LIVE ADA-- MCAP-- TVL-- STAKE-- EPOCH-- Cardano & Midnight 総合情報ポータル
SIPO
𝕏 🔍
HOME › Weekly Brief EN › Weekly Brief #24 | Ikigai Cleared the Committee and Expired 1.81 Points Short
Weekly Brief EN

Weekly Brief #24 | Ikigai Cleared the Committee and Expired 1.81 Points Short

2026-09-19SIPO

At 06:44:51 JST on September 17, at the epoch 656 boundary, two pending governance actions expired. DRep support for Reimburse Ikigai Info Governance Action Deposit. stood at 65.19%, 1.81 points short of the 67% threshold.

The Constitutional Committee had passed it. Five of seven members voted yes (71.43%), satisfying the two-thirds quorum. The only column that came up short was the DRep one.

The previous issue (#23) closed on these two like this: “If nothing moves before the deadline, they expire even with DRep support in place.” Something did move. It only came from the committee’s side.

And one more action has appeared, the mirror image of last week’s core story: a poll asking SPOs whether stakePoolTargetNum (k) should be raised from 500 to 1000. Last week the story was that the standalone minPoolCost action has no SPO voting column at all. This week, for a different economic parameter, an SPO column has been placed there — in a non-binding form.

Executive Summary

  • Two actions expired at the epoch 656 boundary: 06:44:51 JST on September 17. Reimburse Ikigai Info Governance Action Deposit. (requesting 103,000 ada) and Governance Incentives Framework 2026 (requesting 4,207,967 ada).
  • Ikigai fell 1.81 points short at 65.19% DRep support: 129 votes holding 2,890,889,668 ada in favour. Recorded opposition stopped at 5 votes and 54,984,919 ada.
  • The committee had passed Ikigai: 5 yes, 1 no, a 71.43% approval rate. The only column that failed was the DRep one, and this is expiry rather than rejection. This is Ikigai’s third expiry — submissions in epochs 590, 636 and 649 all ran out of time.
  • Growth since the previous issue was +6.82 points: from 58.37% on the morning of September 12 to 65.19%. Exactly the same margin as the preceding inter-epoch gain.
  • Governance Incentives Framework 2026 ended at 1.72%: 12 votes in favour against recorded opposition of 98 votes and 2,354,868,515 ada. The committee was 1 yes, 4 no.
  • The committee’s first vote landed about eight and a half hours after the previous issue went live: 17:10:17 JST on September 12, when one member with a term through epoch 799 voted yes on all four pending actions at once. Four members voted during the period. The committee votes on minPoolCost and OpenZeppelin are still just that one vote.
  • An SPO poll on raising k from 500 to 1000 was submitted: 21:18:25 JST on September 17. It is an Info Action, expiring at epoch 663 = 06:44:51 JST on October 22. The proposer is Chris [STR8].
  • The test is “explicit yes votes exceeding 50% of eligible SPO voting stake”: the proposal text declares this itself. Registered stake that does not vote behaves the same as a no, explicit abstentions are excluded from the denominator, and DRep votes do not count toward the test. Right now explicit support is 2 pools (0.02%) and explicit opposition is zero. SIPO’s three pools have not voted yet.
  • At k=1000 the saturation point works out to roughly 36.75 million ada: a mechanical calculation dividing circulating supply by k. SIPO (live 57,052,790 ada) sits above that line, while SIPO2 and SIPO3 (about 25 million ada each) sit below it.
  • SIPO’s DRep abstained on minPoolCost 75 ada: 08:57:38 JST on September 15. The action stands at 6.26% DRep support (22 votes) against recorded opposition of 20 votes and 469,841,434 ada, so the opposing stake outweighs the supporting stake.
  • cardano-node 11.1.2 shipped: 12:29:26 JST on September 17. It contains two changes, and the release notes contain no mention of security.
  • All four instruments returned to positive: ADA +9.71%, NIGHT +14.63%, BTC +5.23%, ETH +4.31%. Last issue, “only ETH was up.”

1. Market Pulse

Morning fixed-point observations from September 12 to September 19.

Instrument9/129/19Weekly
ADA$0.205667$0.225627+9.71%
BTC$77,199$81,238+5.23%
ETH$2,521.90$2,630.63+4.31%
NIGHT$0.02087497$0.02392995+14.63%

Last issue, “only ETH was up.” This week the usual shape returned, with all four instruments positive. Most of the gain arrived on the final day. Day-over-day on the morning of September 19, ADA was +12.82% and NIGHT +12.22%. But at the same moment SOL was +12.2% and Coinbase stock +11.66%, moving by the same margin, so this does not read as an ADA-specific catalyst.

