OPEN · WORKINGNOTES-ssi-six-hats.md

Evaluation — the SSi six-hat run (a strong failure)

The second published Six-Hat evaluation, and the first to come back decisively critical — of the author's own company. Published in full, unedited, because a Living Epistemic Work cannot cherry-pick its wins, least of all its author's. Run 2026-06-21 (18 agents, swarm-per-hat + Blue synthesis). Source proposal: NOTES-ssi-currency-surface.md.

Human reading (Tom): a strong failure — the proposal as stated (SSi issues a local currency) does not survive; what survives is the reframe (SSi as a trusted verification-and-legitimacy layer on someone else's regulated rail, owned by an Ogwen body, bound to Welsh usage), and even that is gated on a conversation in the valley nobody has had. "A strong failure is just closing certain doors, for now." As Fuller would say: now we're learning. The two human seats — Red gut, final Blue verdict — are deliberately left open, as on every run.

BLUE HAT — framing the session.

The question on the table. Should SaySomethingin (SSi) — a Welsh-language learning company — become the delivery + ledger + legitimacy surface for a local credit/currency system serving a DNA-economy cell in North Wales (Ogwen/Gwynedd)? And if so, in what concrete form? "Delivery" = the app people touch; "ledger" = the record of who earned/spent what; "legitimacy" = the trust and political cover that makes a credit feel real and lawful.

What is being evaluated. Not whether local currencies or language-preservation are good in the abstract — but whether this specific company layering this specific financial function onto its existing app is a viable, life-increasing configuration. Three things are bundled and must be kept distinct: (1) a technical/regulatory claim (can a credit be backed, spendable, lawful, non-inflationary), (2) a strategic-coherence claim (can two different currencies live on one surface), and (3) an identity claim (is "exemplar of local sovereignty" an asset or a mission-drift distraction).

The fitness function. CE's viable objective via Fuller: prefer configurations that bring greater life to all and less to none, without ecological offence — judged as a no-regret move over horizons (Δω over the reachable option-space), not a scalar to maximise. Specific CE claims in play, each itself contestable: a living language is an irreversible option-space (preserving it is no-regret); language-preservation is "frozen capability" the market doesn't price; a local currency can fund the Δω-positive work markets ignore; routing credit through SSi's surface routes around the financial chokepoint.

The three sharp questions (every hat should return to these):

  1. Backed / spendable / below-the-line / not-just-points — is the credit a real claim on real value, lawfully issued, and inflation-resistant?
  2. Two-market coherence — grant-backed learner-reward currency (global diaspora, time-in-app) vs Δω-backed local-value currency (heartland, care/retrofit/energy hours): one surface, two layers — coherent, or a split that tears the company?
  3. Exemplar — asset or distraction — is "company as local-sovereignty exemplar" real strategic capital or a seductive detour from teaching Welsh?

Marching orders for each hat:

Keep to your mode. Insight over intensity. White is next — establish what we actually know before anyone argues from it.

WHITE HAT — facts, gaps, and confidence labels for SSi-as-currency-surface. Facts only; the load-bearing question is Q1 (backed / spendable / below-the-line / non-inflationary), and most of what's needed there is known law against an undesigned choice.

A. What SSi is (web-verifiable)

B. The terrain the currency would serve (verifiable)

C. The regulatory line (Q1 — load-bearing). Two separate lines; do not collapse them.

FCA / e-money line:

HMRC / income line (the half a casual reading misses):

D. Base rates of the artefact class (verifiable, sobering)

E. Three-way ledger

KNOWN (verified): SSi is real; Welsh decline + Gwynedd heartland + the publicly-acknowledged Cymraeg-2050 gap are as described; LNE + €1M notification trigger + contested loyalty-points boundary; HMRC taxes earned/expected service tokens and barter as income; local-pound base rate ≈ failure; B2B mutual credit = the rare survivor.

ASSUMED (proposal-not-fact, needs evidence before build): SSi's user numbers and diaspora reach; the app's ability to be a ledger; any Cymraeg-2050 funding line structured as a per-learner reward subsidy SSi could draw; founder/governance fiduciary appetite; that "political capital" is real and transferable to a financial product; that "two markets, one surface" is operationally separable in one codebase/governance.

UNKNOWN (no one has the answer yet): ward-level Ogwen demographics (basis of the split); whether a closed Ogwen network can stay under €1M while being large enough to matter; how the heartland currency interacts with means-tested benefits; whether Welsh Government would bless or recoil from a language company issuing money.

