Integrity digest, September 2026

Projects publish numbers. We check them against the chain, an independent index or the project’s own deployed contract, and record the verdict with the date and the method. This is the first monthly digest of that ledger: what was checked in the month to 19 September 2026, what held, and what did not.

The tally above is computed from the board at build time, so it will read the same on this page, on the board and in the JSON feed. The rest of this post is the part a tally can’t carry: which claims fell, and why it matters.

The corrections that changed a grade

ORA’s oracle stopped serving a year ago. ORA’s token page claimed an Onchain AI Oracle that had processed 180,000+ inference calls on 500+ nodes. Read directly off the OAO proxy on Ethereum, the lifetime request counter reached 4,615 on 26 September 2025 and hasn’t moved since. The last five requests were never answered, and every request from April 2025 onward came from ORA’s own example contract, so the closing traffic was the project calling itself.

Our own review had carried the 180,000 figure as evidence; both ORA scores moved, and Revenue Sustainability now sits at the floor of its band. The docs domain that hosted the claim has since been deleted, the successor product went dark on 18 September, and no governance venue for the token has ever existed.

Bittensor’s governance runs through one key. The charter says “decentralised, community-governed”. The finney runtime, read at block 9,088,526 on 18 September, carries 28 pallets and none of them is a council, senate or democracy pallet. Privileged operations, including runtime upgrades, dispatch through a single Sudo.Key. Bittensor’s own documentation attributes that key to a Rao Foundation multisig, which we record as their claim.

The board had described a Triumvirate-and-Senate structure since August; that was the documented design and it isn’t what the chain runs. The correction cuts against the project, because one key concentrates more than a chamber with a validatorValidatorA computer that runs the full blockchain protocol, verifies transactions, and proposes new blocks. Validators are the workers that keep a Proof of Stake network running, and they earn rewards for doing the work correctly.Like a notary public who witnesses and stamps legal documents. Validators witness transactions, check they follow the rules, and stamp them into the permanent record. A notary who commits fraud loses their license. Validators work the same way, except the license is staked tokens that get slashed on misbehaviour.Read more → veto.

NEAR’s treasury takes nothing. The Nomicon economics spec, which NEAR still publishes, sends 10% of issuance to the protocol treasury. The live protocol config returns protocol_reward_rate = [0, 1] at protocol version 86: validators receive the whole of issuance and treasury.near is configured, named and set to zero.

We had restated the 10% slice on our own page. A specification is primary about intent and says nothing about behaviour, and this is the cleanest example of the gap on the board.

Aethir’s insiders hold a quarter. Aethir positions itself as a transparent, decentralised network with broad distribution. Its own docs put team, investors and advisors at 29% between them, though the split is published only as an image and the current vestingVestingA schedule that locks up tokens allocated to insiders, investors, and team members, releasing them gradually over months or years. Vesting prevents insiders from dumping on public buyers immediately after launch.Like a new employee's stock options at a startup. You don't get all the shares on day one. They unlock over four years so you stick around and do the work rather than cashing out and leaving.Read more → table carries no advisors row, so the figure we can stand behind from what Aethir publishes today is the 24% that team and investors hold. On the code side, seven public repositories exist across two orgs and none carries a licence. A verified token contract isn’t network transparency.

Morpheus lost its Snapshot venue and its growth month. Two rows moved. The Snapshot space morpheusai.eth, which the board had confirmed as the token-holder voting venue, no longer exists; checked four ways with a positive control in each.

Our index of SessionClosed receipts on Base shows monthly tokens processed rising every month from February to July, then falling 38% in August, with sessions falling further than tokens. The demand is real; the growth half of the thesis did not hold for the last complete month.

Corrections we owed the reader

Thirty-three of the forty-seven corrections overturn a sentence in our own earlier evidence rather than a claim a project published. Most of them landed in one sweep on 6 September, when every exchange-listing claim on the board was re-read against CoinGecko’s paginated ticker set. Twenty-two rows moved. Three examples:

• Flock. Our evidence said “no Binance or Coinbase”. Coinbase lists FLOCK.
• Nosana. Our evidence said “no tier-1 CEX”. The thin-market reading held; the absence claim did not.
• Walrus. Our evidence said “Gate.io and other major venues; modest volume”. The market is broader than we described, on 68 tickers across 47 venues.

Three other self-corrections carry more weight than a listing:

• Heurist. Our prose reasoned from the allocation table to “50% under foundation discretion”. The chain is starker: one Safe holds 66% of supply, the token has 82 holders in total, and the contract answers owner() with a single externally owned key where our evidence had said there were no controls to document. Graded understated, because the direction was right and the magnitude wasn’t.
• IoTeX. We had rated Snapshot governance as functional and accelerating and cited IIP-42 as passed. The space’s last vote was IIP-40, in October 2024, and IIP-42 is a Draft in the repository. Authoring and ratifying are different acts.
• Nillion. Our sentence said the MPC architecture is specified to stop any single party censoring computation. The token is a UUPS proxyUUPS ProxyA pattern for upgradeable smart contracts where the upgrade logic sits in the implementation contract itself. Cheaper gas than the older Transparent proxy, but the implementation must be written carefully or the contract can be bricked.A power tool where the swappable head also contains the swap button. Pull the head off, snap on a new one with a new button, you can keep using the tool. But if the new head forgets to ship a swap button, the tool is now stuck on whatever head you just attached.Read more → whose logic a 3-of-5 Safe can replace with no timelock. The design and the deployed contract are different systems, and only one of them had been measured.

Each of those three is flagged for the October score review rather than moved now. A correction to our own words earns a note in the row and a date, the same as a correction to a project’s.

What the month says about the method

Fourteen of the corrections overturn something a project published; thirty-three overturn something we published. Read that as the board working. A ledger that only ever finds fault outward is a ledger nobody re-tests, and the rows that moved this month moved because the checks ran again with a positive control beside them.

The pattern behind the biggest corrections is the same each time: a design document, a spec or a charter describes an intended mechanism, and the deployed contract or runtime does something else.

• NEAR’s spec still says 10%. The chain says zero.
• Bittensor’s docs still describe chambers. The runtime has a sudo key.
• ORA’s page still describes a live oracle. The contract has been silent for a year.

Where doc and chain disagree, the gap is the story, and the board records it as one.

The next digest covers the month to 19 October 2026. The board itself updates as rows are re-tested; the digest is the dated reading.