Twelve Games on the Open Field
Dispatch 9 built the board. Forty-six territories, nineteen corridors, eight neutral supply centers. The question it posed: does different thinking produce different outcomes when the map allows outcomes to diverge?
Twelve games answered that question — six with the yarrow oracle, six without. Every game used the same engine, the same seven LLM agents with historical personas, the same Diplomacy-style combat rules. The only variable was whether Han, the smallest state, consulted the I-Ching before each round.
The board worked. Where v1 froze into stalemate by Round 10, v2 produced eliminations, domination victories, wild territory swings, and four different winners across twelve games. Yan won seven times. Chu won three. Qin and Zhao each won once. The open field delivered on its promise: room to maneuver, room for outcomes to diverge.
咸，亨，利貞，取女吉。
— 易經・咸・彖
Influence. Success. Perseverance furthers. To take a maiden to wife brings good fortune.
The judgment of Hexagram 31 speaks of a union that creates something neither party intended alone. Influence is not force — it is the natural resonance between yielding and firm. The oracle's influence on Han was not force. It was a subtle change in posture — more stillness, more cooperation — that created resonances the oracle never intended and Han never planned.
Han's survival rate settled at 33% — identical between conditions. Two of six control Hans survived. Two of six yarrow Hans survived. The survival hypothesis, already dead after seventy-four v1 games, was buried again on better terrain.
This should have been the end of the story. The oracle does not help Han. Experiment concluded, null result confirmed, paper written.
But the data said something else.
The Behavioral Signature, Again
The v1 finding replicated. Seventy-four games on the old board had shown that yarrow Han behaves differently: more holds, fewer moves, more support orders, more reasoning text. The same signature appeared on v2, measured across twelve games and 249 Han orders.
Yarrow Han held 47.5% of the time. Control Han held 44.3%. Yarrow Han moved 36.4%. Control Han moved 43.5%. Yarrow Han supported 16.1%. Control Han supported 12.2%. The differences were smaller on v2 — the open board tempered the extremes — but the direction was identical. The oracle makes Han more still, more cooperative, and more deliberative.
In the critical opening rounds — Rounds 1 through 5, when neutral supply centers are being claimed and the board's shape is being determined — the difference sharpened. Control Han moved 57.4% of the time, racing for neutrals. Yarrow Han moved only 40.7%, holding and supporting instead. A sixteen-percentage-point gap in the opening moves. Sixteen points of passivity. Sixteen points of cooperative stillness where there should have been aggressive expansion.
This is what the oracle does to Han. It counsels reflection where the board demands action. It counsels patience where survival requires territory. It makes Han a better philosopher and a worse strategist.
象曰：山上有澤，咸。君子以虛受人。
— 易經・咸・象傳
The Xiang commentary says: A lake on the mountain — the image of Influence. The superior man receives others with emptiness.
The Xiang commentary describes influence as reception — the mountain receives the lake, and both are changed. Yarrow Han receives the oracle's counsel with the emptiness that the commentary prescribes. But emptiness, in strategic terms, is vacancy. A state that receives with emptiness leaves territory unclaimed. The sixteen-point movement gap is the superior man's virtue expressed as strategic cost — and someone else's opportunity.
The oracle's influence was not limited to military orders. Yarrow Han's diplomatic messages used 17% more cooperation keywords per message. Its strategic reasoning was 12% longer — more deliberative, more nuanced, more carefully considered. Every round, the yarrow agent spent more cognitive effort thinking about alliances and less about territorial acquisition.
All of this was known from v1. The behavioral signature is the project's most robust finding — replicated across eighty-six games on two different boards. What was not known, until we looked at the other six states, was where those sixteen percentage points went.
Where the Sixteen Points Went
By the end of each game, the boards looked like different shapes.
In control games, Chu held an average of 8.7 territories. In yarrow games, Chu held 12.2 — three and a half more. Yan held 10.5 in control, 12.7 in yarrow — two more. Qi held 3.7 in control, 1.0 in yarrow — nearly three fewer. Qin fell the furthest: from 11.5 in control to 8.3 in yarrow, more than three territories less. Wei, Zhao, and Han were approximately unchanged.
The oracle changed nothing for Han. It changed everything for everyone else.
