1. Definitions
Conversion
An operational transfer of a concept, quantity, or procedure from a source system ($S_0$) to a target system ($S_t$).
Conversion Loss ($\Delta$)
The observable distortion, semantic decay, or operational error introduced during a conversion.
Critical Failure
A conversion loss that produces epistemic invalidity, operational breakdown, or safety-critical risk.
2. Inclusion Protocol
A case is included in the Conversion Index only if the point of failure can be isolated at the level of the conversion interface. The Index does not document general misunderstandings; it documents broken transfers between systems.
Each valid case must satisfy all of the following:
- Identifiable Systems
Source and target systems are distinct and explicitly defined. - Specified Conversion Operation
The mechanism of transfer (e.g., manual calculation, textual translation, software parsing, institutional rule-mapping) is documented. - Observable Impact
The conversion loss produces a detectable epistemic, operational, or safety-relevant consequence.
Purely hypothetical cases or failures without downstream effects are excluded.
3. Classification Architecture (Taxonomy)
Each case is classified along a fixed set of axes. This taxonomy is designed to support cross-domain comparison without collapsing domain-specific differences.
Axis A: Domain
Identifies the primary knowledge system in which the conversion occurs.
- Physical — measurement- and material-constrained systems (
- Computational — software, algorithms, formal abstractions
- Linguistic — natural language, translation, interpretive systems
One primary domain is required; secondary domains may be noted.
Axis B: Conversion Interface
Identifies where the transfer between systems occurs.
- Human Interface — manual calculation, judgment, translation
- Instrumental Interface — sensors, measurement devices, lab apparatus
- Software Interface — code, file formats, pipelines
- Institutional Interface — rules, conventions, policy frameworks
Axis C: Failure Mode
Identifies how the conversion degrades.
- Protocol Mismatch — incompatible rules or assumptions
- Silent Failure — output appears valid but is incorrect
- Lossy Abstraction — relevant structure discarded during transfer
- Semantic Drift — gradual meaning shift without explicit breakage
Impact Category
Each case is additionally tagged by dominant impact:
- Epistemic Distortion
- Operational Breakdown
- Safety-Critical Failure
Each case must have one primary classification per axis. Reclassification is permitted as evidence evolves.
4. System Boundaries
Prototype Scope
The current dataset is a curated pilot set intended to validate the classification framework rather than serve as an exhaustive catalog.
Interpretive Margin
Physical and computational failures are often binary; linguistic and institutional conversions involve interpretive judgment. Where applicable, cases are constrained through cross-referencing multiple sources.
Current Constraint
The Index currently lacks a computational method for quantifying semantic drift across large textual corpora. Addressing this requires NLP and digital humanities resources beyond the scope of the v1.0 prototype.type.