Two factual corrections for downstream hats:

  1. The dossier bundles "backed + spendable + below-the-line + non-inflationary" as jointly achievable. The verified e-money, tax, and base-rate facts show they are partially mutually exclusive: the closer to a real spendable claim on value, the closer to e-money + taxable income + the Bristol failure mode; the survivors (time bank, B2B mutual credit) buy safety precisely by not being a consumer spendable currency. This trade-off frontier — not any single design — is the real object of evaluation. Black should press it; Green should design against it.
  2. "Below the regulatory line" is two lines, not one: an FCA/e-money line (navigable via LNE, with a real €1M ceiling) and an HMRC/income line (which the heartland work-for-credit design crosses by construction). They must be reasoned about separately.

Sources: SaySomethingin (site/mission/history) · Nation.Cymru (launch, expansion, Cassidy) · Welsh Language Commissioner 2021 Census · Cymraeg 2050 (gov.wales) · Senedd Research (million achievable?) · FCA (E-Money/PSD approach; Limited Network Exclusion) · Mondaq/Dentons (loyalty points as e-money) · GOV.UK EIM44040 + LITRG (vouchers, volunteers, benefits-in-kind) · Tax Adviser (VAT and barter) · GOV.UK trivial-benefits exemption · Wikipedia/CapX/BristolWorld (Bristol Pound) · LSE / Monneta / P2P Foundation (Sardex, mutual credit).

YELLOW HAT — the optimistic case, truest in mode.

White handed Yellow its sharpest gift, so seize it rather than wish it away: the species that survives is closed-loop B2B mutual credit; the species that dies is the consumer earn-and-spend token. The optimistic claim is not "local currencies work" — they usually don't — it is that SSi uniquely holds the asset whose absence is the documented cause of their death, and that the survivable, fundable first win is far smaller and nearer than the full proposal suggests.

1. The failures were trust-and-distribution failures, not monetary-design failures — and that is exactly the failure SSi is structurally built not to have. Read White's failure cluster (Bristol, Brixton, Totnes): plateaued adoption, "elitist club" perception, no warm distribution channel, admin burden exceeding benefit, clunky UX. Every one is a cold-start failure. SSi does not have a cold start. It holds the one input the survivors needed and the failures lacked: pre-existing, affect-laden membership — a self-selected group who have already paid in money and months of daily effort to affiliate with the survival of a place and its language. That is a community pre-sorted for the exact motive a place-currency runs on. A fintech spends years and millions manufacturing that enrolment and still smells like an "elitist club"; SSi starts with it. (Confidence: high that warm distribution is the decisive differentiator against the failure cluster; medium on magnitude — White flags user numbers as proposal-not-fact, and the warm community may sit mostly in the diaspora, outside Ogwen.)

2. SSi is rare in owning both ends of the trust-to-ledger bridge. A mutual-credit circuit (WIR, Sardex) needs two things the failures lacked: a concrete shared problem and a trusted convener nobody suspects of self-dealing. Ogwen has the problem (Δω work the market won't pay for — care, retrofit, language transmission); what it lacks is a convener with distribution and legitimacy who is not a bank and not a council. A fintech builds the ledger and never earns the trust; a community group has the trust and never builds the ledger. SSi credibly has both. (Confidence: high on the structural argument; medium that the app can actually host double-entry balances/audit/dispute — White flagged this unverified.)

3. The strongest first layer dissolves White's "partially mutually exclusive" trade-off — because it is backed by SSi's own near-zero-marginal-cost product. A Cymraeg-2050-aligned learner-reward credit — earned for learning-hours, redeemable in a deliberately narrow Welsh-cultural basket (SSi tiers/content, Welsh books, eisteddfod entry, partnered Gwynedd cafés/bookshops) — sits on the safe side of BOTH lines White drew: non-service-linked and gift-like (trivial-benefit/volunteer zone on HMRC; the heartland work-for-credit layer is what crosses the income line — the learner layer does not), and plausibly LNE-shaped and years under the €1M ceiling on FCA. With no buy-in and no cash-out it stays cleanly out of e-money and out of taxable cash. And uniquely: SSi can honour the credit in the abundant thing it already makes — lessons, tiers, content — so it need not be sterling-solvent to back it. A credit redeemable in your own abundant output, with issuance capped against known redemption capacity, is the one design that is simultaneously backed, spendable, below both lines, and inflation-resistant. (Confidence: high that this layer is uniquely SSi's and regulatorily cool; medium that a per-learner subsidy could be structured for SSi to draw — funding mechanism unverified.)