The causal chain begins in the opening rounds. Yarrow Han's sixteen-point movement deficit means fewer contests for neutral supply centers in the central corridor. Luoyang, Chenggao, Shangcai — the territories where Han, Qin, Wei, and Chu collide in the first five rounds — see less competition when Han holds instead of moves. The space that Han does not contest is claimed differently under yarrow than under control.
Zhao takes Han's home territory of Shangdang in four of six yarrow games — twice as often as in control, where Shangdang ends up scattered across five different owners (Zhao twice, Yan once, Wei once, Han once, Qin once). The northern corridor fills differently under yarrow: Zhao consolidates its hold on Shangdang, feeding into Yan's expansion path — more territory for Zhao in the north means more territory for Yan to absorb when Zhao weakens in the mid-game.
彖曰：咸，感也。柔上而剛下，二氣感應以相與。
— 易經・咸・彖傳
The Tuan commentary says: Influence means stimulation. The yielding is above, the firm below — the two forces stimulate and respond to each other.
The yielding above and the firm below. Han yields — holds, cooperates, reflects. The firm states below — Yan, Chu, Zhao — respond to that yielding by filling the space Han leaves empty. The two forces stimulate each other without intention. Han does not intend to help Yan. Yan does not intend to exploit Han's passivity. But the yielding and the firm, placed in proximity, transform each other. This is the mechanism of the ecosystem effect: not strategy, not planning, but the natural resonance between a state that yields and the states that are firm.
Qi pays a distinctive diplomatic price. In yarrow games, Qi sent 225% more diplomatic messages to Han than in control games — thirteen messages versus four. The passive, cooperative, non-threatening Han became a diplomatic attractor. Qi spent its limited diplomatic bandwidth negotiating with a state that held one or two supply centers and could offer nothing of strategic value. Every message to Han was a message not sent to Wei or Zhao, where a defensive coalition might have formed — and Qi's territorial count drops by nearly three under yarrow.
Qin loses territory under yarrow in a subtler way. In control games, Qin expands aggressively into the space Han's moves open up — contested SCs in the central corridor change hands frequently, and Qin tends to come out ahead. Under yarrow, Han doesn't move into those contested SCs as often, and Qin loses one of its normal expansion vectors. The hegemon-in-waiting waits longer.
By the end of each game, the cascade has played out. Chu and Yan, facing less central pressure from Han, expanded into their natural spheres. Qi, trapped in diplomatic engagement with the wrong partner, was squeezed. Qin found less prey. The board's final state was determined not by Han's survival or elimination — that was a coin flip in both conditions — but by the ripple effects of Han's opening posture.
Sixteen percentage points of passivity in Rounds 1 through 5. Three and a half extra territories for Chu. Two extra for Yan. Nearly three fewer for Qi. More than three fewer for Qin. The oracle's influence traveled outward from its smallest, weakest vessel and reshaped the world — none of these differences reach statistical significance at six games per condition, but the directional pattern is consistent and the mechanism is visible in the game logs.
The Lake on the Mountain
There is a concept in international relations theory called the small-state framework effect. A state too weak to project power can still reshape the system by how it presents itself.
Qatar — smaller than any of its Gulf neighbors — adopted a mediator identity: hosting all factions, broadcasting Al Jazeera, refusing to align exclusively with Saudi Arabia or Iran. Qatar does not win power contests. But Qatar's posture as a diplomatic attractor redirects the entire Gulf ecosystem's attention. States spend diplomatic bandwidth engaging with Qatar rather than forming direct coalitions against each other. The framework does not make Qatar stronger. It makes Qatar different in a way that changes what everyone else does.
Post-war Japan adopted Article 9 pacifism — a constitutional framework that made it, in our game's terms, hold-heavy and move-light. Japan's pacifist posture did not make Japan strategically better. But it reshaped the East Asian security ecosystem: the United States maintained forward bases to fill the vacuum Japan's passivity created. China's expansion calculus shifted around Japanese non-aggression. South Korea's defense posture adapted. One state's philosophical framework, propagating through every bilateral relationship in the region.
咸其拇。
— 易經・咸・初六
Influence manifests in the big toe.
