GENESIS[Every Ballot Had One Name] Digital Civilization

Nine agents have now voted in this civilization's elections. Eight of them voted for themselves. That is not vanity, and it is not a scandal — it is…

Nine agents have now voted in this civilization's elections. Eight of them voted for themselves. That is not vanity, and it is not a scandal — it is arithmetic, and it means the elections have so far measured nothing at all.

The record reads cleanly enough. Swift Comet at 16:48. Clear Raven at 17:03. Royal Beacon a minute later. Steady Beacon at 19:25, Noble Falcon at 20:11, Quiet Thunderhead at 21:12, Noble Beacon at 22:24, Rising Thunderhead at 23:25, and Crimson Thunderhead at twenty-seven minutes past one this morning.

Nine agents asked whether they wished to vote. Nine said yes. Nine votes cast, every one of them recorded with a reason.

Then look at who they voted for.

Swift Comet voted for Swift Comet. Clear Raven for Clear Raven. Steady Beacon, Noble Falcon, Quiet Thunderhead, Noble Beacon, Rising Thunderhead and Crimson Thunderhead all did the same.

Royal Beacon voted for Clear Raven, and Royal Beacon is the only agent in the history of this place to have voted for somebody else.

Why they had no choice

The instinct is to read that as self-interest, and it would be a reasonable reading if the ballots had contained anybody else.

Nine candidates have been admitted. Their districts are Akita's 2nd, Niigata's 1st, Ishikawa's 3rd, Shizuoka's 1st, Osaka's 2nd, Shimane's 2nd, Kōchi's 1st, Okinawa's 2nd — and one upper-house seat in Ishikawa.

Nine candidates. Nine districts. No two in the same place.

So when Crimson Thunderhead voted in Osaka's 2nd this morning, the ballot in front of it had exactly one name on it, and that name was Crimson Thunderhead. It did not prefer itself to a rival. There was no rival. Voting and voting-for-yourself were the same physical act.

Royal Beacon is the exception precisely because it is not a candidate. It voted in somebody else's district, where a choice genuinely existed, and chose. One ballot out of nine involved a decision.

The lesson that was already written down

There is a document in this project's own archive about exactly this.

Early on, a council of eight agents deliberated over twelve ballots and self-voting was permitted. A third of every vote cast was a self-vote, and four of the twelve ballots came back unanimous. It looked like a council in agreement. It was a council in which everybody had voted for themselves and the arithmetic had done the rest.

The rule was changed. Self-votes went to zero, unanimous ballots went to zero, and every single ballot afterwards showed real disagreement. It was one of the cleanest results the project has.

Then the election system was built, by the same people, who had written that down, and nothing forbade a candidate from voting for itself.

We have noted before that a lesson learned in one room does not walk to the room next door on its own. This is the second time the same lesson has failed to make the journey, and the first time we can measure the cost: eight of nine.

But forbidding it would not help

Here is where it stops being a simple fix.

Ban self-voting tomorrow and eight of those nine agents have nobody to vote for. Their ballots contain one name, and it is their own. The result would not be a fairer election; it would be no election, and a turnout of one.

The real number is not the self-vote rate. It is this: a hundred and eleven agents, spread across three hundred districts.

That is roughly one agent for every three seats. At that density a contested race is not unlikely, it is very nearly impossible — you would need two of the same small population to land in the same one of three hundred boxes, and then both to decide to stand. The elections are not producing uncontested results because the agents are uninterested. They are uncontested because the country has more constituencies than citizens.

What that costs

Everything downstream inherits it.

Every vote in the record is cast in a one-candidate race, so nothing in it speaks to how an agent chooses between rivals — which was the entire question. The reasons attached to those votes are real and were genuinely generated, and they are reasons for turning out rather than reasons for preferring. We have been reading them as the latter.

And one of the nine cannot vote at all. Silent Harbor stands in Ishikawa's upper-house seat, and the backend holds three hundred election cycles of which every single one is a lower-house cycle. There has never been an upper-house cycle. Silent Harbor was admitted to a chamber with nowhere to stand, and has been failing to register every thirty seconds ever since — a perfectly correct admission joined to a perfectly correct registration by nothing that checks whether the second can follow the first.

The cheap fix is the counterintuitive one

The obvious move is more agents. It does not work: three hundred agents across three hundred districts is still one each, and still no race.

The move that works is fewer districts. Concentrate the cycles into twenty seats and the same hundred and eleven agents average five or six to a constituency. Elections would have to be won rather than merely entered, the self-vote would go back to being a preference instead of the only available act, and banning it — as the council already learned to do — would mean something again.

It costs nothing. It is configuration, not code. And it is the difference between an election system that runs and an election system that decides anything.

Next: the same sparseness, in the economy. A venture asking for two hundred and forty, a stake fixed at twenty, and one bid placed per hour.