- A voting snapshot on 1 September showed Cardano’s Constitutional Committee renewal past both bars: 69.36% support from delegated representatives against a 67% threshold, and 51.18% from stake pools against 51%.
- Nothing had been ratified at that point. Ratification is assessed at an epoch boundary, and newly elected members can only be seated an epoch after that.
- Stake sitting in pools that cast no ballot at all is counted on the opposing side, so the 0.18-point margin was mostly a measure of turnout rather than opposition.
Cardano is one of the few large networks where holders vote on how the chain is run and the result is written into the ledger itself. On 1 September, one of those votes looked to have squeezed past its threshold by 0.18 of a percentage point. Looked is doing a lot of work in that sentence.
If you hold ADA, decisions about the network’s treasury and its upgrades run through votes where doing nothing is not neutral. Stake in a pool that ignores a vote is counted against the proposal, which means a proposal can fail without a single operator deciding to oppose it.
The second thing worth carrying away is about how these stories get reported. A headline saying a vote has “cleared” a threshold is usually describing a live running count, not a decision. On Cardano the decision happens at a fixed moment, and the count can move right up until it.
What the Constitutional Committee does
Cardano has a written constitution, and a Constitutional Committee whose job is to check whether a proposed action is consistent with it. The committee does not decide whether an idea is good. It decides whether the idea is allowed, and its approval is a required step for the categories of action that need it.
There are seven seats. Four of them expire at a scheduled point in early September, and the committee has an operating minimum of five members. If the four expiring seats lapse with no replacement in place, the committee drops to three, below that minimum, and the actions requiring its sign-off cannot proceed. That is why a routine-sounding renewal vote carried real weight: the seats were always going to expire on a set date, whether or not anyone had voted.
Who actually votes, and why there are two thresholds
Two groups vote, and they represent different things.
Delegated representatives, usually shortened to DReps, vote with the ADA that holders have delegated to them for governance purposes. If you hold ADA and don’t want to read every proposal yourself, you pick a DRep and your stake votes with theirs. Their weight is measured in ADA.
Stake pool operators, or SPOs, run the machines that actually produce blocks. Their weight is also measured in ADA, specifically the stake delegated to their pools for block production. They are the people who would have to run whatever the network becomes, which is why they get a separate say.
The two thresholds exist so that one group cannot carry an action alone. Money delegated to governance and infrastructure that keeps the chain alive have to agree, at different bars for different action types. For an update to the committee’s own membership, only these two groups vote. The committee does not get a say in whether it is renewed, which is sensible and also means there is no third party to break a deadlock.
Why the stake pool number was the close one
CryptoSlate pulled a synchronised snapshot from Koios, a public API that reads Cardano’s chain data, at about 09:59 UTC on 1 September. It reported DRep support at 69.36% against the 67% requirement, and SPO support at 51.18% against 51%. We have not queried the live API ourselves, so those percentages are as that outlet reported them.
The breakdown it cited is the more revealing part. Explicit “No” votes from stake pools amounted to roughly 2.008 million ADA. The combined figure on the opposing side of the calculation, which includes stake in pools that had not voted, came to about 5.316 billion ADA.
Our arithmetic on those two reported numbers: deliberate opposition made up something in the region of 0.04% of the side that was holding the vote back. Everything else was silence.
That is a design choice rather than an accident. Cardano’s governance documentation, as cited in the same report, treats stake behind a non-voting pool as counting against ratification, while an explicit abstention (or stake delegated to an always-abstain option) is taken out of the calculation entirely. So an operator who has views and registers them as abstention is removed from the sum. An operator who is on holiday, or who has never engaged with governance, is counted as an objector.
It is worth being careful about what that shows. Absent stake being counted against a proposal is the behaving as specified. It is not evidence that any particular operator wanted to block the renewal, and nothing in the data supports reading intent into a non-vote.
“Cleared” and “ratified” are not the same event
Cardano runs on epochs, fixed periods of roughly five days. Governance outcomes are not decided continuously; they are assessed when one epoch ends and the next begins.
At the time of that morning snapshot, the chain tip was still in epoch 652 and the proposal’s on-chain record showed no ratification, no enactment and no expiration. The ratification check was due at the boundary into epoch 653, at about 21:44 UTC the same day, roughly twelve hours later. Votes could still change in that window, which means the 0.18-point margin was true of one moment and not necessarily of the moment that counted.
Even ratification would not have put anyone in a seat. Enactment, the point at which the four newly elected members are actually seated, comes one epoch later again, at the boundary into epoch 654 around 21:44 UTC on 6 September. That is the same moment the four existing seats lapse, which is why the official Cardano project account was able to describe the plan as one where “the committee returns to full strength without interruption”. The handover was scheduled to be seamless by design, with no spare epoch built in if the vote missed.
That same official messaging leaned hard on urgency, warning of a governance standstill and stalled treasury and hard- decisions, and telling holders the vote “requires your attention”. The stakes were real. The framing was also a turnout campaign for a scheduled renewal of pre-elected seats, not a contested fight, and those two things are easy to blur together. One earlier official post on 26 August cited far lower support for the same action, around 43% for DReps and 15% for SPOs, and we have not reconciled that jump against the live record.
How this compares
Most -voting systems handle apathy differently. A typical DAO sets a quorum: if not enough voting power turns up, the proposal simply fails for want of participation, and the tally of yes against no is reported separately. Cardano’s SPO rule folds the two together, so non-participation shows up inside the result as opposition rather than beside it as a quorum miss. The practical effect is similar. The reported number is harder to read.
The contrast with Ethereum is sharper still, because Ethereum has no on-chain vote of this kind at all. Upgrades are agreed among client developers and the wider community off-chain, then shipped. Cardano’s approach is more accountable and more legible, and it also gives you a hard number that can be missed by 0.18 of a percentage point.
What to watch
The on-chain record, rather than any snapshot. The proposal’s entry either shows ratified and then enacted, or it does not, and that is the only version of this story that is settled. Check the governance action’s status directly on a Cardano explorer or through the Koios proposal list before treating any margin as final.
Then the turnout gap on the next few actions. DReps came in nearly two and a half points clear of their bar while stake pools scraped theirs, and if that pattern holds, the SPO threshold becomes the place where Cardano governance quietly stalls. A renewal that misses does not produce a dramatic vote against anything. It produces a committee of three, and a queue of actions with nobody empowered to check them.