Boundary
Non-claims
What this site does not claim, and why. Download the register.
| ID | Not claimed | Why |
|---|---|---|
| NC-01 | Insurance, underwriting or coverage | Neverthought is not an insurer, broker or managing agent. Nothing here is an offer, quote, binder or coverage determination. |
| NC-02 | Actuarial advice or opinion | Actuarial epistemics is a research lens. No credentialed actuary has reviewed this site. |
| NC-03 | A risk score, price, premium or rating | The appraisal reports checks, archetype readings and gaps. It deliberately produces no number that ranks or prices risk. |
| NC-04 | Loss prediction | No loss data has been linked to any determination or archetype. |
| NC-05 | Validated risk archetypes | All ten archetypes are proposed. None has been measured in a real domain. |
| NC-06 | Real-domain crossing classes | All crossing classes and examples are synthetic. No insurer, auditor, model-risk reviewer or regulator has reviewed them. |
| NC-07 | Insurance-grade evidence | Receipts are computed in the browser and chained, but anyone holding the chain can recompute every hash, so it detects tampering only against a head hash kept or shared elsewhere. There is no custody, no trusted time source, no external anchor and no durable append-only store. Optional signatures use a browser-held key. |
| NC-08 | Independent replay | Replay recomputes this site's own determination from the same inputs. |
| NC-09 | Verification of declared facts | The gate checks what a proposal declares, such as trust-root independence and evidence hashes. It does not verify those declarations against the world. |
| NC-10 | Physical laws of risk | Physics labels mark which quantities are formal measures and which names are analogies. No physical law is claimed for organisations or AI systems. |
| NC-11 | Compliance, certification or legal conclusions | Determinations are about declared crossings under an illustrative policy, not regulatory status. |
| NC-12 | Data collection | Everything runs in your browser. Proposals, receipts and keys are not sent anywhere. |
| NC-13 | Probabilities or near-miss judgements | Headroom and flip statements compare declared inputs with declared limits. They are not probabilities, and no cut-off turns small headroom into a finding. |
| NC-14 | Anonymous insurer views | The insurer view omits names, sources and proposal text, but its accumulation keys are hashes of model, vendor and data-source names. Anyone who can guess a name can hash it and match it. |
| NC-15 | Recognition by an insurer or regulator | No insurer or regulator recognises these receipts. Until one does, they carry no price. Securing recognition is an institutional project measured in years, and if Neverthought acted as certifier it would inherit liability and be paid by the party it assesses. |
| NC-16 | A toll, risk charge or premium | These are three different prices: a charge for passing the gate, a charge for the risk carried, and an insurance premium. This site computes none of them, and the cover trigger it evaluates is not a coverage decision. |
| NC-17 | An adequate consequence model | No class has passed a structural test, so none is supported. A structure status of untested means no observation has challenged the model yet, not that it contains every relevant path. No observed effect is not no effect; no modelled path is not no possible path. |
| NC-18 | Detection of structural omissions in general | The site flags a consequence record whose recorded path is missing or undeclared, and effects on undeclared systems. It does not infer that a path is missing from observations it is not told about. The refund fixture's R29 (PASS, narrow: frozen event kinds and an independent provider-fee challenge) shows that once: an unaccounted event kind in an independent provider log, checked against paths frozen before the observation. Effects absent from every log, or hidden inside kinds already accounted for, are not detected. |
| NC-19 | Domain validation of a user's class | Admission checks the schema, the source's I2 rules and, with a warranted example, the mutation battery. It does not show the class suits its domain or that its consequence paths are complete. |
| NC-20 | Completeness or authenticity from a manifest | An evidence pack's manifest shows its files are the ones listed. It lives in the same bundle, so it cannot show the set is complete, the inputs true, or who produced it. |
| NC-21 | Hosted workspaces, accounts, subscriptions, human sign-off or insurer recognition | All proposed. Pilots are a proposed paid offer. Payment would change access only, never a verdict, warrant, scientific standing, certification or cover. |
| NC-22 | Certified discoveries or causal findings | A promoted claim passed a class's checks. The science example maps a review of software-governance experiments whose methodological inference was not adjudicated. |