND(POSSIBLE_HARM(d))
  CONNECTION_BASIS(c) corresponds to PATHWAY(d)
  the relevant EnablingConditions(d) obtained
  the onset or continuation of h falls within HORIZON(d)

  -- requiring Danger(d) to hold, not merely to be well-formed, prevents a
     Harm from formally "realizing" a false Danger representation merely
     because selected fields resemble one another. d need not have been
     appraised, claimed, recorded, or Recognized before h occurred — a
     previously unrecorded Danger relation can be reconstructed afterward,
     provided it actually held.

  -- optional composition: never required for Harm(h) to hold, and Harm(h)
     remains fully representable when no prior Danger was ever appraised,
     claimed, recorded, or Recognized. The correspondence check runs
     between Consequence's own connection_basis and Danger's own pathway
     — Harm itself carries no pathway or causal-source argument

Examples

Distinctions

Harm ≠ Consequence. Consequence is a value-neutral, general source-effect relation — natural, social, psychological, economic, normative, or institutional, and its own text is explicit that "beneficial or harmful" are independent, unstructured properties it may carry, not ones it structures. Harm is the specific, structured, adverse realized effect a subject bears. A Consequence can be beneficial, neutral, mixed, or harmful; Harm is what fills the effect argument when it is the latter.

Harm ≠ Danger. Danger is prospective — a relation that may hold before anything happens. Harm is realized — a condition a subject already bears. Neither requires the other to have been represented: a Harm can occur with no prior Danger ever appraised, claimed, recorded, or Recognized, and a Danger relation can hold indefinitely with no Harm ever following it at all.

Harm ≠ Significance. Significance is a direction-agnostic magnitude of mattering, assigned to an already-established item within a Frame. Harm's own severity or extent is not the same axis: a severe Harm to a small, easily overlooked population can carry low institutional Significance, and a comparatively minor Harm can carry great Significance for a vulnerable group, exactly as Consequence's own Distinctions already establish for Consequence generally.

Harm ≠ wrongdoing. Authorized, accidental, necessary, or justified acts can still cause real Harm. A medically necessary surgery causes genuine tissue injury and temporary impairment while producing a net benefit and remaining fully authorized; denying the Harm because the act was justified collapses a report about a subject's realized condition into a moral judgment about the act that produced it.

Harm ≠ injury, damage, loss, or suffering. Injury, damage, loss, and suffering ordinarily describe the realized adverse_effect or a family of such effects. kind classifies that effect as physical, psychological, economic, relational, reputational, or another supported category. Neither is synonymous with Harm, which is the fuller subject-relative relation.

Harm ≠ accountability. Establishing that a Harm exists does not by itself settle cause, fault, responsibility, remedy, or standing to contest it. Causal attribution requires a supported Consequence relation and its own connection_basis. Fault and responsibility require applicable Obligations, Rules, roles, or Standards. Answerability and response require Accountability's own structure. Harm alone settles none of them.

Common Failure Modes

Minimum Viable Test Case

Case A — Harm without wrongdoing:

s  = Patient_44
e  = AdverseEffect(tissue injury and temporary functional impairment)
i  = AffectedInterestOrCapacity(physical integrity, short-term mobility)
k  = Kind(physical)
S  = Scope(the patient alone; no other subject affected)
t  = Interval(procedure through expected recovery window)
b  = EvaluativeBasis(pre-operative physical baseline)

Well-formedness:
  Harm(h1, subject: s, adverse_effect: e, interest: i, kind: k,
       scope: S, occurs_over: t, basis: b) is well-formed: every
       argument is supplied and typed.

Truth (checked independently of well-formedness):
  e is realized in or borne by s within S during t                -- yes
  e adversely affects i                                          -- yes
  b supplies the evaluative baseline under which e is adverse    -- yes
  k accurately characterizes the effect                          -- yes
  → Harm(h1) holds.