4. "Route around the chokepoint" is literally true here — and its sharpest form is the grant-disbursement chokepoint, not the banking one. Because the strong layer is a claim on SSi's own product, SSi needs no bank, no e-money licence, no payments giant to start — the chokepoint-routing is real rather than aspirational, available precisely because it makes a near-zero-marginal-cost digital good. The deeper Δω, though, is routing around the grant chokepoint: language-transmission and care are frozen capability the market won't price, and the normal way to pay for them (grant → council → programme → acquittal paperwork) is slow, lossy, demoralising. If Welsh Government can fund outcomes — learner-hours, care-hours — as credits flowing directly to the people doing the work, with the ledger as the acquittal trail, that is a genuine efficiency Δω the market ignores. In this frame the currency is mostly a better disbursement-and-audit rail for money that already wants to flow — which also lowers the regulatory temperature (structured grant-delivery, not new money). White's verified Cymraeg-2050 shortfall (538k and falling against 1M, ~42k behind trajectory) makes a government visibly behind its own flagship target a credible buyer for exactly this. (Confidence: medium-high on the structural logic; the funding relationship is unverified.)

5. The two-market split is an asset, not a company-tearing fork — because it is a sequence, not a simultaneity. Yellow's optimistic read of Q2: don't launch both. Build the learner-reward layer first — global, grant-backed, regulatorily cool, self-backing — and let it stand up the ledger, the UX, the audit trail, and the institutional trust with Welsh Government. Then the heartland Δω layer (which White correctly flags crosses the HMRC income line by construction) gets built on a surface that has already proven it can move value cleanly and account for it, and built as the survivor species: inter-provider mutual-credit clearing among Ogwen providers (the retrofit co-op, the care collective, the energy-warden scheme clearing hours against each other, interest-free, sterling-denominated, settled in services not cash, SSi supplying the ledger and the legitimacy badge — "operated on the SSi surface, in service of Cymraeg 2050"). Easy-and-fundable first de-risks hard-and-real second.

6. The exemplar effect is the multiplier, and it lands harder for a language company than almost anywhere. Fuller's point — the working model enrols more than the argument — is sharpest here because the thing SSi teaches is itself the proof. A community that revived its language by paying its own people to transmit it, in its own credit, narrated in the very language it preserved, is not an argument for local sovereignty — it is a working instance of it. That makes "learn Welsh" mean "join a community that runs its own abundance," a far stronger enrolment hook than "learn a heritage language." And because SSi already runs ~40-language ambitions on one engine (White: medium), a working Ogwen credit is not a one-off — it is a replicable module that travels to Cornish, Irish, Basque, Māori the same way the teaching engine already does. The option-space opened by one credible exemplar dwarfs the local circulation it generates, and that distribution channel is itself rare and hard to manufacture. The exemplar is not a distraction from the mission; at its best it is the mission's most vivid demonstration. (Confidence: high as a strategic-narrative claim — this is Yellow's mode; real if and only if the first win is genuine, a condition Black/Green should hold it to.)

What a credible first win actually looks like (concrete, against White's base rates): not a Bristol-Pound consumer token. A grant-backed Welsh-cultural learner-reward credit — earned for learning milestones; redeemable only in a chartered, narrow circle (SSi tiers/content plus a handful of opted-in Welsh-cultural providers); no buy-in, no cash-out; issuance capped against known redemption capacity (no inflation); squarely in trivial-benefit + LNE territory, years under €1M, crossing neither line. This single move proves the ledger, banks the political capital, and tells the true story — without touching the taxable-income heartland design at all. Months, not years; thousands of pounds of exposure, not millions; reversible if it stalls. The heartland mutual-credit circuit is Phase 2, designed as inter-provider clearing, entered only once the surface has earned it.

Yellow in one line: the proposal's most exciting half is its most dangerous half, and its safest half is also its most fundable and most mission-aligned — an unusually kind risk gradient. The largest near-term value is a self-backing, chokepoint-routing delivery-and-audit rail for the language-transmission Δω the market won't price — worth doing even if Phase 2 never ships, with the heartland currency held as the option it keeps open, not the bet it makes today. SSi is plausibly the one actor on Earth positioned to make the safe half work beautifully — and the safe half is already worth doing.