The ADA figure taken exactly at the epoch boundary points the other way. The epoch 654 boundary (morning of September 12) was $0.20642490 and the epoch 655 boundary (morning of September 17) was $0.19436574, a −5.84% move. In yen terms, ¥31.70 to ¥30.36, over a period in which USD/JPY also moved from 153.58 to 156.22. Boundary values and morning fixed-point observations are taken at different times, so they are not mixed into the same table.

A few network figures as well. The staking ratio moved from 56.49% to 56.45%. Pools that produced blocks fell from 953 to 935, and the delegator count fell by 584 to 1,337,408, a smaller decline than the previous week’s 975. The treasury stands at 1,360,079,767.29 ada (+3,664,494.38 ada from the previous issue).

2. Two Actions Expired at the Epoch 656 Boundary

Two governance actions now carry expired_epoch 656 in the on-chain record. Both are treasury withdrawals, and both were submitted in epoch 649.

ActionRequestedFinal DRep yesDRep recorded noCommittee (7)
Reimburse Ikigai Info Governance Action Deposit.103,000 ada65.19% (129 votes)5 votes / 54,984,919 ada5 yes, 1 no
Governance Incentives Framework 20264,207,967 ada1.72% (12 votes)98 votes / 2,354,868,515 ada1 yes, 4 no

A distinction worth keeping straight: these two were not “rejected” — they “expired.” That means they entered their deadline epoch without reaching the threshold. The substance, though, is entirely different. GIF went down under accumulated opposition: 12 votes in favour against 98 recorded against, a 32-fold gap in delegated stake. SIPO’s DRep voted no on August 26 and held that vote through to expiry.

Ikigai is the other case. Support climbed to 65.19% and did not arrive. The stake behind the opposition did not even reach 2% of the stake behind the support. In the record, both are simply expired.

3. The Committee Had Passed It

Here are the Constitutional Committee’s votes on Ikigai, in time order.

Recorded (JST)VoteMember (key prefix)
August 22, 01:33:22Yescc_hot1qdc65ke… (term 726)
August 29, 01:07:21Yescc_hot1qdjx6xe… (term 726)
August 30, 19:31:25Abstaincc_hot1qde96n2… (retired — outside the tally)
September 12, 17:10:17Yescc_hot1qgsl88a… (term 799)
September 15, 00:20:37Yescc_hot1qvh20fu… (term 726)
September 16, 03:53:31Yescc_hot1qwz0aw5… (term 799)
September 16, 20:32:07Nocc_hot1q2ccqmm… (term 799)

Sitting members’ votes that enter the tally come to 5 yes, 1 no, an approval rate of 71.43%, clearing the two-thirds quorum (66.67%). From the committee’s side, this action had passed. Member names cannot be taken from the on-chain record; what can be taken is the key hash and the term epoch.

The DRep side stopped at 65.19%, 1.81 points below the 67% threshold. When the previous issue was written on the morning of September 12 it stood at 58.37%, so +6.82 points accumulated. The preceding inter-epoch gain was also +6.82 points. It accumulated at the same speed, and fell short by the same amount. Every DRep vote cast on Ikigai after epoch 655 began was a yes; it did not fall because opposition grew.

This is Ikigai’s third expiry. The deposit on a governance action is 100,000 ada per submission, and when an action seeking a refund itself runs out of time, that deposit is treated the same way. An action asking for a deposit back has now added to the pile of deposits three times over.

4. The Committee’s First Vote Covered Four Actions at Once

The previous issue carried a section headed “the new committee has not cast a single vote in five days.” It went live at 08:39:41 JST on September 12. The first vote landed that same day at 17:10:17 — about eight and a half hours later.

Recorded (JST)Member (key prefix, term)Votes cast
September 12, 17:10:17cc_hot1qgsl88a… (799)Four at once — GIF yes / Ikigai yes / minPoolCost yes / OpenZeppelin yes
September 15, 00:20:37cc_hot1qvh20fu… (726)GIF no / Ikigai yes
September 16, 03:53:31cc_hot1qwz0aw5… (799)Ikigai yes
September 16, 20:32:07cc_hot1q2ccqmm… (799)GIF no / Ikigai no

Four members voted during the period: three from the term-799 side and one from term 726. The other two on the 726 side still have their August votes from the previous configuration standing. Note also that the single yes vote on GIF came from the member who voted on all four actions at once on September 12.

That first four-at-once vote has a side effect. The committee votes sitting on Reduce minPoolCost to 75 ada and on the OpenZeppelin action are still only that one vote. Both stand at 1 yes and 6 not voted, an approval rate of 14.29%, a long way from the two-thirds quorum.

The “vote that stopped being counted” described in the previous issue also persists. The GIF no and the Ikigai abstention cast by the retired member cc_hot1qde96n2… remain in the itemised record but do not enter the ratification tally. GIF’s opposition reads as 5 votes in the detail and 4 in the tally. It stays as a record and does not stay as a judgment. Every time the committee turns over, this gap is created.