SEVERITY(h1, moderate)   -- asserted separately, evidentially supported;
  not required for the truth check above

HAS_CONSEQUENCE(source: authorized_surgical_procedure, effect: h1,
  under: cb, within: S, over: T)
  -- the same procedure also has further, separately tracked consequences:
     symptom relief, resolved underlying condition, net improvement in s's
     overall health

Result:
  Harm(h1) holds and is fully compatible with the same source's producing
  a large net benefit. Denying h1 because the overall act was authorized,
  medically necessary, or beneficial would collapse a report about s's
  realized physical state into a moral judgment about the act. Harm(h1)
  entails nothing about wrongdoing, unauthorized conduct, or whether the
  surgery should have occurred.

Case B — Realizing a prior Danger, and the failure of doing so wrongly:

d  = the prior Danger relation from Danger's own Product Liability
     example: source = the vehicle component's latent failure mode,
     pathway = failure under specified load conditions
s' = Driver_9
h2 = Harm(subject: s', adverse_effect: physical injury, interest:
     bodily integrity, kind: physical, scope: Driver_9 alone,
     occurs_over: t2, basis: b2)

c  = HAS_CONSEQUENCE(
       source: component_failure,
       effect: h2,
       under: failure_under_specified_load,
       within: S2, over: T2
     )

REALIZES(harm: h2, danger: d, via: c)
  -- holds: SUBJECT(h2) corresponds to SUBJECT_AT_RISK(d), KIND(h2)
     corresponds to POSSIBLE_HARM(d), c's connection_basis
     (failure_under_specified_load) corresponds to PATHWAY(d), d's
     EnablingConditions obtained, and h2's onset falls within HORIZON(d)

Contrast:
  Driver_12 is separately injured by the same vehicle model through an
  unrelated manufacturing defect in a different component. A genuine
  Harm h3 exists, correctly connected to its own real source via
  c' = HAS_CONSEQUENCE(source: unrelated_defect, effect: h3,
       under: distinct_mechanism, ...), but
  REALIZES(harm: h3, danger: d, via: c') does not hold: c'’s
  connection basis (distinct_mechanism) does not correspond to
  PATHWAY(d). Asserting that h3 realizes d anyway is the
  mismatched-pathway failure this entry names — a mismatch between
  Consequence's own connection_basis and Danger's own pathway, not
  something contained in Harm alone, and a failure of the realizes-claim
  specifically, not of either driver's own independently well-established
  Harm.

Intellectual Lineage

Harm draws on Joel Feinberg's account of harm as a setback to interest, developed in Harm to Others (1984), the first volume of his The Moral Limits of the Criminal Law. Feinberg used the concept to bound what the criminal law may legitimately prohibit. This entry adopts his relational structure, harm as always harm to a subject's interest rather than a free-floating property, and generalizes it beyond the criminal-law question Feinberg was addressing, into a concept usable across legal, institutional, and ordinary-life domains.

Cross-References

• Consequence — connects a source to a Harm as its effect; Harm fills Consequence's existing effect argument rather than requiring Consequence's own formal grammar to change
• Danger — prospective where Harm is realized; a Harm may optionally realize a prior Danger relation's pathway via REALIZES, but neither requires the other to have been represented
• Significance — assigns weight to an established Harm within a Frame, once that Harm has been interpreted and carries Meaning; a Harm's own severity is a different axis from how much institutional attention or weight it receives
• Interpretation — a Harm must be interpreted before it carries Meaning; a wholly uninterpreted condition can hold as Harm without yet being available for Significance
• Meaning — the required intermediate between an interpreted Harm and any Significance assigned to it; Harm's own truth does not depend on any Meaning ever being fixed
• Recognition — institutional or collective uptake of a Harm Claim; a Harm may exist and hold before any Agent has Recognized it, exactly as it may exist before any appraisal
• Evidence — supports or undermines whether a claimed Harm exists, its kind, its severity, and its connection to a source
• Claim — a Harm assertion is a Claim, carrying its own basis, epistemic profile, and contestability, independent of whether the underlying Harm actually holds
• Accountability — may make an Agent answerable for producing, preventing, disclosing, or responding to a Harm, but is never entailed by the Harm's existence alone
• Breach — a Harm may occur with or without an applicable Breach; authorized, accidental, or justified acts can cause real Harm without breaching any Obligation, Rule, or Commitment
• Standard — may supply a baseline or criterion under which an effect is evaluated as adverse; applying a Standard does not by itself establish Significance, wrongdoing, or Accountability