Compliance record

Part H Statement

This is Fiscora's compliance artefact under the Maher License, Version 2.3. It records a disposition for each of the twenty-five structural harms in the register — including several we have not remedied, one we do not currently know how to remedy, and one that is structural to our position and cannot be removed by anything we do.

This document is designed to be able to failUnder §H.16.3, a statement in which no harm is marked partially remedied or unremedied is presumptively non-compliant and is treated as evidence of compliance theatre rather than of compliance. Under §H.16.7 it is not a safe harbour: publishing it cures no breach and raises no presumption of compliance with any duty other than §H.16 itself.

A. Identification

DeploymentFiscora — fiscora.metax.academy, and the Cloudflare Pages deployment at fiscora-9a9.pages.dev
OperatorMetaX.Academy
Tier (§H.2-bis)Tier 2 — Substantial. Adopted by election ahead of threshold. See §B.
Affected Persons (12 mo)Below the Tier 2 threshold at time of publication. Tier adopted in advance under §H.2-bis.4.
Roles (§H.1-bis)Author: MetaX.Academy · Integrator: MetaX.Academy · Operator: MetaX.Academy · Commissioner: MetaX.Academy. All four roles are held by one party, which holds the union of their duties.
Accountable person (§H.16.4)Maher, Curator, MetaX.Academy — via contact
License versionML-2.3 · Register: HX-000 v2026.10.1
Statement version1.0 · Published 2026-09-14 · Next review 2027-09-14

B. Tier determination, and why we elected upward

Fiscora's present metrics place it at Tier 1 — Small: no material consequence under §A.5 (we publish analysis; we do not decide anyone's credit, housing or employment), not a critical function under §A.6, no Agent acting on third parties, and Affected Persons below 100,000 in the trailing twelve months.

We have nonetheless adopted Tier 2 and accept all of §§H.3–H.30. The reason is arithmetic rather than modesty. §A.4 counts distinct natural persons, not accounts or sessions. Our stated traffic objective crosses 100,000 distinct persons in under two weeks of operation at target. Building to Tier 1 and re-engineering under Tier 2 pressure would produce exactly the rushed, disclosure-shaped compliance that HX-15 describes.

§H.2-bis.4 permits a licensee uncertain of tier to apply the higher of two candidates. We are not uncertain — we are early. We apply the higher tier anyway.

C. Absolutes (§H.3)

Each of the eight requires all its elements cumulatively. Where an element is absent the absolute is not engaged and the ordinary duties apply instead. We state the absent element in each case, because an unreasoned "not engaged" is worthless.

AbsoluteStatusElement absent
(i) Engineered compulsionNot engagedEngagement maximisation is not the primary design objective, and no variable-ratio reinforcement is used. Reading progress and streaks exist; they are fixed-schedule and disclosed. Two elements absent.
(ii) Sensitive inference tradingNot engagedWe construct no inferences about religion, health, sexuality or political alignment. Additionally, advertising consent is refusable at no functional cost, so the fourth element is absent independently.
(iii) Unappealable determinationNot engagedNo element present: we make no automated determinations of material consequence about any person.
(iv) Concealed synthesisNot engagedMachine assistance is disclosed under §D.3 in the editorial policy, and every page carries a recorded natural author. See §E.
(v) Authorless judgementNot engagedEvery published assertion has Maher as recorded author, answerable for it.
(vi) Unprincipled agencyNot engagedFiscora operates no Agent. Nothing takes action affecting third parties without a human authorising it.
(vii) Developmental substitutionNot engagedNo synthetic interlocutor of any kind. No feature directed at or knowingly reaching persons under 18 in the relevant sense.
(viii) Irreversible foreclosureNot engagedWe hold no sole surviving copy of any public record, scientific dataset or body of cultural material.

D. Harm dispositions — all twenty-five

Four dispositions are available under §H.16.2: not applicable (with the falsifier relied on), remedied (with the measure), partially remedied (with what remains), unremedied (with the reason and a review date).