5. A Poll Asking SPOs About Raising k from 500 to 1000

At 21:18:25 JST on September 17, Should stakePoolTargetNum (k) be raised from 500 to 1000? (SPO poll) was submitted — 14 hours, 33 minutes and 34 seconds after epoch 656 began. It is an Info Action (nothing executes on chain; the formal threshold is 100%), the proposer is Chris [STR8], the deposit is 100,000 ada, and it expires at epoch 663 = 06:44:51 JST on October 22.

What the proposal text asks is whether there is “sufficient SPO support to proceed with a subsequent parameter change proposal” raising k from 500 to 1000. It builds on the 2023 SPO poll while treating stakePoolTargetNum on its own.

k is the figure that sets, for reward calculation purposes, how many saturated pools the system targets. It is also called nOpt, and the saturation point per pool is circulating ada divided by k. The proposal text puts the saturation point at its reference date at roughly 75 million ada for k=500 and roughly 37.5 million ada for k=1000. Two easily confused points are worth separating: the saturation point caps rewards, not eligibility to make blocks; and it is a different thing from the Ouroboros security parameter k, a confusion the text itself warns against.

This action does not change any parameter. It measures support first and, if that support is judged sufficient, proceeds to a separate parameter change proposal — a two-stage design. The proposer writes that “the submission of this Info Action should not be interpreted as an endorsement by the proposer of the proposed parameter change.” There is also an explanation of why SPOs are being asked. stakePoolTargetNum is not currently among the parameters for which a subsequent parameter change would require SPO approval. Even so, seeking a clear mandate before changing a parameter with a material effect on stake pool economics aligns with the broader principle proposed by CAP #10, “Require SPO approval for changes to stake pool economic parameters.”

Last week’s core story was that the standalone minPoolCost action has no SPO voting column at all. This week, for a different economic parameter, an SPO column has been placed there in a form that is not required. What connects the two goes as far as CAP #10, which the proposal text cites explicitly. Whether the proposer had last week’s action in mind is not stated in the text.

6. The Test Is Explicit Support Above Half of Eligible Stake

An Info Action has a formal threshold of 100% and is not the kind of action that passes and then executes something. So what counts as “supported”? The proposal text defines that itself.

“Sufficient support is established only if explicit SPO yes votes exceed 50% of the applicable eligible SPO voting stake.” “Explicit abstain votes are excluded from eligible voting stake, and registered stake that does not vote behaves the same as a no vote.” “DReps are invited to vote and submit rationales, but DRep votes are not included in determining whether this Info Action established an SPO mandate.”

The relationship to the formal threshold is explained up front as well: the 50% bar is a declared interpretation of the poll result, not an on-chain ratification threshold for the Info Action itself. The text also states that a result falling short must not be read as a mandate for any other intermediate value.

Here is where the votes stand (retrieved on the morning of September 19 JST).

CategoryCurrent
SPO explicit yes2 pools (0.02% / 2,114,251 ada)
SPO explicit no0
DRep yes4 votes (0.06% / 3,061,763 ada)
DRep recorded no3 votes / 171,938,311 ada
DRep abstain1 vote / 445,777,028 ada
Committee0 votes

The two pools voting yes are pool1m83drqw… (September 18, 05:01:02) and pool1tnl3yxm… (the same day, 17:12:47).

Given the design of the test, the protagonists right now are the pools that have not voted. Stake delegated to pools that have not voted comes to 9,604,088,251 ada, and by the proposal’s own declaration that behaves the same as opposition. The 616 pools set to “always abstain,” holding roughly 11.7 billion ada, fall out of the denominator. SIPO’s three pools have not voted yet.

7. The 2023 Poll Answered Differently by Count and by Stake

Asking SPOs about k is not new. The 2023 poll offered combinations of k and minPoolCost as options, and the interim result reported votes from 797 pools. In the tally cited by the proposal text, options including k=1000 took 69.64% by pool count and roughly 41.1% by participating stake, while options including k=500 took 21.08% and roughly 44.2% (abstain 0.63% / roughly 0.7%; none of the above 8.66% / roughly 13.9%).

Counting by number or by stake flips which side is in the majority. That experience is why this poll repeats that it weights by stake and requires explicit support above 50%. The text makes one further point: because the question bundled two parameters together, it did not deliver a clear standalone mandate on k.

There is also a summary of k’s own history. At the start of Shelley, k was 150 and minPoolCost was 340 ada, and k rose from 150 to 500 at epoch 234. The text simultaneously cautions against attributing the subsequent improvement in decentralisation to k alone, since the market and participant numbers moved substantially over the same period.

