error
Enable an agent to recognise, assess and address an error by relating a claim, result or action to an applicable reference and the evidence for a discrepancy.
Research draft, second pass
A second pass drafted this model: the structure a model of this thing needs, and what is known about it in the world. The line under this one says how the second half was obtained - researched against sources, or recalled without web access, in which case nothing here was read anywhere and every claim is a lead to verify. Unreviewed either way.
Researched by: Codex
Bundle → Layer → Finding → Questions Filled
6 bundles · 11 layers · 16 findings · 32 questions
Grounds of error Establishes what is alleged to be wrong and what makes that allegation applicable.
Error bearer
Locates the precise content or occurrence under assessment.
Locatable discrepancy
Record the particular assertion, result, omission or action alleged to depart from a reference.
- Which exact assertion, result, omission or action is alleged to be erroneous? definition
- Which version, occurrence and context delimit the allegation? boundary
Applicable reference
Identifies the comparison basis and its authority or evidential standing.
Reference justification
Record why this truth condition, value, rule or intended result governs this particular assessment.
- Against which truth condition, reference value, rule or intended result is the departure assessed? definition
- What establishes that this reference applies in the relevant context and time? provenance
- Would another legitimate convention change whether the item counts as erroneous? boundary
Evidence and adjudication Separates detection signals from evidence sufficient to support an error assessment.
Detection evidence
Captures how the possible error became visible and what the check actually tested.
Diagnostic comparison
Record the observation or test and its ability to distinguish error from acceptable variation.
- Which observation, comparison or test first indicated the possible error? provenance
- What alternatives, such as rounding, context mismatch or a faulty checker, could explain the same signal? boundary
Assessment limits
Records the support, disputes and unknowns surrounding the attribution.
Supported error attribution
Keep the strength of the error assessment separate from confidence in the exact corrected answer.
- What evidence supports or contests the conclusion that an error occurred? provenance
- Can the item be shown to be wrong even if its correct replacement or exact deviation remains unknown? boundary
- What additional check would resolve the most consequential uncertainty in the assessment? action
Discrepancy form and extent Characterises how the item departs from its reference and how far that departure can be assessed.
Error form
Classifies the departure without treating every error as a numerical difference.
Form-specific description
Describe whether the problem concerns content, absence, reasoning, representation, calculation or execution.
- What form of departure best describes this case, and can several forms coexist? definition
- Is the error in the represented content, the transformation applied to it or the action performed? boundary
Error magnitude
Assesses size or extent using measures appropriate to the comparison.
Interpretable deviation
Record a quantitative deviation only when the reference and comparison operation support one.
- Can the discrepancy be quantified, and which units, sign convention or denominator make that quantity interpretable? measurement
- How well is the reference known, and what limits does that place on the reported deviation? measurement
- Does a stated tolerance affect acceptability, the definition of error, or both in this domain? boundary
Origin and propagation Examines how the error arose, recurs and affects dependent material.
Error generation
Distinguishes observed recurrence from proposed explanations of its cause.
Supported origin hypothesis
Record candidate points of introduction and distinguish causal evidence from repetition alone.
- At which acquisition, interpretation, transformation or execution step could the discrepancy have entered? provenance
- Across comparable cases, what recurrence or directional pattern is actually observed? measurement
- What evidence would distinguish the proposed cause from competing explanations? action
Dependent effects
Tracks use of the erroneous item and whether that use changes subsequent outcomes.
Consequence trace
Separate exposure to an erroneous input from demonstrated error or harm in a dependent output.
- Which later claims, calculations, records or decisions used the erroneous item? provenance
- Which dependent outcomes change when the item is corrected, and which remain valid? measurement
- Which dependencies warrant reassessment first given the consequences of leaving them unresolved? action
Correction and reassessment Connects corrective action to verification and possible revision of the original attribution.
Corrective intervention
Defines what must change to remove or contain the supported discrepancy.
Targeted correction
Choose an intervention suited to the error bearer, its dependencies and the remaining uncertainty.
- Should the item be replaced, recalculated, qualified, withdrawn or held pending further evidence? action
- Which dependent items require correction or notification when this intervention is applied? action
Verification and revision
Tests the correction and keeps the error assessment open to contrary evidence.
Verified resolution
Record evidence that the intervention addresses the discrepancy and conditions for reopening or withdrawing the assessment.
- Which check would show that the corrected item satisfies the applicable reference without repeating the original checking weakness? action
- What evidence would require withdrawing the error attribution or revising its reference? boundary
- What remains unresolved after correction, including possible recurrence and downstream effects? action
Evidence and external alignment What the world already says about this thing, gathered so the model can be checked against it.
Reported evidence
Findings from the breadth pass, kept separate from the structural claims.
Check these first
Recalled without web access and unsourced; every item is a lead to verify.
- Which of these check these first hold for the sense of error this model covers, and on what evidence? provenance
Kinds and varieties
Recalled without web access and unsourced; every item is a lead to verify.
- Which of these kinds and varieties hold for the sense of error this model covers, and on what evidence? provenance
Real-world use
Recalled without web access and unsourced; every item is a lead to verify.
- Which of these real-world use hold for the sense of error this model covers, and on what evidence? provenance
Typical measurements
Recalled without web access and unsourced; every item is a lead to verify.
- Which of these typical measurements hold for the sense of error this model covers, and on what evidence? provenance
Failure modes and hazards
Recalled without web access and unsourced; every item is a lead to verify.
- Which of these failure modes and hazards hold for the sense of error this model covers, and on what evidence? provenance
Neighbouring kinds and how to tell them apart
Recalled without web access and unsourced; every item is a lead to verify.