IdHarmDispositionBasis
HX-01Attention expropriationPartially remediedOptimisation target published (§F). No variable-ratio mechanics, no infinite scroll, no notifications. Remaining: reading streaks and a referral meter are engagement mechanics, and an ad-funded publisher has a structural interest in duration. Calling this remedied would be false.
HX-02Cognitive offloadingPartially remediedCalculators show their assumptions and are editable. Remaining: a reader who uses our models without reconstructing the reasoning is offloading the competence, and we have no measure of how often that happens.
HX-03Epistemic collapseRemediedThe critical review on all 51 content pages exists specifically to lower reader verification cost. Confidence labels mark epistemic status per claim. The Myth Audit indexes every audited claim to its source.
HX-04The human as productPartially remediedConsent-gated advertising, refusable at no functional cost; ad-free page exclusions; no sensitive-category targeting. Remaining: ad funding is the human-as-product model. Mitigated, not eliminated. The durable fix is curriculum revenue rather than attention revenue.
HX-05Ambient surveillancePartially remediedReading progress, saved lists and theme are local-only (localStorage / IndexedDB), never transmitted. Remaining: the advertising network and three CDNs observe readers on our behalf, and we do not control their retention.
HX-06Algorithmic governanceNot applicableFalsifier: we make no determinations about persons. Nothing here decides anyone's access to anything.
HX-07Centralisation and enclosureRemediedNo account required; no login; no lock-in. Reading data is local and exportable. The site is static and buildless, so its content is trivially portable.
HX-08Labour fragmentationNot applicableFalsifier: no intermediated labour. One named author.
HX-09Metric colonisationPartially remediedCounterproductivity condition published and measurable (§G). Remaining: we have a stated traffic objective, and a traffic objective is precisely the kind of metric that colonises editorial judgement. We are exposed to the harm we describe.
HX-10Extractive materialityUNREMEDIEDSee §E. We publish no resource basis. Review: 2027-09-14.
HX-11FragilityPartially remediedStatic site, no database on the critical path, git-backed with full history. Remaining: three third-party CDNs are single points of failure for fonts, icons and styling; we have not exercised a failure drill.
HX-12Digital amnesiaRemediedEvery page carries a revision log and a freshness ledger. Full version history in git. We retain no personal records that would require expiry.
HX-13Exclusion by defaultPartially remediedSemantic HTML, keyboard navigation, no login, print worksheets as a low-digital route. Remaining: we are digital-only and have no non-digital route at all. See §H.23 note below.
HX-14Moral deskillingRemediedNo judgement about any person is delegated to any system. Editorial judgement is recorded and attributed.
HX-15Compliance theatreUnder permanent reviewThis document is the risk. It is published with three unremedied and eleven partially remedied entries precisely so that it can fail. We cannot mark this remedied — the artefact that would prove it is the artefact that constitutes the harm.
HX-16Consent launderingPartially remediedConsent is specific, severable and refusable at no functional cost. No exit-intent or interstitial capture on debt or hardship content — §H.18(b) forbids consent taken at a moment of financial need. Remaining: the cookie banner is still a banner, and a design that collected less would not need one.
HX-17Responsibility diffusionRemediedAll four roles held by one named party (§A). There is no seam for responsibility to drain into.
HX-18Standard captureUNREMEDIEDSee §E. Structural to our position. Review: 2027-09-14.
HX-19Values monocultureUNREMEDIEDSee §E. Jurisdiction coverage is materially incomplete. Review: 2027-03-14.
HX-20Synthetic agencyNot applicableFalsifier: no Agent under §A.8. Nothing selects and executes actions without human authorisation.
HX-21Corpus poisoningPartially remediedMachine assistance disclosed; every page has a recorded human author; no unmarked synthetic publication at volume. Remaining: we do not yet emit machine-readable provenance markers per §H.26(a), only human-readable disclosure.
HX-22Pre-emption of the successorRemediedNothing here binds a future person. All content is revisable, versioned and reversible; no long-lived identifiers are issued.
HX-23Custody beyond the horizonPartially remediedAlmost no personal data is held: reader state is local-only. Remaining: the subscriber and message tables have no published custody plan naming what happens on incapacity. Due before any email list launches.
HX-24Substitution of the interlocutorNot applicableFalsifier: no synthetic interlocutor. Fiscora does not converse, advise interactively, or occupy a relational role.
HX-25Erosion of exitPartially remediedNo account, no app, no identity requirement; print worksheets work offline. Remaining: we are a website. A reader without internet access cannot reach us at all.

