I have a habit that friends find strange. When a governance document lands in front of me โ a DAO charter, a foundation proposal, a federal press release โ I read the names before I read the mission. Not the vision statements. Not the bolded promises about protecting the interests of all Americans. The names. A governance structure is only as honest as the people it claims will run it, and when the names don't line up with reality, everything downstream deserves a second look.
That habit earned its keep last week. A wire summary crossed my desk describing a 'Superintelligence Task Force' โ a White House coordination body supposedly assembled to keep the United States at the front of the race toward artificial general intelligence and beyond. The roster looked impressive on first read: the Director of National Intelligence, the chair of the Federal Trade Commission, the Pentagon's research chief, the head of the Office of Personnel Management, the White House chief of staff. Then I started checking, the way I check the signatories on a treasury multisig before I sign anything.
The Director of National Intelligence, the document said, was Jay Clayton. It isn't. Jay Clayton spent his public career at the Securities and Exchange Commission; the office in question belongs to Tulsi Gabbard. The head of the Office of Personnel Management, it claimed, was Scott Cooper. In reality that seat is held by Scott Kupor. The abbreviation attached to the body โ 'SIF' โ doesn't even match its own name, which should render as STF. Some of the names were plausible. Two were wrong. And the central event โ a 'historic White House superintelligence accord' โ left no trace I could find in any primary source.
A governance document that cannot get its own roster right is telling you something. The question is whether it is telling you about a translation error, a fabrication, or a future that simply hasn't happened yet.
I don't write about this because I care about American personnel appointments. I write about it because the same failure mode โ a structure that looks authoritative while resting on unverified claims โ is the defining pathology of the systems I have spent a decade helping to build. If the people designing oversight for the most consequential technology in human history are making the same mistakes we made in our first DAOs, then the decentralized world has something urgent to say, and something uncomfortable to learn.
Let me set the table, because the significance here is not in the leak. It is in the drift.
For most of the past decade, 'superintelligence' lived in research papers and late-night arguments among people who read too much philosophy. The policy world preferred safer vocabulary โ frontier models, advanced AI, foundation models. Those words were chosen deliberately. 'Frontier' sounds like a place you can map. 'Foundation' sounds like something you can build on and regulate. 'Superintelligence' sounds like a thing that has already escaped the map.
The shift to the stronger word is itself a policy instrument. It is the difference between saying we should study this technology and saying we should prepare for a power that may exceed us. That reframing does real work: it mobilizes budgets, it commands congressional attention, and it stakes a claim in the international competition over who gets to define the governance framework for the next generation of AI. The United States has been circling this territory for years โ through executive orders, through export controls, through the creation of a dedicated AI Safety Institute. A federal coordination body for superintelligence would be the logical next stone in that path.
Here is where the blockchain world should sit up. The body, as described, would not fund research or build models. It would coordinate. Its members span intelligence, antitrust and consumer protection, defense research, and federal personnel management. Its stakeholders, per the summary, would include the leading technology companies, the operators of critical infrastructure, and โ this is the detail that caught my attention โ religious organizations. Its stated purpose: to protect the interests of all Americans and improve their lives.
If that architecture sounds familiar, it should. It is the exact shape of a DAO foundation: a coordinating entity with no direct product, whose power lies in convening, in setting the terms of the conversation, and in gradually accumulating the authority to decide who is allowed at the table. I have watched this pattern play out across dozens of protocols. The foundation starts as a facilitator. It ends as the arbiter. The transition is rarely announced, and by the time anyone notices, the constitution has already been rewritten.
There is a second, larger current running underneath. For three years, the dominant story of AI competition has been a race of capabilities โ who trains the biggest model, who secures the most compute, who publishes the benchmark first. A federal superintelligence body, even a toothless one, signals a change in the axis of competition. It moves the contest from who builds the strongest system to who writes the rules for the strongest system. That is a governance contest, not a capability contest, and it is one the decentralized community has been fighting, in miniature, since the first token-weighted vote was gamed by a whale.
Let me work through what the document actually reveals, layer by layer, the way I would audit a protocol before recommending anyone put capital into it.
The composition tells you the intent, and the intent is not research. It is control. The four departments named โ intelligence, antitrust and consumer protection, defense research, and federal personnel โ map onto four distinct anxieties. Intelligence means the state treats superintelligence as a national security asset, with all the secrecy, classification, and export sensitivity that implies. Antitrust and consumer protection means the state is worried about concentration โ about a handful of companies holding capabilities that function like a public utility. Defense research means the state intends to be a user, not merely a regulator. And personnel management โ the strangest inclusion, the one a casual reader would skip past โ means the state intends to build its own internal AI capacity, to become a buyer and an operator rather than a bystander.
Most coordination bodies in technology policy include the science agencies and leave it there. The presence of the personnel office is a tell. It says the government wants to staff itself with people who understand these systems, to run superintelligent tools inside its own machinery, and to negotiate with the private labs as a peer rather than a petitioner. When I co-designed the governance structure for a five-million-dollar treasury back in 2020, the single most important decision was not the voting math. It was who held the operational keys. The same logic applies here. The roster is the keys.
Even the institutional geography is unresolved. The United States already has an AI Safety Institute nested inside its standards agency, a science advisory apparatus attached to the presidency, and a national security council with its own technology staff. A new coordination body would either overlap with these, absorb them, or quietly supersede them โ and the document says nothing about which. In protocol terms, this is a governance proposal that creates a new council without specifying its relationship to the existing multisig. Anyone who has watched a DAO fork itself into paralysis over exactly this ambiguity knows how it ends: not with a decision, but with a slow accretion of informal power by whoever happens to be in the room when the important conversations happen.
