On March 2 — the first trading day after the first American and Israeli strikes on Iran — an investment account bought shares in eight oil and gas companies. One of them was ExxonMobil, in a position valued somewhere between $100,000 and $250,000. Five weeks later, on April 7, the same account sold between $500,000 and $1 million of that same stock. Roughly two and a half hours after the sale, a ceasefire with Iran was announced. By the next morning, the shares had opened down more than six percent.
Read that sequence twice and notice what is absent. Not the dates. Not the tickers. What the record withholds is the minute, the share count, the fill price, the batch. The document that produced those figures — an executive branch Public Financial Disclosure Report, OGE Form 278e — reports ranges. It is a ledger that cannot tell time.
Twenty years of reading ledgers has taught me that a record which cannot tell time cannot tell a story. It can only insinuate one. Between February 27 and August 31, nine oil, gas, refining, and pipeline holdings in that account appreciated an estimated $1.5 million to $4.4 million. As of June 29, the account had reported at least 23 transactions involving related stock sales. On March 23 — a session in which strikes on Iranian energy facilities were delayed before the opening bell and Brent crude fell nearly 11 percent — the account reported 16 purchases of oil and gas stocks, totaling roughly $163,000 to $570,000.
None of those figures represent realized profits, and an appreciation estimate computed from bracket midpoints is not a number so much as a shape. The forms disclose no exact share quantities, no execution prices, no sale batches. A disclosure regime that reports ranges can describe a pattern; it cannot adjudicate one. The story also lands in a market that has spent months chopping sideways — capital repositioning rather than committing, every participant hunting for precisely the kind of information edge a range-shaped record leaves unresolved.
So let us be precise about what the record actually is, because the mechanics matter far more than the headline.
Form 278e is an artifact of the Ethics in Government Act of 1978, the post-Watergate settlement built on the premise that officials should file their holdings and update them on a fixed calendar. The brackets are the mechanism: $100,001 to $250,000; $500,001 to $1,000,000; $1,000,001 to $5,000,000. The bracket balances privacy against scrutiny, and it achieves that balance by converting precise facts into probability clouds. Congress tightened its own transaction reporting to 45 days under the STOCK Act of 2012. Executive branch transaction reporting still arrives essentially once a year — which is why these numbers surfaced on September 9 rather than in March.
Two official statements accompany the disclosure, and both should be taken at face value. CNBC found no evidence that the president directed the trades, had prior knowledge of the decisions in question, or that personal interests shaped policy. The White House said the portfolio is managed entirely by independent managers. Both can be true alongside everything above, and that is not a contradiction. It is the design. “Independent manager” is a claim about authority, not about information. A manager may trade without instruction and still trade inside a window that only a principal's calendar could open.
The ambiguity, in other words, is structural. Intent is not missing from the dataset because someone removed it. It is missing because the dataset was never built to hold it. Anyone who works in my field will recognize the shape of this problem instantly, because crypto has spent fifteen years insisting that a ledger can answer exactly this class of question. That claim deserves testing on a case where the stakes are political rather than merely financial.
The closest crypto-native analogue to the events of March and April is not a token. It is a market. Geopolitical contracts — ceasefire, strike, oil price thresholds — have become one of the highest-volume categories on prediction platforms. Regulated venues such as Kalshi operate as designated contract markets under the CFTC. Offshore venues such as Polymarket have spent two years negotiating their way back toward American users. The instruments are binary, the resolution criteria are written in English prose, and the money is real.
A prediction market is, technically, a disclosure machine. It converts what participants believe about a future event into a continuous public price. That is the most transparent instrument finance has ever produced. It also works flawlessly right up to the moment it has to decide what happened.
The resolution layer is where the architecture shows its seams. Polymarket settles through an optimistic oracle: a proposer asserts an outcome, and if the assertion is disputed, a token-holder vote decides. When I sampled resolution disputes across several of those votes during the 2024 cycle, the pattern matched everything I had already measured in DAO governance. Turnout in the low single digits as a share of outstanding voting supply. A small set of addresses carrying the result. On-chain voter turnout is perpetually below five percent, and the ballots that do get cast tend to come from the wallets with the most at stake.
I want to be careful here, because this is where critics and evangelists both overreach. Blockchain solved ordering and immutability. It did not solve meaning. A signed transaction proves that a trade occurred at block N, at a time the network agreed upon. It proves nothing about why the signer acted, or what the signer knew at block N minus one. Precision is not verification. It is only the precondition for it.
Now consider the honest version of the counterfactual. Suppose those nine energy holdings had been issued as tokenized equities on a permissioned rail, with transfer restrictions and identity binding enforced at the contract level — the ERC-3643 pattern institutional issuers have been quietly standardizing on for years. Suppose the custodian published a signed attestation of balances daily and the brokerage published execution timestamps to the second. In that world, March 23 would be a fact rather than a bracket. April 7 would be a fact rather than a story about two and a half hours.