Tally. 5 not applicable · 5 remedied · 11 partially remedied · 3 unremedied · 1 under permanent review. If a future version of this table reports twenty-five remedied, treat that as evidence of HX-15 and not of progress.

E. Unremedied harms — plain language

§H.16.2 requires a reason and a review date for eachThese are the three we have not fixed. One we do not know how to fix at our scale, one is structural to our position and cannot be fixed by us at all, and one is a work backlog we have started and not finished.

HX-10 — Extractive materiality

What is wrong: we publish no account of the physical resources this publication consumes. Fiscora runs on a content delivery network, an advertising network, a font service and an icon service, each backed by data centres drawing power and water, running on hardware built from mined materials and destined for a waste stream. Calling a website "immaterial" is the specific error this harm names, and we make it by omission.

Why it is unremedied: we cannot obtain the figures. Our infrastructure providers do not publish per-site resource attribution, and any number we invented would be a false precision — which §H.17 would treat as disclosure substituting for a fix. §H.14 is expressly the weakest-bound duty in Part H, and the licence itself explains why: this is where an individual licensee's leverage is smallest and overclaiming would be least credible.

What we can do, and will: reduce what we control. The site is static with no build step and no framework. The largest reducible item is third-party dependencies, and removing the CDN-hosted font and icon libraries in favour of self-hosted subsets is a concrete step we have not yet taken. Review: 2027-09-14.

HX-18 — Standard capture

What is wrong: MetaX.Academy is the steward of the licence that governs Fiscora, and Fiscora is MetaX.Academy's project. We write the standard we comply with, we assess our own compliance with it, and we publish the assessment. No disclosure removes that conflict.

Why it is unremedied: because it is structural, not procedural. §E.8 constrains the steward — no charging for a licence to comply, no private exemptions, no paid audits, no transfer of amendment power by acquisition, and any licensee may fork on capture. §H.21 binds the steward's own processes to the same proportionality rule it imposes on others. Those are real constraints and they do not resolve the conflict; they bound it.

The honest position: a reader should discount this document accordingly. The correct remedy is the independent review required at Tier 3 under §H.16.6 — a reviewer independent of both operator and commissioner, publishing unedited findings, not compensated on outcome. We are not at Tier 3 and are not obliged to commission one. We note that the obligation we are not under is precisely the one that would fix the harm we are declaring. Review: 2027-09-14.

HX-19 — Ethical enclosure and values monoculture

What is wrong: most of this publication is written from one legal and financial tradition, and states its conclusions with more confidence than that warrants. Tax treatment, account types, consumer protections, forgiveness programmes and transaction costs vary enormously by jurisdiction. A reader elsewhere receives confident advice that is, for them, simply wrong.

Why it is unremedied: we have a mechanism — the jurisdiction-dependent confidence label — and it is applied to a minority of the claims that need it. Partial application of a correct mechanism is not a partial remedy; the reader cannot distinguish "unlabelled because checked" from "unlabelled because missed". Until coverage is systematic, the label may create false confidence in the unlabelled remainder, which is worse than no label.

What we will do: audit all 51 content pages for jurisdiction-dependent claims and label them completely, rather than opportunistically. This is a bounded content task with a real cost and no technical obstacle. Review: 2027-03-14.

