The Reconstruction Requirement Framework
Understanding can no longer be verified through production. Every system that relies on it continues to function. This framework defines what remains measurable — and what does not.
This page is not an introduction. It is a reference.
It defines the terms, axioms, and constraints of the Reconstruction Requirement as a verification standard — the specific language through which the framework operates and through which its claims can be evaluated, applied, and built upon.
All definitions are canonical. All are released under CC BY-SA 4.0.
The Core Axioms
These are not opinions. They are structural consequences of what AI assistance is and what it has done to the relationship between the signals verification systems measure and the capability those systems were designed to certify.
Production can be simulated. Persistence cannot.
Where reconstruction is absent, verification does not exist. Only its appearance does.
A test that cannot detect absence cannot certify presence.
Before the novelty threshold, everything looks correct. At the threshold, only what is real survives.
If it cannot be reconstructed without assistance, it was never understood.
The Canonical Terms
Each term below is defined as it functions within this framework. Terms that exist elsewhere with other meanings are marked with their specific framework definition — which is the only definition that applies here.
Reconstruction Requirement The condition under which structural comprehension can be verified as genuine: that it can be rebuilt from first principles, without assistance, after a minimum of ninety days of temporal separation, in a context that differs meaningfully from the original acquisition context. The only verification standard whose validity AI assistance cannot compromise.
Structural Comprehension Understanding that persists independently of the assistance that may have been used to produce its appearance. An internalized model of why something is true — one that survives temporal separation, generates reasoning from first principles, and adapts to genuinely novel contexts. What genuine cognitive encounter with difficulty builds and what borrowed explanation never builds.
Borrowed Explanation Explanation that performs as structural comprehension while assistance is present and collapses when assistance is removed. Syntactically correct, semantically coherent, and structurally absent. The default mode of AI-assisted professional and educational performance — not because practitioners intend to borrow, but because borrowed explanation and structural comprehension are indistinguishable by every signal available under contemporaneous assessment.
The Signal Test Any assessment that measures what can be produced under conditions where AI assistance is available, accessible, or recently present. Every existing educational examination, professional certification, peer review process, and organizational competency assessment is a Signal Test. Signal Tests were reliable evidence of structural comprehension before AI assistance made production independent of structural comprehension. They are no longer reliable evidence of what they were designed to measure.
The Novelty Threshold The point at which borrowed explanation fails and only structural comprehension remains. The specific moment when a situation falls outside the distribution that AI-assisted performance covered — when established patterns fail, when the framework stops governing, when the correct response requires recognizing that the established reasoning has become wrong. The novelty threshold cannot be detected before it is crossed. It arrives with consequence, not with warning. Before the threshold, the difference between structural comprehension and borrowed explanation is invisible. At the threshold, it is complete.
In this framework, the novelty threshold is an epistemological term — the moment when the difference between genuine understanding and its simulation becomes consequential. This definition is unrelated to the legal concept of ”novelty threshold” in patent law, which concerns prior art requirements for patentability.
Verification Collapse The structural event in which the correlation between explanation quality and structural comprehension was broken by AI assistance. Verification collapse does not mean existing assessment systems stopped functioning. It means they continued functioning — accurately measuring signals that no longer correlated with the structural comprehension those signals were supposed to indicate. The measurement continues. The measurement’s validity does not.
Temporal Separation The minimum interval — ninety days at standard verification — between original acquisition of material and reconstruction testing. The mechanism that removes short-term memory, residual familiarity, and pattern recall from the evaluation. Not a difficulty parameter. The condition that makes the difference between structural comprehension and borrowed explanation observable.
The Gap The outcome of reconstruction testing in which structural comprehension does not return. Not a verdict. Not a failure in the conventional sense. Accurate information about what was never built — the specific, honest starting point for genuine formation. The Gap is more valuable than the comfortable certainty of contemporaneous assessment, because it reveals what contemporaneous assessment cannot: that the structural comprehension the credential claimed to certify was never developed.
The Constraint
This is not a proposal. It is a constraint.
When AI assistance makes the production of every signal of structural comprehension independent of structural comprehension itself, every verification system that measures those signals loses the validity that once justified its certification claims. This is not a gradual degradation that better Signal Tests could address. It is a structural decoupling — the specific condition in which the signals and the property they were supposed to indicate are no longer produced by the same cognitive process.
The constraint is this: the only verification that remains valid is the one that tests what AI assistance cannot provide in the human mind — the independent structural persistence that genuine cognitive encounter produces and that borrowed explanation never produces.
There are no domains exempt from this constraint.
Every other approach measures something. None of them measure structural comprehension. The Reconstruction Requirement measures structural comprehension. This is not a claim about relative rigor. It is a claim about what different measurements can and cannot detect.
The Consequence
Every domain that depends on genuine structural comprehension — medicine, law, engineering, AI development, governance, finance, education — is now operating with verification infrastructure that cannot confirm whether the structural comprehension it certifies is present or absent.
This does not mean all practitioners lack structural comprehension. It means no current verification system can determine which practitioners possess it and which do not.
The consequence is not incompetence. It is unverifiability — which produces the specific failure mode that the novelty threshold reveals: correct performance during normal operations, invisible deficit accumulation, and complete failure at the moment when structural comprehension is most required and least replaceable.
The Reconstruction Requirement is the instrument that closes this gap — not by making verification more rigorous within the existing Signal Test paradigm, but by restoring measurement of the specific property that Signal Tests can no longer reach.
Where this framework is not applied, verification is not occurring.
This framework is maintained at ReconstructionRequirement.org under CC BY-SA 4.0. All terms may be freely used with attribution. No entity may claim proprietary ownership of the framework’s definitions, conditions, or verification methodology.
ReconstructionRequirement.org — The verification standard AI cannot defeat
ReconstructionMoment.org — The test through which the standard is administered
PersistoErgoIntellexi.org — The protocol that formalizes the standard
TempusProbatVeritatem.org — The foundational principle: time proves truth
2026-03-26