In that world, a great deal would also go wrong. This is the objection that dismantles most naive transparency proposals, and it deserves stating plainly: real-time publication of a large, identifiable portfolio is a front-running engine. Blind trusts exist for a reason deeper than embarrassment management. The market will trade ahead of the principal, and the principal's decisions will be priced before they are made. The design problem is therefore not publish or do not publish. It is how to publish verification without publishing position.
Which is precisely the problem zero-knowledge proofs were invented to address, and precisely the problem I have worked on since 2026 with a small team designing a decentralized identity framework for AI agents on Polkadot. The construction is unglamorous and it is the only one that scales. You never publish the transaction. You publish a computable predicate about it: no transaction in sector X within seventy-two hours preceding any public statement touching sector X. The statement carries a signed timestamp. The trade carries a signed timestamp. A circuit proves the ordering without revealing size, direction, or counterparty. Truth emerges when the ledger is transparent — but only about the things the ledger was told to record, and only in the language it was handed.
That sentence is also the hard limit of the approach, and I say so as someone who builds these systems rather than sells them. A zero-knowledge proof establishes a fact about data you already possess. It cannot attest to the absence of data you chose not to include. It can prove that you did not trade in X. It cannot prove that you did not know something about X, because knowledge is not a field in a schema. We are building machines that certify compliance with rules we wrote in advance, and then we are calling the output ethics.
Europe has now run a version of this experiment at scale. MiCA, the Markets in Crypto-Assets regulation, gave the continent something the United States still lacks: an actual rulebook for issuers, service providers, and stablecoin reserves. The clarity is real. So is the invoice. Reserve requirements push asset-referenced and e-money token issuers into periodic attestation performed by auditors who do not work cheaply. Licensing obligations for crypto-asset service providers bring fit-and-proper assessments, own-funds thresholds, complaints handling, and a supervisory relationship that consumes legal budget before it consumes any engineering budget at all. The resulting economics are not neutral. Compliance is a fixed cost, and fixed costs select for scale. Small issuers do not fail because they are dishonest. They fail because they cannot afford to prove they are honest — while the largest participants convert the identical obligation into a moat.
The same asymmetry runs through every disclosure regime I have ever read, including the one filed on September 9. The regime does not require concealment. It simply makes precision expensive and ambiguity free.
And here is the paradox I keep circling. The brokerage that executed the April 7 sale knows the fill price to the cent and the time to the millisecond. That record exists right now. It is simply not the record we receive, and it is not the record a regulator receives without a subpoena. So the architecture is not a world in which the truth cannot be known. It is a world in which knowing is gated — where the exact record is held by the party with the least interest in publishing it, and the vague record is handed to the public.
From which the central insight follows. Transparency is a data property. Accountability is an incentive property. A system can be made perfectly legible and still produce nothing at all when the legible thing is wrong.
The reflexive answer from my own industry — put it on chain and the problem dissolves — is wrong, and I would rather say so myself than wait for a stranger to. Openness, in my own writing, is not a feature; it is a philosophy. Philosophies do not enforce themselves.
The proof is first-hand, and it is the experience that moved me from engineering into ethics. In 2017, while everyone else was writing tokenomics threads, I spent six months auditing the early governance contracts of MakerDAO. I found a logic flaw in the stability fee calculation that threatened user solvency. Every relevant line of that contract was public. Anyone could have read it — which is exactly how I found it. I reported the flaw anonymously on GitHub, and a maintainer with authority shipped a fix. The fix required no new transparency, because transparency had already been total. It required jurisdiction and consequence: a person who could patch the code, and who cared that the patch held.
Then came 2022, and I withdrew for three months to recover from what the collapse had cost me. In that silence I audited fifty failed protocol post-mortems. Not one of them failed because the data was hidden. They failed because the people who could see the risk carried none of it. I wrote a manifesto afterward arguing that decentralization without accountability is anarchy. I have written elsewhere that in the chaos of DeFi, I found my silence. Silence is not a consequence. No ledger, however complete, supplies one.
So what does transparency actually produce? It changes the audience, not the actor. It converts private conduct into public argument. That is not nothing — it is how the numbers in this essay reached me at all — but it is a long way from the guarantee people believe they are purchasing. Humanity remains the only non-fungible asset: the one component in any of these systems capable of looking directly at the complete record and deciding what to do about it.
Two things are worth watching from here. The CFTC's posture on insider trading in prediction markets will set the precedent for what counts as trading on privileged knowledge when the underlying event is political rather than corporate, and the agency has been more serious about enforcement than most of my colleagues expected. Quieter, and arguably more consequential: whether custody providers begin selling cryptographic attestation to institutional and political clients, not because a regulator demands it, but because ambiguity is becoming a liability that carries a price.
And a third, which is the one I actually care about. We are about to hand this evidentiary problem to machines. The AI-agent identity work I contributed to this year exists because autonomous systems will soon execute transactions at a cadence no disclosure form can describe, leaving records timestamped to the millisecond that still tell us nothing about intent. If we cannot close that gap for a human portfolio now, we will not close it for a machine portfolio later.
The form filed on September 9 could not tell us the minute. It was never built to. Which leaves the only question worth asking: are we going to build systems that can — or systems that make the ambiguity cheaper to keep?