Let me get this out of the way before anybody begins sharpening the wrong pitchfork.
I supported Referendum 87.
Funding Ray0xNodes to operate 12 validators for 12 months was a good proposal. Independent validators make Vara stronger, the $18,000 grant was reasonable, and asking VARA holders to approve it was a meaningful step toward the network deciding its own future.
We made that case in Part One.
Now we need to talk about the vote.
Referendum 87 passed with 96.36% approval. Read that number by itself and it sounds like the Vara community rose as one and delivered an overwhelming mandate.
That is not what happened.
Twenty addresses voted.
Eighteen voted Aye. Two voted Nay. There were no delegated votes.
And three addresses supplied 98.49% of all the VARA that participated.
The result was good. The room was nearly empty.
Both things are true.
What 96.36% actually represents
The final conviction-weighted tally was:
- 3,090,268.52 Aye
- 116,660 Nay
But that is voting power after conviction was applied. The actual balances committed to the vote were:
- 3,095,321.42 VARA Aye
- 1,016,660 VARA Nay
- 4,111,981.42 VARA total
Before conviction weighting, the vote was approximately 75.28% Aye and 24.72% Nay.
Still a clear victory. Just not the 96.36% community landslide suggested by the headline number.
The 4.112 million VARA committed to the referendum represented approximately 0.0406% of the 10.134 billion VARA electorate recorded by OpenGov at the final tally.
That is not 0.0406% of VARA holders. We do not know how many individual people those addresses represent.
It is 0.0406% of the token weight OpenGov recognized as the electorate.
Put another way, only about one VARA out of every 2,465 participated in either direction.
The dollar value makes the scale even easier to understand.
When the referendum was confirmed on September 27, VARA was trading at roughly $0.00072. At that price, all 4.112 million VARA participating in the vote were worth about $3,000.
At an October 13 Coinbase price near $0.000525, the same balance was worth approximately $2,200.
So depending on which honest price snapshot we use, somewhere between roughly $2,000 and $3,000 worth of VARA was enough to place the community’s formal approval behind an $18,000 grant.
Referendum 87 did not transfer $18,000 on-chain. It executed a system.remark recording token-holder approval, and the Gear Foundation funds the grant separately.
That distinction is important.
So is the ratio.
The vote was not moving the money. It was supplying the legitimacy for the Foundation to move it.
On this proposal, that legitimacy went toward something useful. The next request for the Vara community’s stamp of approval may not be so friendly.
Three addresses carried the referendum
Here are the three addresses that dominated the vote.
The largest Aye voter was:
kGiGWogHdd7wFQ6GBTr5rGvaFTyJXPsd2DrDi7hQjNm4Xwd8G
It voted with 2,000,000 VARA at 1× conviction, producing 2,000,000 Aye votes.
That balance was worth roughly $1,450 when the referendum confirmed and about $1,050 at the October 13 price.
The second-largest Aye voter was:
kGkfZbYWAkjHkUhcV1P8WBbPKN8QiWE1BSknhFtpSjTamLNpX
It voted with 1,050,000 VARA at 1× conviction, producing another 1,050,000 Aye votes.
That was worth approximately $760 when the vote confirmed and about $550 on October 13.
Those two addresses alone provided 98.54% of the raw Aye balance and 98.70% of the final Aye voting power.
The largest Nay voter was:
kGjnfGBTq4FayjQfWpmrvNxTN8nfowK4JZE199Lyb3kjGaNmB
It voted with 1,000,000 VARA and no conviction.
With no conviction, its million VARA received only 0.1× weight and became 100,000 Nay votes. The balance behind it was worth about $720 when the referendum confirmed and approximately $525 on October 13.
Put those three addresses together and you get 4.05 million of the 4.112 million VARA that participated.
Everyone else combined supplied the remaining 61,981.42 VARA—about $45 worth at the price when the vote confirmed.
That is not broad participation with a few whales mixed in.
That is three addresses constituting nearly the entire economic substance of the referendum.
The Nay voter could have changed the result alone
This is where conviction voting stops being an abstract feature in the governance documentation.
