Bangladesh's healthcare failures are identity failures

The centralized alternative has already failed publicly. In 2023 a government website leaked the personal data of roughly 50 million citizens, and national ID information subsequently circulated on Telegram. That is not an argument against the state. It is the documented reason a national health record cannot be one more honeypot. The architecture must assume breach and make breach unrewarding: records encrypted to patient-held keys, no central trove whose theft exposes a population's diagnoses.

The law now says what the architecture already assumed. The Personal Data Protection Ordinance 2025 recognizes every citizen as the owner of their personal data, mandates explicit consent, regulates health data with special strictness, and takes key effect around May 2027. Glyph's design is that ordinance rendered in cryptography rather than policy, arriving compliant before the enforcement date.

Every patient, physician, pharmacy, diagnostic center, hospital, factory, and NGO holds a decentralized identifier resolvable over plain HTTPS. No blockchain, no token. Ed25519 signatures, W3C standards throughout. Every clinical fact is a credential signed by its issuer and held in its subject's wallet.

Enrollment meets every Bangladeshi where their documents are, through eight paths: national ID, birth certificate, passport, BMET registration for migrants, embassy attestation, institutional vouching, NGO-mediated enrollment for refugees and stateless people, and an anonymous mode for stigmatized care with no civil anchor at all. Every credential carries its anchor provenance, so a verifier always knows what was actually checked, and no one is excluded from care.

Professional identity strengthens in declared phases: self-issued with provenance first, institution-vouched second, authority-anchored third as BMDC and DGHS join as issuers. No phase pretends to be a later one. The credential says what it is.

Bangladesh never had family physicians, so the system cannot reconnect patients to their doctors. It must create relationships that never existed. On top of identity sits matching: clinical need, geography, language and dialect, gender preference, and price tier, with the strongest signal being continuity itself, the doctor you have seen before. Success is measured in repeat visits to the matched physician, not consultations brokered. Ranking is transparent and paid placement is structurally forbidden.

Bangladesh's health ministry already operates a national Grievance Redress System, and a 2025 PLOS Digital Health analysis of 11,604 messages submitted to it tells the familiar story: 67% forwarded to another department, 30% closed, 2.55% resolved. The channel exists. The accountability does not.

In this network a grievance is a credential. It is filed by an identified patient, or through anonymous mode for the complaints people are afraid to sign. It is bound to the encounter it concerns. It is tracked to a resolution that is attested rather than self-reported, and it cannot be quietly forwarded into oblivion. Aggregate resolution rates are visible to funders and regulators through selective disclosure, with no complainant exposed. For donor-funded programs a functioning grievance mechanism is a standing requirement; here it is a property of the record architecture itself.

An earlier draft of this vision overclaimed and was corrected, and the correction is now doctrine. Glyph is a single-operator system built on portable open standards, not a decentralized network. The honest claims are exactly three. Portability: every wallet exports in W3C-standard form, and credentials verify with or without KhaM Health's cooperation. No lock-in: a future operator, including the government, can assume the namespace, and signed credentials remain valid. Succession: KhaM Labs' governing documents oblige transfer of keys and infrastructure to a designated successor if the operator fails, in preference order a Bangladeshi public authority, a consortium of participating institutions, or an international digital-public-goods custodian.

The cryptographic core runs in production today: physician identities, signed visit notes and prescriptions, and pharmacy-counter verification with revocation. The eight enrollment paths, wallet custody modes, selective disclosure, recovery, and break-glass protocols are specified and under build, adapted from identity infrastructure already serving supply-chain credentials in production. This layer is the single critical-path dependency for Pharmacy, Lens results, Hospital, and Bridge.