8. SIPO’s First Pool Sits on the Side Where the Saturation Point Halves

This section touches SIPO’s own interests, so the numbers come first. The calculation below is mechanical: the on-chain circulating supply of 36,752,924,102 ada on the morning of September 19, divided by k.

kSaturation point (supply ÷ k)SIPO’s three pools (epoch 655 boundary)
500 (current)roughly 73,505,848 adaall three below
1000 (proposed)roughly 36,752,924 adaSIPO 57,052,790 ada = above / SIPO2 25,256,040 ada / SIPO3 25,589,076 ada

SIPO sits on the side that becomes oversaturated for reward purposes if k moves to 1000. For anyone delegating, that is something checkable today, without waiting for the outcome.

The proposal text leaves the position of operators like this open, as a question rather than an answer.

“Would lowering the saturation point disproportionately disadvantage existing independent or community-run pools that have accumulated meaningful delegation through years of consistent uptime, reliability, reputation and community engagement?”

SIPO is a party to that question. So, before casting a vote, here is the position in writing. k=1000 is a change that disadvantages SIPO’s first pool, and SIPO is on the side that bears that disadvantage. At the same time, it is a fact stated in the text that eligibility to make blocks survives a lower saturation point, and what the excess stops increasing is only the ceiling on potential rewards. The text also lists what will not happen automatically: new pools and independent operators do not automatically multiply; nothing prevents exchanges and custodians from registering new pool IDs; delegation that moves may land on another pool in the same operating group. The ledger records pools, but it does not comprehensively disclose which pools are in the same hands.

9. minPoolCost 75 ada — SIPO’s DRep Abstained

The action at the centre of the previous issue is still pending. It expires at epoch 661 = 06:44:51 JST on October 12.

CategoryPrevious issue (9/12)This issue (9/19)Threshold
DRep yes0.54% (3 votes)6.26% (22 votes / 294,844,280 ada)67%
DRep recorded no0 votes20 votes / 469,841,434 ada—
DRep abstain0 votes2 votes / 533,897,596 adaoutside denominator
SPO0 votes (not required)0 votes (not required)0%
Committee0 yes, 7 not voted1 yes, 6 not voted2/3

By vote count it is 22 yes to 20 no, but by delegated stake the opposition outweighs the support. And the five most recent votes (22:07 on September 17 through 18:14 on September 18) were all opposed. At the time of the previous issue not a single no vote had been cast, so the direction changed over the past week.

One of those two abstentions is SIPO’s DRep. At 08:57:38 JST on September 15, SIPO cast a constructive abstention with the equivalent of 87,500,000 ada in delegated stake. The rationale is recorded on chain in English and Japanese, and the full text is published at SIPO DRep: abstaining on the standalone minPoolCost 75 ADA action. It rests on three points.

  1. The 2023 move from 340 to 170 did not change the structure. Input Output Research’s SPO incentives report (November 2025) shows that 340 ada remained the most common fixed cost in every size band, that no market-wide price reduction followed, and that the market split into a 340 incumbent bloc and a 170 challenger bloc. When SIPO read the ledger on September 15, of the 2,672 pools with active stake, 1,747 pools declared 340 ada and accounted for 67.6% of active stake, while 170 ada covered 503 pools and 22.7%.
  2. The band where it bites is 0.5 to 3 million ada. Delegator yield only moves materially within that band: at 1 million ada from roughly 0.90% to roughly 1.60%, and at 10 million ada from roughly 2.02% to roughly 2.09%. That covers 342 pools, about 2.2% of active stake.
  3. The fixed cost moves before the minimum margin does. The same report recommends abolishing minPoolCost while pairing that with a minimum margin parameter to prevent a race to the bottom on fees. This action moves only the fixed cost, before that safeguard exists. The proposal’s own withdrawal plan states that pools which have already lowered their cost cannot be made to raise it again. It is one-way.

Disclosure. All three SIPO pools declare a fixed cost of 340 ada, so this change has no mechanical effect on them. And SIPO sits with the “340 incumbents.” If SIPO voted no, it would be fair to read that as a no cast to protect incumbents. An abstention records the concern without directing that weight at the action. Here are the conditions under which SIPO would switch to yes: a proposal pairing the reduction in fixed cost with a minimum margin mechanism, or one accompanied by an operator revenue model, broken out by pool size, showing that independent operators remain viable at the new floor.

10. Opposing Stake Is Accumulating Against OpenZeppelin’s 11.78 Million ada

The other pending action, Withdraw 11,787,063 ada for the OpenZeppelin Stack administered by Intersect, also expires on October 12.

Category9/15This issue (9/19)Threshold
DRep yes0.88% (6 votes)2.14% (7 votes / 111,152,249 ada)67%
DRep recorded no17 votes29 votes / 930,769,896 ada—
Committee1 yes, 6 not voted1 yes, 6 not voted2/3

Against 7 votes and 111.15 million ada in support, recorded opposition is 29 votes and 930.77 million ada — a gap of more than eight times in delegated stake.