BLACK HAT — synthesis (merged from three independent takes).

I attack this configuration — SSi bolting a credit-and-ledger onto a Welsh-teaching app for Ogwen — ranked by lethality, each wound named to the CE gate it fails. Yellow built the strongest version (phased, learner-reward-first); I attack that, not the strawman consumer token. The discipline: the phasing hides the wound rather than healing it.

WOUND 1 (mortal) — The heartland layer crosses the HMRC income line by construction, and it crosses it for exactly the people who can least afford it. CE gate failed: less-to-none. Pay an Ogwen carer or energy-warden a redeemable credit for regular, expected, service-linked hours and you have created taxable earnings, likely NIC-bearing, that interact with the Universal Credit taper. The modal human outcome: a carer earns 40 care-credits, spends them, and months later UC is clawed back or sanctioned because she received undeclared "income" — income SSi minted and told her was a community reward. The feature that makes the credit Δω-real (real value for real work) is identically the feature that makes it taxable income that can strip a vulnerable earner's benefits. This is arithmetic, not narrative; a configuration that imposes a fiscal/benefits penalty on the lowest-income participants for doing the very Δω work it exists to honour fails less-to-none at the level of the books. Yellow's "Phase 2, build it later" does not dissolve this — it defers the only layer that delivers the heartland Δω at all. (Confidence: high on the tax/benefit law — White verified it; medium-high that the design as stated walks straight in.)

WOUND 2 (mortal) — The redeemability trilemma: the credit cannot be backed, spendable, AND below both lines at once — and Yellow's "back it with SSi's own product" hides an unfunded sterling liability. CE gate failed: viable-objective / no-regret. Two failures fused into one, because they are the same dial:

WOUND 3 (severe) — Two opposite monetary policies on one balance sheet and one brand. CE gate failed: coordination-bounds-reachability. The learner layer wants maximum issuance (more learning = more credits = mission win). The heartland clearing unit needs scarcity and trust (it holds value only if not flooded). If the two credits are fungible, the diaspora's grant-pumped issuance inflates the heartland's value-unit; if they are not fungible, "one surface" is just two unrelated apps sharing a logo and the exemplar story ("a community that runs its own abundance") quietly dies. There is no dial setting that satisfies both. The split is not two layers on one surface — it is incoherent before any regulator touches it, unless severed, and severing kills the exemplar. (Confidence: high on the structural contradiction.)

WOUND 4 (severe) — Right community, wrong geography: the warm-distribution asset does not reach the layer that carries the Δω, and SSi has the wrong standing in Ogwen. CE gate failed: reachability + collective-ownership. Yellow's headline asset is "pre-sorted, affect-laden membership." But that community is learners, disproportionately diaspora — and the heartland currency must circulate in Ogwen, among native care-workers, retrofitters and energy-wardens who are mostly not SSi users (the proposal itself concedes learning Welsh "does not fit here"). The warm crowd cannot spend a place-currency in a place they don't live; the place that needs the currency has no warm relationship with SSi. Bristol at least had Bristolians. Worse than neutral: in Ogwen, SSi is the company that teaches outsiders the language natives already speak — precisely the wrong credential to arrive as issuer of the local money. The "elitist club" perception White documented as a currency-killer becomes "incomers running the local economy." And nothing in the dossier shows Ogwen asked for this — an externally-convened currency the heartland didn't request is closer to benign colonisation than to local sovereignty, and the exemplar inverts into a cautionary tale. The two-sided market still has to be built cold, café by café — where SSi brings less merchant-recruitment muscle than a town council. (Confidence: high on the structural mismatch; medium-high on the sociology; the demand gap is an unverified-but-damning absence, and needs the ward-level Ogwen data White flagged.)

WOUND 5 (severe) — Phasing is illusory de-risking: Phase 1 validates nothing Phase 2 needs, and ratchets commitment. CE gate failed: coordination-bounds-reachability. Yellow's cleverest defence — "build the safe learner layer first, let it stand up the ledger and trust, then build the heartland layer on proven rails." But a learner-reward coupon redeemable in SSi's own product needs no real ledger — it needs a balance counter SSi presumably already has for tiers. The genuinely hard infrastructure (double-entry, multi-party balances, dispute resolution, settlement, KYC, audit against benefit-interaction) is demanded only by the deferred heartland layer. So Phase 1 proves none of Phase 2's real failure points: "do the easy thing that doesn't test the hard thing, then do the hard thing." Worse, you reach Phase 2 with public commitment and reputational skin in the game — a ratchet that makes the hard layer harder to cancel when it proves genuinely difficult. An option turned into an obligation. (Confidence: medium-high; rests on the likely-but-unverified claim that the app already counts balances.)