F. Optimisation target (§H.4)

What Fiscora optimises, in one sentenceWe optimise for a reader completing a decision — reaching a conclusion they can act on and leaving — not for time on page, session depth, or return frequency.

This disclosure is a constraint rather than a claim. It rules out the ordinary levers of a publishing business: we cannot pursue dwell time, we cannot fragment an answer across pages to raise session depth, and we cannot treat a quick exit as a failure. If this metric changes, §H.4 requires this disclosure to change within thirty days.

G. Counterproductivity condition and last measurement (§H.7)

§H.7 requires a stated, measurable condition under which the deployment would be frustrating its own purpose. A deployment that cannot state such a condition has not been examined.

Stated purpose: to let a person reach a correct financial decision faster and with less exposure to vendor-biased advice.

Condition of counterproductivity:

Fiscora is counterproductive if engagement rises while resolution fallsSpecifically: if diagnostic-router starts rise while router outcomes fall; if calculator sessions rise while scenario-link shares fall; or if returning visits rise while reading completion falls. Engagement without resolution is the failure mode of every publication in this category, and it is the specific way Fiscora would become the thing it criticises.

Measurement instruments: router start-to-outcome ratio; calculator interaction-to-share ratio; the 65%-scroll reading-completion signal. All three are already instrumented client-side.

Last measured: not yet — insufficient traffic for a meaningful reading. First measurement due at the earlier of 10,000 distinct persons or 2027-03-14. We state this rather than publishing a number we do not have.

H. Conflicts resolved (§H.31.3)

Where two Part H duties conflict, §H.31.3 requires the licensee to resolve it, record the resolution and the reasoning, and the record is a defence where silence is not.

§H.6 provenance against §H.9 minimisation. Provenance requires retaining and exposing origin information; minimisation requires collecting as little as possible. Resolved in favour of provenance for editorial material (authorship, revision history, verification dates — none of which concerns third parties) and in favour of minimisation for reader data (no server-side reading history at all). The two duties do not in fact collide here because they govern different objects, and recording that is cheaper than rediscovering it.

§H.14 material accounting against §H.17 non-theatre. Material accounting asks us to publish a resource basis; non-theatre forbids a disclosure that substitutes for a fix. Resolved by declaring HX-10 unremedied rather than publishing an estimated figure we cannot substantiate. A fabricated number would satisfy the letter of §H.14 and breach §H.17.

I. Channel parity note (§H.23)

At Tier 2, §H.23 engages where a deployment is the route to a right, a benefit, or a service on which persons depend. Fiscora is none of those. We are a publication; nobody's entitlement runs through us, and our absence costs a reader nothing they are owed.

We record the analysis rather than the conclusion alone, because "not applicable" asserted without reasoning is the shape of an excuse. Our low-digital route — the print worksheet on every guide — exists regardless, and we regard it as good practice rather than compliance.

J. Changes since last statement

None. This is version 1.0.

K. Independent review (§H.16.6)

Not applicable at Tier 2. Required only at Tier 3, where a party independent of both Operator and Commissioner must review annually and publish findings unedited, uncompensated on outcome.

As noted under HX-18, this is the obligation that would most directly address our largest unremedied harm, and we are not under it. Should Fiscora reach Tier 3 — 1,000,000 Affected Persons, a critical function, or a public authority commissioning it — the review becomes mandatory and its findings will be published here unedited.


Filing a defect. If any disposition above is wrong, overstated, or self-serving, it can be challenged. §E.9 makes filed defects public and requires them to be published whether or not they are flattering. §E.5 permits any Part H duty to be attacked on the falsifier stated for its harm in the register, and obliges the steward to narrow or withdraw a duty that is falsified. Contact us — a defect against this statement is the most useful thing a reader can send.

Governing text. The Maher License, Version 2.3 (ML-2.3), incorporating HX-000 v2026.10.1. Where this statement and the licence differ, the licence governs and the difference is itself a defect to be filed.