The amount is USD 1,831,000 (11,443,750 ada at a rate of 0.16) plus Intersect’s 3% administration fee (343,313 ada). The term is 12 months with quarterly milestones, and 20% is paid up front on contract signature. Deliverables are three reference implementations (liquid staking, self-repaying loans, tokenised MMF) and a contracts library. The contract language is listed as “to be determined in the initial evaluation phase” and is not yet fixed. From 01:00 JST on September 16, the Cardano Foundation’s Governance Hour #10 covered this action in full, and the following day’s recap post says it is worth watching for any DRep still deciding.

SIPO’s DRep has not voted on this action yet. Two things are outstanding: an analysis of overlap with existing contract libraries, and the choice of contract language. SIPO will check the state of things around October 5 and decide by October 9. The reason overlap matters comes from SIPO’s own past vote. SIPO voted yes on IO: Developer Experience Initiative (3,601,926 ada) on May 9, 2026, and one of its deliverables was an “OpenZeppelin-style ContractsLibrary.” The deliverables of funding SIPO supported and the deliverables of the action now under consideration sit next to each other, at least as described.

A treasury figure to go with it. Treasury withdrawals enacted since epoch 613 come to 34 actions and 457,395,629 ada, leaving roughly 42.6 million ada against the 2026 cap of 500,000,000 ada. If this action passes, the total reaches 469,182,692 ada, or 93.8% of the cap.

11. Constitutional Amendments Are Still at the Consultation Stage

The previous issue noted that nothing had been submitted by epoch 655, the stated target. This week too, no constitutional amendment action has appeared on chain. The actions submitted in epochs 654, 655 and 656 are just three: the standalone minPoolCost action, the OpenZeppelin withdrawal, and the k poll.

What is moving instead is the stage before submission. According to the Cardano Foundation’s community digest (September 16), six weeks after the constitutional amendment portal went live, five constitutional amendment proposals (CAPs) and two constitutional issue submissions (CIS) have been filed, and all seven are at the consultation stage. The first review call was held on September 17, and the plan is to alternate biweekly between 07:00 and 14:00 UTC. Intersect’s weekly update #128 says the most active discussion is on CAP-3 (Net Change Limit).

CAP #10, cited by the k poll, is among those seven. In sequence: a proposal to create an SPO approval requirement on the constitutional side is in consultation, and a voluntary poll consistent with that principle reached the chain first.

12. Nodes and Tooling

cardano-node 11.1.2 — Two Changes and Nothing Else

Released at 12:29:26 JST on September 17. Unlike 11.1.1, covered in the previous issue, this is not a pre-release. The changes are exactly what the release notes list: on the node side, “update the protocol version minor in the block header”; on the ledger side, “optimise memory usage of time-lock scripts.” Every other component is listed as unchanged, and the benchmarking and system test results are stated to carry over from 11.1.1.

One point, stated plainly. Do not read this release as a security fix. The release notes contain no instance of “security” or “CVE,” and no security advisories have been issued for cardano-node or cardano-ledger. The expectation in #128 of “11.2 within a month” has not been updated this week either.

Downstream packaging moved too. Daedalus 11.4.0 (04:12:49 JST on September 19) bumps the bundled node to 11.1.2 and states that “all users are recommended to update to this version.” Mithril 2630.1-hotfix (20:50:02 JST on September 17) “fixes a bug in the 2630.0 distribution that could prevent Mithril signers and aggregators from retrieving blocks and transactions from the Cardano chain,” with instructions to make sure the signer is on 1.1.9. Anyone running 2630.0 has something to do this week.

Ouroboros Genesis — It Is Already the Default in the Distributed Configuration

A change affecting sync mode is visible first on the configuration side. The mainnet configuration file distributed as a pre-release sets ConsensusMode to GenesisMode. Preprod and preview are the same, and the topology side points at peerSnapshotFile. Meanwhile, the mainnet configuration on master in the cardano-node repository is still PraosMode. The developer portal’s wording also remains “expected to become the default on mainnet in a future release.“

So the state is this: it is the default in the pre-release distribution, and no formal announcement has been made. The difference matters operationally. The same page states that “Genesis mode is incompatible with bootstrap peers. Enabling it overrides the bootstrapPeers configuration. Replace bootstrapPeers in the topology with a peer snapshot file” (SIPO’s writeup).

The Leios Demo, the Plutus Handover, and Dingo’s First Block

At 16:35:18 JST on September 15, IOG’s Carlos Lopez de Lara shared a demo by the Leios architects.

“Watch a Leios node peak at 250 TxkB/s (about 1,000 transactions per second of simple 250-byte transactions) on a local cluster with simulated round-trip times. Mainnet today is capped at 4.5 TxkB/s. The work ahead is in the networking stack, where tuning and optimisation will carry these numbers from the lab to real internet conditions.”