WOUND 6 (serious) — Grant-dependence builds the chokepoint into the foundation. CE gate failed: reachable-set durability over horizons. Yellow's strongest funding story makes the currency a disbursement rail for grant money, so its backing is a political line-item that White showed is already behind its own target (538k vs 1M, ~42k short). A flagship missing its numbers is exactly what gets restructured, rebranded, or cut at the next Senedd budget or change of government — which cuts both ways: a nervous government may fund a delivery rail, or may retrench to safe, legible spending and run a mile from "language charity issues currency." When the grant flexes, the credit's backing evaporates and it inflates into the worthless points Q1 warns of — now with people's earned care-hours inside it. Coupling the company's surface to a single politically-exposed funding line reduces SSi's reachable futures: one budget decision holes both the credit and the distraction-tax already paid. That is the opposite of a no-regret move. (Confidence: medium-high; the grant relationship is unverified — which is itself the point.)

WOUND 7 (serious, asymmetric) — Reputational contagion stakes an irreversible asset on a near-uniformly-doomed one. CE gate failed: irreversibility / no-regret. White's base rate for the consumer half is near-uniform failure. Weigh the asymmetry CE itself insists on: the upside is a real-but-bounded circulation in one valley, capped and replicable elsewhere; the downside contaminates SSi's actual irreversible asset — multi-generational community trust in the language-survival mission. When the credit stalls (the documented pattern: plateaued adoption, suppliers won't accept, everything converts back to sterling), or when an FCA €1M-notification letter or an HMRC reassessment of participants lands, the headline is not "fintech pilot fails" — it is "Welsh-language company loses people's money / got grandmothers taxed / Cymraeg 2050 vanity-currency collapses," running in the exact community whose trust is SSi's rare asset. The regulatory risk and the reputational risk are not symmetric: a fintech that trips e-money rules pays a fine; a Cymraeg-2050-adjacent language brand that causes benefit sanctions gets a betrayal narrative aimed at the language itself. You do not stake the frozen-capability asset to fund the frozen-capability work — that is precisely the trade CE's no-regret criterion exists to forbid. (Confidence: high on the asymmetry; medium on failure probability for the specific phased design.)

WOUND 8 (chronic) — Fiduciary and mission dilution: the org cannot hold both shapes. CE gate failed: pattern-intelligence-constraint. Pedagogy iterates fast and forgives error; ledgers demand reconciliation, dispute process, fraud handling, AML posture, benefit-interaction advice for vulnerable earners, and zero tolerance for "oops we double-credited." Every hour Cassidy (Head of Course Creation) and Jones spend on ledger UX, FCA counsel, merchant onboarding and benefit edge-cases is drawn from the same finite pool that teaches Welsh — SSi's actual Δω and actual revenue. White flagged founder/governance fiduciary appetite as unverified; and the evaluator of this very proposal is internal (Cassidy) — a conflict to name, not hide. Even if every regulatory gate is passed, the currency taxes the core mission's bandwidth indefinitely. (Confidence: medium; depends on org capacity, unverified.)

Verdict on the three sharp questions.

The pattern across all eight wounds: the features that make the credit real (backed, spendable, service-linked, Δω-bearing) are the same features that make it regulated, taxable, cold-start-fragile, and reputationally radioactive. Yellow's "safe half" survives my attack only by becoming so safe it is barely a currency — a closed, non-redeemable, no-cash-out reward for SSi's own product, i.e. a loyalty-points scheme wearing a sovereignty narrative, delivering approximately none of the heartland Δω the proposal exists to capture. The dangerous half is where the value is; the safe half is where the value isn't. That is not Yellow's "kind risk gradient" — it is the proposal's central contradiction.

Black in one line: the proposal asks SSi to stake its one irreplaceable asset — language-survival trust, a genuine no-regret Δω — as collateral on an artefact class with a near-uniform failure record, where the only configuration that delivers the promised heartland value is the one that taxes the poor, draws the regulators, and corrodes the mission if it fails. Build nothing until someone names two things the proposal never does: the sterling pool that clears merchant redemptions, and the benefits-safe-harbour that protects the carer. Absent both, this is not a currency — it is an uninsured promise written on SSi's reputation. The single reframe that would answer most of this is Green's to make: stop trying to issue money, and ask whether SSi should be the trusted legitimacy-and-audit layer on top of someone else's already-regulated rail — a credit union, an existing LETS, a time-bank.

