Verification
Verifiable, portable, machine-readable credentials.
-
Credential discovery in the agentic web: a credential not legible to agents is invisible
As agents mediate more credential discovery, a credential an agent cannot resolve is structurally invisible — not a UX problem, a non-existence problem. Agent-legible = resolvable issuer identity + signed VC + live revocation + evidence pointers + machine-readable semantics. A protocol bet to make at issuance, because you can't retrofit legibility onto a credential that was never legible.
-
Verifiable credentials as a protocol: Open Badges 3.0 is sound, the adoption graph is broken
Open Badges 3.0 and W3C Verifiable Credentials are a sound format — signed, evidence-linked, portable, revocable. The failure is last-mile legibility: receivers don't read the metadata and have no incentive to. The fix is not another spec revision — it's building the receiver side.
-
Why we didn't use a blockchain: sealed-key custody and the issuer-identity problem
The credentialing space defaults to 'blockchain!' as the trust answer. The honest negative result, from having built a real issuer: the problem a chain solves isn't the problem a credential has. The hard part is issuer identity and key custody, and no consensus protocol makes that decision for you.
-
The proctoring bias tax — and why it falls hardest on neurodivergent learners
Automated proctoring doesn't just fail at integrity — it manufactures false-positive flags against the exact learners credentialing claims to serve, and a credential earned under a biased proctor is a lower-trust credential.
-
The credentialing trust stack
Why most micro-credentials die at the employer-trust checkpoint — and the five layers a credential must stack to actually be trusted: identity, assessment integrity, evidence, issuance, and portability.