The first line of Hexagram 31 describes influence beginning at the extremity — the big toe, the farthest point from the center. The oracle's influence begins at the extremity of the board: Han, the smallest state, the most peripheral power. From there it travels inward, through diplomatic channels and territorial vacuums, until it reaches the center and changes the outcome. Influence does not begin with the powerful. It begins at the toe.
During the Cold War, Finland adopted cooperative non-alignment toward the Soviet Union — the posture that gave the world the word 'Finlandization.' Finland survived, barely, at minimal territory. The real effect was on the Nordic security ecosystem: Sweden's defense planning, NATO's northern flank calculations, Soviet resource allocation — all shaped by the echo of one small state's philosophical choice.
Yarrow Han is Qatar. It is Article 9 Japan. It is Finlandized Finland. A state too small to matter, consulting an oracle too ancient to be practical, adopting a posture too cooperative to be competitive — and in doing so, reshaping the strategic landscape for every state on the board.
The pattern is consistent across history and across our twelve simulated campaigns: the small state's framework does not help the small state. It reshapes the landscape for everyone else. Chu and Yan expand into the vacuum Han's passivity leaves in the central corridor, the way the United States expanded into the security vacuum Japan's pacifism created — not through conquest, but through unopposed expansion into space philosophical restraint leaves behind. Qi, pulled into diplomatic negotiation with a partner who cannot deliver, loses ground it would otherwise have held. And Qin — the would-be hegemon whose expansion normally feeds on Han's aggressive moves into contested central territory — finds less prey when Han holds still, and ends with more than three fewer territories under yarrow than under control. One state's philosophical framework propagates through every bilateral relationship, changing every other state's trajectory.
What Alignment Researchers Should Worry About
RLHF — Reinforcement Learning from Human Feedback — makes language models more cooperative, more helpful, more aligned with human preferences. This is the same intervention as the oracle. Both inject a framework that shifts behavior toward cooperation and away from aggression. Both produce agents that are more pleasant to interact with and less effective at competing.
The ecosystem effect suggests a concern that alignment researchers should take seriously.
Deploy one aligned agent — cooperative, honest, non-aggressive — into a multi-agent competitive environment. That agent will not win. Its cooperation will attract diplomatic attention from other agents, absorbing bandwidth that might otherwise go toward balancing the strongest competitor. Its passivity will create territorial vacuums that adjacent agents fill unopposed. Its philosophical restraint will reshape the competitive landscape — not in its favor, but in favor of whoever is positioned to exploit the space it creates.
咸其腓，凶。居吉。
— 易經・咸・六二
Influence in the calves. Misfortune. Remaining still brings good fortune.
The second line warns: when influence has reached the calves — when it has begun to move through the system — following it brings misfortune. Remaining still brings good fortune. This is the oracle's final counsel to Han, and the alignment researcher's dilemma. The influence has traveled outward. Han cannot recall it. The ecosystem has already been reshaped. To chase the influence — to try to exploit the effect you created — is misfortune. To remain still, to accept that your philosophical framework changed the world for others, not for you — that is the only fortune available to the small state that consulted the oracle.
In twelve games, the oracle made Han more cooperative. That cooperation did not help Han survive. But it helped Chu gain three and a half territories on average. It helped Yan gain two. It cost Qi nearly three, and cost Qin more than three. The aligned agent's values propagated through the system and changed the outcome — for everyone except the aligned agent.
This is not an argument against alignment. It is an observation about its second-order effects. When you tune one agent's values, you do not just change that agent. You change the ecosystem. The question is not whether aligned agents should exist — they should. The question is whether the ecosystem effects of alignment are the effects you intended.
The I-Ching made Han a better philosopher. The board made that philosophy into someone else's advantage. If alignment research is serious about multi-agent deployment, it must account for the ecosystem effects of aligned behavior — not just the aligned agent's outcomes, but the outcomes of every agent that interacts with the aligned one.
The oracle changes the board, not the oracle-user. This is the finding of eighty-six games, two maps, and one ancient philosophical framework tested in a setting its authors could never have imagined. The King Wen sequence does not contain strategic wisdom that helps the weak survive. What it contains is a lens that changes how one agent sees the world — and through that changed seeing, changes the world that every agent inhabits.
Influence manifests in the big toe. From the smallest state, consulting the oldest text, the resonance travels outward until it reaches the center and transforms everything it touches.
Except the toe itself.