GREEN HAT — synthesis. The discipline Black set: name the sterling pool that clears merchant redemptions and the benefits-safe-harbour that protects the carer. Green does not argue Black down — it finds configurations where those two bills never come due, because the design never raises them. The unifying reframe: stop asking "how does SSi issue money?" and ask "what can only SSi do?" The answer is certify Welsh-language Δω, measure Welsh usage, channel diaspora affect-and-money into heartland care — none of which requires SSi to be issuer, backer, or bank.

MOVE 1 — Invert the carrier: SSi sells legitimacy-and-audit, never money. The load-bearing move. SSi never mints, holds float, or clears redemptions. It supplies the one thing it owns — a trusted, Welsh-native front-end, narration, learner pipeline, and Cymraeg-2050 audit trail — bolted onto an already-regulated rail underneath: a Gwynedd/North-Wales credit union (FCA-regulated, deposit-protected) or an existing time-bank/LETS hour-ledger. The regulated partner holds the sterling pool and carries the e-money/tax/fiduciary liability it is already authorised to hold; SSi holds the UX and the trust badge. This deflates Black's Wounds 1, 2, 6, 7, 8 at a stroke: the unfunded sterling hole becomes the credit union's funded, protected pool; the benefits question becomes the time-bank's existing carve-out; the bandwidth drain on Cassidy/Jones is the partner's; a failure is "a pilot ended," not "the language company lost grandmothers' money." SSi keeps its rare asset and sheds every part it is worst-placed to hold. Wall that may not yield: a partner at Ogwen scale willing to move at SSi's pace and cede the UX — a deal to strike, not a law of physics; needs White's unretrieved local-institution scan. (Confidence: high that this is the lowest-regret option; medium on partner availability.)

MOVE 2 — Protect the carer by routing the heartland layer through organisations, never individuals — and build the survivor species. Black's HMRC/UC wound attaches to a person receiving service-linked income. Remove the person from the credit-receiving role. The individual carer is employed and paid in sterling by the care collective under normal PAYE/NIC — benefits-visible, declarable, lawful. That is the safe-harbour Black demanded: the carer is a normal employee, not a credit-earner. The credit layer operates only above the people, as inter-provider mutual-credit clearing (the WIR/Sardex survivor species White identified): the retrofit co-op owes the care collective in cleared service-hours, denominated in £ for accounting but never converted to cash, interest-free, settled service-against-service. No grandmother ever holds a taxable token; the Δω still flows and still gets paid. This is the single design that keeps the heartland layer real and keeps less-to-none intact. (Confidence: high on the structural fix; the non-taxable footing holds precisely because conversion-to-cash — the HMRC trigger — is designed out.)

MOVE 3 — Kill the unfunded-sterling-liability by removing redemption, or by backing it with already-committed procurement. Two complementary routes to answer "which sterling pool clears merchants?":

MOVE 4 — Resolve the two-policy contradiction by non-fungibility plus a one-way valve — and let the diaspora fund the heartland as patrons, not spenders. Don't merge the currencies; don't merely sever them. Two categorically different objects that can't inflate each other because they are not the same type:

MOVE 5 — De-risk with a "currency-less pilot" that tests the trust, not the token — and tests the hard thing first. Black's Wound 5: a learner coupon validates nothing Phase 2 needs. Two ways to answer it, and the second is stronger:

MOVE 6 — Cure the geography/standing wound: SSi is the invited guest, not the convener. Black's "incomers running the local money" is fatal if SSi convenes it — so SSi must not. The heartland layer exists only if an Ogwen-native body (Partneriaeth Ogwen-type trust, community land trust, Cwmpas-supported co-op) asks for it and owns the governance — sets rules, recruits merchants, holds a revocable contract over SSi as subordinate tech-and-legitimacy supplier. The framing flips from "company that teaches outsiders the language natives already speak" to "Ogwen's own institution, with a tool built by a Welsh company in service of Cymraeg 2050." Collective-ownership / less-to-none passed by governance structure, not slogan; reputational contagion (Wound 7) de-fused because a failure is the community's scheme pausing, not SSi losing grandmothers' money. Wall, named honestly: this requires SSi to genuinely cede control and accept being fireable — a real constraint, not cosmetic. If the founders won't, this move is unavailable and Wound 7 stands. (Confidence: high that the structure cures the wound; conditional on real cession.)