These are local-cluster numbers, not mainnet numbers. The line in the quote about the work ahead being in the networking stack is precisely what remains.

On September 17, IOG announced that maintenance of Plutus moves to Midgard Labs, with the transition beginning October 1. One sentence in the announcement captures the character of this series of handovers: “responsibility is moving out of a single organisation and into specialised, independent teams.” It is the fifth, after Se7en Labs, Teragone, ICAN Group and BlockPQR (SIPO’s writeup).

And at 04:12:15 JST on September 16, Dingo, a node implementation written in Go, made its first block on mainnet (block height 13,945,217, pool pool1eqj3dzp…). That said, Dingo’s repository still says not to use it on mainnet with real funds, and it was run by an independent operator. On top of that, a block on chain carries no field indicating which node implementation produced it. That the block exists and that the named pool made it can be verified on chain. What software made it cannot (SIPO’s writeup).

13. Two Unchecked Assumptions in a Row

Between 00:47 and 00:48 UTC on September 13, Splash’s ADA/OADA StableSwap pool was drained in two transactions. The second transaction removed 2,434,648.42 ADA and 1,988,222.18 OADA, and the LP supply was unchanged. On cause, the report (Splash, 16:55 UTC on September 13) is precise.

“The cause is in the validator. The validator subtracts accrued protocol fees from the actual balance to obtain ‘tradable reserves,’ but never checks that the result is positive. The attacker first inflated the fee counter, then spent the actual ADA balance down below that counter. The invariant polynomial accepted negative reserves and released both assets from the pool.”

This code had been audited. It falls within the scope of Anastasia Labs’ StableSwap audit report (v1.0, August 5, 2024). The report describes the relationship to the eight findings as follows: “none of them concern fee accrual in the swap branch, the sign of tradable reserves, or swap direction.” The flow of funds is evidenced as well: 2,552,176.54 ADA moved to six deposit addresses across four custodial clusters, of which KuCoin accounted for 786,851.54 ADA and Gate.io 550,000 ADA. Optim Finance, which issues OADA, paused the protocol at 15:57:59 JST on September 13 (SIPO’s analysis).

The shape overlaps with last issue’s Hydra story. A path or formula meant to make something faster had dropped one precondition for correctness. Last issue it was a skipped signature check; this week it is an unverified sign on a balance.

Two more items in the same week. On the morning of September 16 JST, the DEX VyFi disclosed unauthorised outflows in “Safe Swaps,” its feature for direct peer-to-peer token exchange, affecting the RISE and PALM communities. Safe Swaps has been paused, and the scale and cause are under investigation. And IOG’s official YouTube channel was hijacked. On September 18, Charles Hoskinson called on people not to trust videos or links on the channel. The assumption that a link appearing on an official account can be trusted unconditionally was tested twice this week.

14. Midnight Watch

One post went up on the official blog this week: “Building a seamless user journey through abstraction on Midnight,” September 14. With tooling from Dynamic (recently acquired by Fireblocks), application builders can embed a Midnight wallet directly in their software, and users simply sign in with a standard email address or social account. Behind the scenes three things are provisioned automatically: an unshielded address for public transfers, a shielded address for private transfers, and a DUST window for tracking transaction costs.

Two Node Lines Are Running in Parallel

VersionReleased (JST)Position
node-2.1.0-rc.1September 16, 00:53:11For devnet / qanet, governance action required
node-2.1.0-rc.2September 17, 23:42:54No protocol behaviour change, first *.wasm artifacts
node-1.0.2September 19, 04:34:21Mainnet maintenance line (1.0.x)

rc.1 layers 29 changes on top of 2.1.0-beta.1, and the change operators should read first is a consensus-affecting change to the order in which observed cNIGHT UTXOs reach the runtime. On the bundled ledger, the text says this:

“8.1.2 is a security patch hardening low-level deserialisation, and an 8.1.2 node accepts strictly less than an 8.1.1 node.”

This one does state “security patch” explicitly. Which is to say it is handled differently from cardano-node 11.1.2 in the same week. The notes add that “switching binaries is a flag day, not a staged update,” and state explicitly that for SPOs, “non-public release candidates are out of scope.” On node-1.0.2, one fact: the release body is identical to node-1.0.1 from July 14 apart from the tag name.

The Proposal Index Reached 0042, and MIP-0017 Is Still Held

Three improvement proposals landed this week. MPS-0040, “Provenance of cross-contract calls in Compact circuits” (September 16) documents the gap that when a contract is called by another contract, the callee has no way to know the caller (SIPO’s writeup). Then MPS-0041 (custodian proving servers) and MPS-0042 (serialisation format). The index, which is the authoritative record of numbering, was updated on September 18 and now lists through MPS-0042. Meanwhile MIP-0017, covered in the previous issue, still has its document in place while the index stops at MIP-0016. That discrepancy has stood unchanged for a week.