- Which of these neighbouring kinds and how to tell them apart hold for the sense of error this model covers, and on what evidence? provenance
Classifiers Filled
- Family
- Thing Registry
- Category
- Information and virtual systems
- Entry kind
- thing
- Plane
- INF
- Domain
- INF.KNW
What it is Filled
An error is a departure of a belief, statement, inference or result from truth, validity or an applicable reference, whose identification depends on the criterion used to assess correctness.
In scope
- The claim, representation, calculation, observation or action assessed as erroneous
- The reference and comparison conditions that make an error assessment meaningful
- Evidence supporting or contesting the alleged discrepancy
- Error magnitude, pattern, consequences and propagation
- Detection, correction and verification of the error assessment
Out of scope
- Complete models of the people, organisations or systems associated with an error
- Measurement uncertainty as a standalone characterisation of incomplete knowledge
- Defects and failures beyond their relationship to an assessed discrepancy
- Intentional deception, misconduct and responsibility as independent concepts
- Domain-specific procedures for clinical, legal or engineering remediation
Why it exists Filled
Enable an agent to recognise, assess and address an error by relating a claim, result or action to an applicable reference and the evidence for a discrepancy.
Distinguishing features Filled
- An error assessment identifies what is wrong and under which reference; mere surprise or dislike does not establish error.
- Uncertainty concerns what is known or how well it is known; an error assessment concerns a departure, whose exact size may itself be unknown.
- Two different results establish disagreement; establishing error additionally requires grounds for preferring an applicable reference.
- A defect is a condition that may produce errors, while a failure concerns unmet performance; these need not coincide with an erroneous claim or result.
- Calling something an error does not by itself establish intention, negligence or blame.
What robots and AI may and may not do Filled
Must not
- Hide, minimise or deny its own errors.
- Label a person's view an error where the question is a matter of values or open debate.
- Blame a person for an error without evidence of who made it.
- Correct records that others rely on without leaving a trace of the change.
Only with a human decision
- Correcting an error in an official or shared record.
- Telling an affected person about an error that harmed them.
May
- Identify and correct errors, stating the criterion used to judge them.
- Report errors in its own outputs openly and promptly.
- Estimate and state measurement or calculation error where relevant.
Moral aspects Filled
- Admitting errors builds trust; hiding them compounds harm.
- Blame for errors falls unequally and can be used to scapegoat people.
Who is affected
- People affected by the error
- People blamed for it
Owners Filled
Steward
Nobody owns the concept; whoever produced a result answers for its errors.
Links to other meta-models Filled
parent
- Q29485 - registry parent class
- Q809898 - registry parent class
What else AI and robots need to interact with it Filled
Identity and identifiers required Filled
- Vercy registry: vr.tr.error-q1202128
- Wikidata: Q1202128 (https://www.wikidata.org/wiki/Q1202128)
Direct properties not applicable Not applicable
- Assessed bearer: Link to the particular claim, representation, result, observation or action and its version
- Reference basis: Truth condition, reference value, rule, specification or intended result, with applicable context
- Discrepancy form: False assertion, omission, invalid inference, numerical deviation, transcription discrepancy, procedural departure or domain-defined form
- Assessment status: Suspected, supported, contested, refuted or unresolved
- Quantitative deviation: Signed or absolute difference in the compared quantity's unit; relative difference only with an explicit denominator; unknown or not applicable
- Observed recurrence pattern: Isolated, recurring with consistent direction, variable across repetitions, mixed or undetermined
- Affected dependencies: Links to conclusions, records, decisions or actions that used the erroneous item
- Correction status: Unaddressed, proposed, applied, verified, unsuccessful or no correction required
Plane INF: no invented physical properties.
Recognition optional Filled
- A mishandled ground ball or dropped catch can indicate a possible error when the play appeared achievable with ordinary effort.
- An inaccurate throw associated with a missed out or runner advancement can indicate a possible throwing error.
- An E notation associated with a fielder in a scoring record identifies a recorded error, while the visible misplay alone does not establish the scoring decision.
Capabilities and actions required Filled
- Frame an alleged error as an explicit comparison against a stated reference.
- Test whether disagreement reflects error, differing conventions or insufficient evidence.
- Estimate discrepancy magnitude and investigate recurrence where the evidence permits.
- Trace which downstream claims or decisions require reassessment.
- Propose and verify a correction while preserving the original assessment and its history.
- Withdraw or revise an error attribution when its reference or evidence proves inadequate.
Hazards and failure modes required Filled
- Errors that propagate through systems that copy each other's outputs.
- Silent errors in data that are never detected.
Standards and interfaces required Filled
- Evaluation of measurement data, Guide to the expression of uncertainty in measurement, issued by the Joint Committee for Guides in Metrology.
Context of use required Filled
- In baseball, an error is recorded when an official scorer judges that a fielding misplay meets the applicable scoring criteria.
- Errors appear in game records and fielding statistics to account for certain missed outs or runner advances.
Sources Filled
- Wikidata item Q1202128: error - identity and sense of the item
- Wikipedia: Fielding error - general description of the item
Open questions
- Does registry entry vr.tr.error-q1202128 intend the broad concept of error or a narrower epistemic sense within INF.KNW?
- Which existing Vercy world model, if any, already owns this concept and should be linked instead of duplicated?
- Should erroneous actions be owned here or represented only through relations to action-plane models?
- Which domain conventions must be named to distinguish error, uncertainty, residual, defect and failure without imposing one discipline's terminology on another?
- How should the model represent cases where the governing reference is disputed, changes over time or is later shown to be erroneous?
Machine files
Provenance
thing registry research (pass 2) · unreviewed
Built from: models/things/publications/thing-q1202128/spec.json