That is the decision this desk will keep repeating. Automation does not erase the manager. Code can run a strategy. A person can still choose the strategy, change the parameters, collect the fee, and sell the expected return. The Commission's larger March 2026 interpretation, which is Commission-level rather than a single official's view, already treats protocol staking and investment contracts as fact-specific. Together, the materials move the question off the marketing claim of decentralization and onto control and expectation. The failure model is not abstract. Depositors can lose through code faults, oracle failures, liquidation, borrower default, slashing, governance capture, key compromise, or a manager's strategy change. Withdrawal queues and unstaking periods can make an apparently liquid product slow to exit. Past returns can depend on incentives that disappear. Managers and promoters bear disclosure and compliance risk when they select strategies or sell an expected return. Users bear the loss when those facts were compressed into a single APY. A proper profile starts with the control surface, not the yield. Identify control keys, upgrade authority, allocation rules, fee flows, withdrawal terms, audits, dependencies, and loss allocation. Distinguish protocol (what the deployed code can do without a person), operator (who can change parameters, pause, or reroute), user (what a depositor was told and what claim remains after loss), and legal wrapper (fund, note, investment company, adviser relationship, or none of those). Immutable code and adjustable rules and personal discretion are three different machines. They should not share a headline. Onchain lending sits on the same spectrum. Peirce described strategies that let participants deposit assets into systems that lend them for a fee, and flagged the people who set interest rates, eligible assets, loan-to-value limits, and liquidation thresholds. Those choices can look like credit administration. They can also look like the kind of managerial effort that securities analysis has always cared about. The statement does not classify every vault. Courts and examiners have not either. Code can change faster than a legal memo. That residual uncertainty is a closing beat, not a reason to treat the label as empty. The regulated example is useful because it does not pretend otherwise. Grayscale's Ethereum Staking ETF prospectus supplement, filed on EDGAR, is a Commission-facing disclosure of staking custody, validator selection, liquidity constraints, and trust-level risk. It is not a vault. It is a reminder that when staking is wrapped in a registered product, the control and liquidity facts have to be written down. Unregistered products that route user assets into staking or lending do not become safer by omitting the same page. Disclosure standards for strategy drift, composability, and emergency control remain uneven. Annualized yields are not analysis. The standing story is the control and failure model: what can change, who can change it, what dependencies can break, and what claim a user has after loss. Updates belong to audited code, governance votes, prospectus changes, incidents, and binding agency actions. A commissioner's spectrum is a map. It is not a hall pass, and it is not a ban.