On the wallet side, Gero Wallet 2.7.1 made it possible to pay Midnight transfer fees with “DUST held in another of your own wallets” (September 16 JST). On the disbanding of the DevRel team, the person who had handled protocol operations full time announced their own departure, and Charles Hoskinson published a roughly 55-minute video, “Devs versus Builders,” at 03:50 JST on September 14 defending the decision. The argument is that “builders” who make commercially viable things should replace developer headcount as the metric (SIPO’s full translation and commentary). NIGHT’s price was +14.63% for the week, the largest move among the four instruments.

15. Risk Dimensions

  • Governance (time). What Ikigai showed is that even when the approach to the threshold is fast enough, the deadline overtakes it. 65.19% was the highest figure across three submissions, but deadlines do not extend. The next action that could take the same shape is Reduce minPoolCost to 75 ada, expiring 06:44:51 on October 12.
  • Governance (committee). The committee votes sitting on minPoolCost and OpenZeppelin amount to the single vote from September 12. Unless the remaining six move, neither passes regardless of what happens on the DRep side.
  • Governance (SPO). By the design of the test, not voting on the k poll works the same as voting no. Stake delegated to pools that have not voted comes to 9.6 billion ada. The decision whether or not to express a view lands directly in the result.
  • Pool economics (SIPO itself). The k=1000 saturation point is roughly 36.75 million ada, and SIPO’s first pool (57.05 million ada) sits above it. Anyone delegating can check which side of that line their current pool is on. SIPO is on the side disadvantaged by this change, stated up front, and will decide its vote in the coming epoch.
  • Treasury (cap). If the OpenZeppelin action passes, 2026 withdrawals reach 93.8% of the cap. With only about 42.6 million ada left, the room for another withdrawal within the year is limited.
  • Nodes (operations). 11.1.2 is not a pre-release, but it is not a security fix either. Mithril’s 2630.1-hotfix fixes a bug that can stop blocks and transactions from being retrieved, and that one is worth acting on this week. Also, the distributed pre-release mainnet configuration is in Genesis mode while master remains Praos. Mixing configurations means Genesis mode overrides bootstrapPeers.
  • DeFi (audit scope). The Splash flaw was in code within the scope of a 2024 audit, and different in character from all eight findings. “Audited” means “examined from these angles,” not “examined from every angle.”

16. What to Watch Next Week

  1. The epoch 657 boundary — 06:44:51 JST on September 22. Whether the run of five consecutive no votes on minPoolCost continues is readable in how votes line up across this boundary. Check the individual action pages on gov.tools.
  2. SPO votes and rationales on the k poll — how far explicit support moves from 2 pools and 0.02%. The proposal text also asks operators who disagree to cast an explicit no with a rationale. The deadline is 06:44:51 JST on October 22; check that Info Action’s page on gov.tools.
  3. Committee votes on minPoolCost and OpenZeppelin — whether the remaining six move. Both expire 06:44:51 JST on October 12. On OpenZeppelin, SIPO will check the state of things around October 5 and decide its vote by October 9.
  4. The Intersect board election result — voting closes 21:00 JST on September 25, with results published September 28. Eleven candidates for two seats. Check Intersect’s news page.
  5. The 2027 NCL roundtable — 22:00 JST on September 21, titled “Treasury Funding Allocation – The Case for Reform.” Announced as a session on allocation models.
  6. Intersect weekly update #129 — unpublished as of writing. When it appears it will sit after #128 on Intersect’s news page.
  7. cardano-node 11.2 — the “within a month” expectation in #128 lands around October 11. Check the releases page for IntersectMBO/cardano-node.
  8. A formal announcement on Ouroboros Genesis — whether the gap between the distributed configuration and master is resolved. Check the developer portal’s Topology page.
  9. A final 2.1.0 release for the Midnight node, and MIP-0017 reaching the index — check the releases page for midnightntwrk/midnight-node and NUMBERS_INDEX.md in the improvement proposals repository.
  10. Splash’s fixed scripts and the rebuilding of OADA — check splashprotocol/splash-core and Optim Finance’s announcements.

In Closing

The previous issue wrote this about the two actions nearing their deadline: “If nothing moves before the deadline, they expire even with DRep support in place.”

Something did move. It came from the committee’s side, after five days of silence. At 17:10:17 on September 12, one member with a term through epoch 799 cast yes votes on four actions at once, and three more followed on September 15 and 16. On Ikigai, the committee passed it five votes to seven members.

It expired all the same. The DRep side was at 65.19%, 1.81 points below the 67% threshold. It accumulated +6.82 points from the previous issue, at exactly the same speed as the preceding inter-epoch gain, and fell short by exactly as much. For the third time.

