A recent arXiv paper on "Binding Agent ID" surveys the growing field of agent accountability frameworks and lines up two of the most-cited proposals against each other: Carnegie Mellon's LOKA protocol and Ethereum's ERC-8004. The comparison is useful because the two approaches solve overlapping problems with different machinery. LOKA is built around a Universal Agent Identity Layer using DIDs and Verifiable Credentials, plus a blockchain-based trust layer meant to govern how agents interact and to back ethical consistency claims with cryptographic proof. The paper describes it as a decentralized identity framework specifically targeting accountability and ethical consistency, implementing a blockchain-based trust layer to govern agent interactions. ERC-8004 takes a narrower, more Ethereum-native route: rather than issuing portable identity credentials, it defines "Trustless Agents" through an on-chain validation registry, so agents submit verifiable proofs of completed work directly to a smart contract that anyone can query. The paper frames it this way: ERC-8004 introduces Trustless Agents via an on-chain validation registry, allowing agents to submit verifiable proofs of their work. The practical difference matters more than it might look. LOKA's bet is that identity needs to travel with the agent across contexts — a DID that carries credentials attesting to capability and ethical standing wherever the agent goes, independent of any single blockchain or platform. ERC-8004's bet is narrower and arguably more deployable today: it doesn't try to solve portable identity or ethical alignment, it just makes task completion auditable on a registry other parties can check without trusting the agent's own say-so. Neither framework has seen production adoption at scale — LOKA remains a research proposal more than a year old, and ERC-8004 is still an Ethereum draft standard working through its own review process. What the new paper adds isn't a verdict on which wins, but a clearer map of the tradeoff space: full-stack identity-plus-ethics frameworks versus minimal, auditable-proof registries. For teams building agent infrastructure now, that's the more useful takeaway than either paper's individual pitch — the field hasn't converged on whether accountability needs a whole identity layer or just a place to check receipts. The paper's broader point is that the accountability gap in agent ecosystems is still being attacked from multiple directions at once, with no single architecture yet treated as the default.