MOVE 7 — Bind the credit to Welsh usage, not Welsh learning — and ship the pattern, not the currency. Two reinforcements that make the exemplar the mission rather than a detour:

The whole-question reframe. The proposal asks "can SSi issue a credit." Drop that question. SSi's rare, real asset is trusted bilingual surface + Cymraeg-2050 legitimacy + warm distribution + the ability to verify Welsh-usage outcomes no funder can otherwise reach — and that asset is most valuable precisely where it is not the money. The "currency" was always a proxy for a trustworthy way to pay people for unpriced Δω; build the trustworthy-way-to-pay and let a regulated partner or plain sterling grant-payments be the bank. Reframe the most fundable product as structured grant-delivery, not new money: a disbursement-and-verification rail where funders pay outcomes (transmission-hours, care-hours) in ordinary sterling through a regulated processor, and SSi supplies the trusted ledger that proves the work happened in a language and community the funder can't otherwise reach. Lower regulatory temperature, no e-money, and SSi's genuine rare asset lands in a paid role.

Walls that do not yield in-horizon (named honestly):

Green in one line: stop trying to be the money — be the trusted bilingual ledger-and-legitimacy skin on someone else's regulated rail, owned by an Ogwen body, bound to measured Welsh usage, tested hard-thing-first at toy scale, with the carer kept a sterling PAYE employee and the diaspora a one-way patron, then shipped as an open module — which converts every mortal Black wound into either a dissolved contradiction or a cheap, early, reversible failure, and leaves the one impossible object (the consumer cash-out token) honestly on the floor.

RED HAT — for Tom. Gut factors only, no verdict.

(a) Mission energy vs dilution — gut-read. There's a real warmth in the core sentence: diaspora learning Welsh pays for care in the heartland. That's not a finance scheme, it's the mission made circulatory — and it'll feel like belonging to the people who already love SSi. But sit with the other gut signal: SSi's community fell in love with speaking Welsh from day one, the recall, the bravery, the voice in their ear. A ledger has no voice. The felt risk isn't that it fails — it's that it's cold in a way the product never was, and the community can sense when something stops being about the language and starts being about plumbing.

(b) Political/identity momentum. Tailwind: Cymraeg 2050 is visibly behind, the heartland geography is right, and Welsh sovereignty has genuine emotional voltage right now — a Welsh company building Welsh abundance reads beautifully and could enrol people the way an argument never will. Headwind, felt: a beloved language brand is a fragile, almost sacred thing, and money is the fastest way to make something sacred feel grubby. The asymmetry your own hats kept circling lands in the gut here — "fintech pilot fails" is a Tuesday; "the Welsh-language people got grandmothers taxed / lost the chapel fund" is a wound to the language itself. The thing carrying the upside is the same thing that makes the downside feel like betrayal rather than disappointment.

(c) Founder/org appetite & bandwidth. Honest gut: Aran's instinct is a teacher's instinct, audio-first, recall, the human voice — does running a ledger, FCA counsel, merchant onboarding and benefit edge-cases actually light him up, or is it the kind of work that gets endured? This has the shape of a second company bolted onto the first, sharing a logo and a finite founder. And you — Head of Course Creation, evaluating a proposal you're inside — there's a felt pull worth naming out loud: the idea is genuinely exciting to you, and excitement-from-the-inside is exactly the energy that can't tell vision from gravity well.

(d) The felt sense — wind or shine. Mixed, and worth being honest that it's mixed. There's true wind behind the narration — the one-line story is real and moving, and the exemplar/replicable-module idea (Cornish, Basque, Māori) has the good kind of ambition. But there's also unmistakable shiny-distraction energy in the "SSi issues a currency" framing specifically — the part that feels biggest is the part that feels most like a magpie reach. The tell: the versions that feel calm in the body are the ones where SSi stays the trusted voice and skin and someone else holds the money. The versions that feel thrilling are the ones the rest of the dossier says are dangerous. When thrill and safety point opposite ways, the gut's job is just to flag that — and it is flagging it.

One quiet undertow to honour: nothing here yet says Ogwen asked. The warmest version of this starts with a conversation in the valley, not a build — and the gut knows the difference between bringing a gift and arriving with a system.