What this reads out to is how ratification requirements work. Of the four columns — DRep, SPO, committee, deadline — an action stops if any one of them goes unfilled. Last week we looked at an action whose SPO column had disappeared. This week we are looking at one whose committee column filled while the DRep column came up 1.81 points short.

And then a conversation about the columns themselves began. The k poll is an action that placed an SPO voting column there itself, in a non-binding form, where none is required. The test is explicit support above 50% of eligible stake, non-voting behaves as opposition, DRep votes are not counted. What it cites is CAP #10, the attempt to create an SPO approval requirement on the constitutional side.

SIPO is a party to this one. If k moves to 1000, the saturation point becomes roughly 36.75 million ada, and SIPO’s first pool (57.05 million ada) comes out above it. So the position went in writing before the vote. The same applies to the minPoolCost action: all three pools declare a fixed cost of 340 ada, so there is no mechanical effect, and SIPO sits with the 340 incumbents.

Before votes are counted, who gets counted has already been decided. This week that underlying layer became visible three times.

Sources

Where this issue refers to SIPO’s own articles, those are linked inline. The following are primary sources.

Transparency Notes

  • “Expired” versus “rejected”: both actions carry expired_epoch 656 in the on-chain record. They are not described as “rejected.” For Ikigai, the fact that the committee passed it 5 to 1 is stated alongside.
  • The 1.81 points and the +6.82 points: 1.81 is the 67% threshold minus 65.19% DRep support; 6.82 is the difference between the 58.37% recorded by the previous issue at 08:20 JST on September 12 and the final 65.19%. That this matches the preceding inter-epoch gain is consistent with what was recorded in Epoch Diary 655.
  • Why no “opposition is X%”: the opposition rate returned by the public API includes non-voting stake and “always no confidence.” So in the body, recorded opposition is given only as vote counts and ada amounts. Support rates are the aggregated values as returned.
  • Committee member names: all that can be taken from the on-chain record is the key hash and term epoch. Who cast the four simultaneous yes votes on September 12 is not stated here. The reasons for that vote, and for the single yes on GIF, are also not stated, because the record contains no explanation.
  • The relationship between the k poll and last week’s action: what the poll’s text cites explicitly is CAP #10; it does not mention the minPoolCost action. Accordingly, this issue does not describe it as “a reaction to last week’s action.” The proposer’s position is likewise described only within the quoted text (“should not be interpreted as an endorsement”).
  • The saturation point calculation: roughly 73,505,848 ada at k=500 and roughly 36,752,924 ada at k=1000 are mechanical figures, dividing the on-chain circulating supply of 36,752,924,102 ada on the morning of September 19 by k. They do not match the roughly 75 million / 37.5 million ada the proposal text cites for its own reference date, because the supply is taken at a different time. SIPO’s three pool figures are as of the epoch 655 boundary.
  • SIPO’s interests: for the k poll, that SIPO’s first pool sits above the k=1000 saturation point is disclosed in the body; for minPoolCost, that all three pools declare 340 ada and are mechanically unaffected and that SIPO sits with the 340 incumbents. The OpenZeppelin vote is outstanding, and the judgment on how far the overlap extends is deliberately reserved.
  • The word “security”: cardano-node 11.1.2’s release notes contain no “security” or “CVE” and no related advisory was issued, so it is not described as a security fix. Midnight’s ledger 8.1.2, by contrast, states “security patch” in its own text, and is described as such.
  • Ouroboros Genesis, Leios, Dingo: on Genesis, this issue does not say it “became the mainnet default” (what it states are the three facts about the distributed configuration, master, and the developer portal). Leios’s 250 TxkB/s is a local-cluster figure and not a like-for-like comparison with mainnet’s 4.5 TxkB/s. Dingo’s attribution rests on statements by the operator and the developer.
  • Intersect weekly update #129: unpublished as of writing. Everything in this issue sourced from Intersect rests on #128.
  • Prices versus actions and incidents: no connection is drawn. On ADA’s +12.82% day-over-day move on September 19, this issue goes only as far as noting that SOL was +12.2% and Coinbase stock +11.66% at the same moment. The −5.84% at the epoch boundary is taken exactly at the boundary time, and because it is retrieved at a different time from the morning fixed-point observation, it does not match the +9.71% weekly figure.

This is not investment advice. For information and research purposes only.

関連記事

Weekly Brief EN
Weekly Brief #23 | minPoolCost Returns Without an SPO Vote
Weekly Brief EN
Weekly Brief #22 | The Deadline Arrived — The Committee Renewal Passed at 71.95% and 56.54%, minPoolCost Fell on the SPO Side Alone, and Hydra Shipped a Critical Fund-Theft Fix
Weekly Brief EN
Weekly Brief #21 | The Votes Came In, the Thresholds Did Not Fall — The Deadline Is 21:44:51 UTC on September 1, the Committee Renewal Is 2.76 Points Short on DReps, and NIGHT Lists in Japan for the First Time