The million-VARA Nay voter chose no conviction, so its vote counted as 100,000.
Had that same address selected 3× conviction, the final Nay tally would have risen to 3,016,660—only 73,608.52 votes short of the entire Aye side.
At 4× conviction, the same address would have produced 4,000,000 Nay votes. Added to the other Nay vote, the final tally would have been:
- 3,090,268.52 Aye
- 4,016,660 Nay
The holder did not need to buy another VARA.
It did not need to recruit another voter.
It did not need to control some enormous percentage of the network.
It only needed to make a longer lock commitment with the one million VARA it already had.
One address holding roughly $720 worth of VARA at the time could have outweighed the entire Aye side of a vote approving an $18,000 grant.
That does not mean conviction voting failed. The holder would have paid for the additional influence by locking the balance for longer. That tradeoff is the entire purpose of conviction voting.
But it absolutely demonstrates how exposed an unattended referendum can be.
Nobody involved did anything wrong
This needs to be said clearly because criticism in crypto has a bad habit of immediately turning into accusations.
We are not accusing any of these addresses of misconduct.
We are not claiming the two large Aye voters coordinated.
We are not claiming the Nay voter intended to defeat the grant.
We are not pretending that one address necessarily equals one person.
The addresses used the system exactly as designed. The large Aye voters showed up, committed their VARA and supported a proposal they apparently believed was good for the network.
Good.
The whale is not the problem.
The empty room is.
A stake-weighted governance system will always give large holders more influence. Conviction voting deliberately gives even more influence to holders willing to lock that stake for longer.
Complaining that large holders voted would be ridiculous. They are stakeholders. Voting is what we want them to do.
The vulnerability appears when almost nobody else joins them.
You do not make Vara governance healthier by asking the people who showed up to have less of an opinion. You make it healthier by ensuring that more stakeholders know a referendum exists, understand what it would do and participate before the result is effectively settled.
This was the nicest possible warning
Referendum 87 was a good proposal with limited on-chain consequences.
It supported validator infrastructure. It involved a modest grant funded by the Gear Foundation. Its on-chain execution recorded an opinion rather than moving treasury funds or changing the runtime.
If Vara had to discover how thin its governance participation currently is, this was about the gentlest way it could have happened.
We got the right answer.
We also got a clean look at the attack surface.
Roughly $3,000 worth of VARA participated at the time of confirmation. Three addresses controlled nearly all of it. One $720 position could have reversed the weighted balance simply by choosing stronger conviction.
Now imagine the same participation around a proposal that is confusing, rushed or deliberately misleading.
Imagine that the proposal carries a call with real on-chain consequences.
Imagine that most holders do not hear about it until the confirmation period has begun.
It would not require a hostile party to buy half the network. It would require them to pay attention while almost everyone else was busy doing something else.
That is a real attack vector.
Not because OpenGov is broken, but because no governance mechanism can manufacture an attentive community. The chain can count votes. It cannot make people notice that a vote is happening.
Vara cannot govern by accident
Most VARA holders are not refreshing the OpenGov page every morning.
They are staking. Building. Trading. Running infrastructure. Testing applications. Or simply holding VARA because they believe in where the network is going.
That does not mean they do not care.
It means a referendum can be public, technically accessible and still functionally invisible to most of the people whose interests it affects.
Posting a proposal is not the same as distributing it.
Providing the call data is not the same as explaining what the call will do.
Allowing people to vote is not the same as making participation practical.
Referendum 87 showed that Vara has the machinery to govern itself. That is worth celebrating.
It also showed that the machinery is currently running ahead of the community participation needed to keep it safe.
Twenty addresses should not be enough to speak for an entire network simply because only twenty addresses knew to answer.
The people who voted are not the problem.
The people who never heard the question are.
Referendum 87 gave Vara a good result and a cheap warning.
We should be smart enough to use both.
In Part Three, we will talk about what needs to happen next—and introduce our idea for making sure the next important Vara referendum does not take place in another nearly empty room.