Decentralized Identity for Student Credentials: Simulation-Based Comparison of Revocation in Verifiable Credential, AnonCreds, and Polygon ID Stacks

Academic credentials are issued once, verified many times, and revoked rarely, yet the revocation mechanism decides most of the cost, latency, and privacy of a decentralized-identity deployment for students. This paper compares three deployable stacks: a W3C Verifiable Credential with did:web issuer keys, SD-JWT selective disclosure, and a Bitstring Status List; an AnonCreds-style accumulator registry on a permissioned ledger; and an iden3 (Polygon ID) sparse-Merkle identity state anchored on a public chain. A lifecycle model is derived with four propositions: an entropy bound on status-list size, the holder-side update burden of accumulators, a closed form for anchoring transactions under batching, and a linear law for false acceptance of revoked credentials under caching. Cryptographic primitives are measured in a reference implementation, and the three stacks are replayed against one simulated population of two million credentials, 17.6 thousand revocations, and 3.7 million verification requests over five years. The status-list stack costs 2.9 kilobytes and 0.2 milliseconds per verification, but every presentation is linkable and its herd privacy collapses for small issuers: one third of list fetches expose anonymity sets below 1000, far from the 131 072-entry list length the specification assumes. Federating lists across issuers restores a median anonymity set of 130 thousand for 2.7 times more list bytes per fetch. The accumulator stack is unlinkable but shifts up to 120 witness updates per presentation onto the holder, and the Merkle-state stack is unlinkable at an anchoring cost governed by whether issuance is on-chain rather than by batching. All code and result logs are released.