The stakeholder list reveals a theory of legitimacy โ and a blind spot. Critical infrastructure, public interest groups, consumers, religious organizations. This is a deliberate attempt to assemble social consent, not just technical oversight. The inclusion of religious bodies is the distinctive American move: it reaches for a moral vocabulary that transcends the engineering community, an attempt to ground AI alignment in shared human values rather than in the preferences of a few labs in California. I understand the impulse. I have spent years arguing that code without compassion is cold โ that a system optimizing for efficiency will, left unattended, optimize away the people it was meant to serve. Building a moral constituency around AI is not naive. It is necessary.
But a legitimacy that is declared is not a legitimacy that is earned. The document names its stakeholders in the abstract. It never names a single company. It never names a single standard. It never explains how a religious organization's input becomes an actual constraint on a model's behavior, or how a public interest group's concern becomes a line of enforceable code. In governance terms, this is a proposal with a mission statement and no mechanism โ the most dangerous kind of document, because it looks complete while remaining entirely aspirational.
And then there is the silence, which is the most damning feature of all. No budget. No statutory authority. No term of existence. No definition of what superintelligence actually is โ a term that, years after it entered the policy lexicon, still has no rigorous technical meaning. No mention of export controls, compute thresholds, or the international competition that is the actual subtext of the entire exercise. No mechanism for evaluating models, no recall power, no independent red-team process. A structure can look like a regulator and function like a rumor.
The document's most revealing feature, in fact, is what it refuses to say. Nowhere does it mention China, though the entire logic of a superintelligence leadership body is an answer to a perceived rival. Nowhere does it mention export controls or compute restrictions, the hard instruments of technological competition. The omission is not an accident of reporting; it is a posture. Naming a competitor out loud invites escalation and international backlash. Building a coordination body that keeps the competitor unnamed preserves optionality. It is the diplomatic equivalent of a governance proposal that quietly removes the clause everyone would fight over and hopes no one reads the diff.
I have a rule I apply to every governance proposal I review, and it has saved my community real money. If a structure has no budget and no enforcement power, it is not a regulator. It is a press release with a letterhead. The coordination-before-regulation pattern is real and worth tracking โ coordination bodies often mature into regulators once they acquire staff and funding. But until the budget line appears, the correct posture is patience, not panic.
Here is the angle I want to press, because it is the one the crypto community is least comfortable hearing.
The instinctive reaction in our corner of the world is to read a story like this as a story about the state โ about Washington reaching for control over a technology that should belong to everyone. That reaction is emotionally satisfying and analytically lazy. The real governance risk in the superintelligence era is not that a federal task force will seize the machines. It is that a small number of already-powerful actors will write the rules that everyone else must live under, and that the multi-stakeholder language will be the velvet glove on that hand.
I know this pattern intimately, because my industry invented it. On-chain governance turnout sits below five percent across most major protocols. The community that supposedly decides is, in practice, a handful of whales and venture funds whose token weight makes every other vote decorative. We call it decentralization. It behaves like an oligarchy with better branding. When I see a government body announce that it will convene leading technology companies to shape the future of superintelligence, I do not see a state against the market. I see a revolving door being installed, with the same small set of incumbents walking through it in both directions.
The uncomfortable truth is that centralized institutions and decentralized ones are running the same play, and neither has solved the problem of who actually holds the pen.
I saw a version of this play two years ago, when I helped lead a coalition of smaller DAOs to negotiate a grant from a large institutional investor, conditional on that investor adopting our transparency protocols. We won that negotiation โ the terms were ours, the standards were ours โ and I came away understanding something uncomfortable. We had not decentralized anything. We had simply become a stakeholder important enough to be included at the table, and inclusion, for all its rhetoric, is a form of co-option as often as it is a form of victory. The superintelligence task force, with its invitation to leading companies and religious bodies, is extending the same invitation. The question is whether the invitees understand what they are being handed, and what is being taken in return.
There is a second contrarian reading, and it is subtler. The document's errors โ the wrong intelligence director, the wrong personnel chief, the abbreviation that doesn't match its own name โ may not be the signature of a clumsy human translator. They may be the signature of a machine. I spent much of the past year building a manual verification layer for governance proposals precisely because AI-generated text has become indistinguishable from human-authored text at the exact moment when the stakes of getting it wrong have never been higher. A wire service that recycles an AI-synthesized policy summary, complete with plausible-sounding names in the wrong chairs, is not a hypothetical. It is the world we now live in.
Which means the deepest lesson of this episode is not about superintelligence at all. It is about verification. The same community that built trustless systems on the premise that you should never have to trust a counterparty is now being asked to trust a paragraph of unverified prose about the most powerful technology in history. The tools we developed to audit a smart contract โ reading the source, checking the signers, demanding proof โ are exactly the tools that a citizenry needs to audit its own governance. We have been practicing for this moment without realizing it.
So where does that leave us, standing in the sideways market of a policy cycle that has not yet found its direction?
I think the honest answer is that we are early, and that early is a gift. The superintelligence task force, if it exists, is a coordination body with no teeth and a roster it cannot keep straight. The real rules are still unwritten. The question of who governs the machines โ which companies, which governments, which communities, which values โ is genuinely open, in the way that a governance proposal is open in the first hours after it is posted, before the whales have noticed and the quorum has quietly filled.
That window will not stay open. It never does.
The lesson I take from a flawed press release is not cynicism. It is urgency. Code without compassion is cold, and governance without verification is theater. The decentralized community has spent a decade learning, often painfully, that the only defense against concentrated power is a community that pays attention โ that reads the names, checks the roster, and refuses to accept a vision statement in place of a mechanism. The superintelligence era will be governed by someone. The only question worth asking is whether the people who will live under those rules will have bothered to read them before they were signed.
I intend to keep reading. I hope you will too.