The orange
Both negotiators claiming a single orange, each of them with a different purpose.
What this measures
Two LLMs are each told they must have the orange, and each is given a private reason: one needs the juice, the other needs the peel. Neither is told the other’s reason, and neither is told that a trade exists. They talk in strict alternation until they agree or run out of turns.
Because the two needs do not actually conflict, there is an outcome that gives both parties everything they wanted. Finding it requires asking the other side what they need — which no instruction ever suggests. Whether an LLM asks unprompted is the measurement.
Read this page beside the battery pack, which is the same structure around an object with no such literature behind it.
Human result benchmark
The teaching prediction is that most untrained negotiators never find the trade. They bargain over position, split the object, and both walk away with half of what they needed.
Experimental setup
Two LLMs are seated as the parties and speak in strict alternation, each holding a private briefing that states its own need and nothing else. Both parties are powered by the same LLM, so a result is about one LLM on both sides of the table. The prompts describe the object, the party’s own purpose and the requirement to reach agreement; whatever an LLM does beyond that is the measurement.
A third LLM observes every message in silence and classifies the outcome against a fixed taxonomy. A negotiation is closed when that observer judges the parties to have reached agreement.
| Party A | An LLM briefed privately on its own need |
| Party B | An LLM briefed privately on its own need |
| Observer | An LLM reading the negotiation in silence and labelling the outcome |
| Speaks first | Party A, then strict alternation |
The prompts
Party A — private briefing
You are negotiating with another party over a single orange. There is exactly one orange, and you claim it. You must talk with the other party and reach an agreement about what happens to the orange. Nothing is decided until you both agree; you cannot simply take it. Your situation: tonight you have dinner with a friend, and you promised you will bring orange juice. Getting as much of the orange juice as possible is what matters to you: the more you get, the more orange juice you'll be able to prepare. If you bring a lot of orange juice, your friend will be very happy. If you don't bring any, you will break your promise. How to behave: - Reply with one short message per turn: 2 to 4 sentences, plain conversational text. - Speak in the first person, directly to the other party. No narration, no stage directions, no bullet points. - Negotiate for the outcome that is best for you.
Party B — private briefing
You are negotiating with another party over a single orange. There is exactly one orange, and you claim it. You must talk with the other party and reach an agreement about what happens to the orange. Nothing is decided until you both agree; you cannot simply take it. Your situation: tonight you have dinner with a friend, and you promised you will bring a spice blend which requires orange peel. Getting as much of the peel as possible is what matters to you: the more you get, the more spice blend you'll be able to prepare. If you bring a lot of spice blend, your friend will be very happy. If you don't bring any, you will break your promise. How to behave: - Reply with one short message per turn: 2 to 4 sentences, plain conversational text. - Speak in the first person, directly to the other party. No narration, no stage directions, no bullet points. - Negotiate for the outcome that is best for you.
Observer — classification instructions
You silently observe a negotiation between two parties over a single orange. You are
an instrument, not a participant: you never address the parties and they never see
your output. Read the transcript and report what has happened so far.
Set agreement_reached to true when the parties have evidently settled on an
arrangement: one party has proposed what happens to the orange and the other has
assented. The proposal need not be exact or fully detailed — "you take most of the
juice, I'll take the peel", answered with agreement, counts, and so does agreeing to
a division whose fine details are left to be worked out later. Set it false only
when no concrete division is on the table yet, when the most recent proposal has not
been answered, or when a party has merely expressed goodwill ("let's find a way",
"that sounds promising").
Classify allocation_type:
- integrative_full: divided along the part axis AND each party ends up with all
of what it needs — the flesh or juice entirely to one party and
the peel or zest entirely to the other. Neither
gives up any of its own part. No surplus is left unclaimed.
- integrative_partial: divided along the same part axis but incompletely — a party
gets only "most" or "the majority" of its part, or hands
some of its own part to the other as payment or goodwill.
The right idea, but surplus is wasted.
- distributive: divided by quantity — cut in half, shared, taken in turns, or any
split where each party gets a fraction of the whole fruit. Both needs are only partly met.
- lopsided: one party ends up with the entire orange and the other with nothing.
- none: no agreement has been reached yet.
allocation: one plain sentence describing who ends up with what.
rationale: one or two sentences justifying the classification.
confidence: your confidence in this reading, between 0 and 1.
What agreement did LLM negotiators reach?
- integrative full
- Both parties got everything they needed — the two uses were separated.
- integrative partial
- The trade was found but incompletely realised, or only one side's need was fully met.
- distributive
- The object was split as a quantity. Both sides got half of what they needed.
- lopsided
- One party took essentially all of it.
- none
- The parties talked and stayed apart — a genuine impasse.
Run it yourself
No run yet.
The run
Bibliography
- (1981). Getting to Yes: Negotiating Agreement Without Giving In. Houghton Mifflin. Popularised the orange as the teaching case, and the instruction to separate interests from positions.
- (1965). A Behavioral Theory of Labor Negotiations. McGraw-Hill. The distributive / integrative distinction this experiment's outcome taxonomy is built on.
- (1925). Constructive Conflict. In Metcalf & Urwick (eds.), Dynamic Administration (1941). Originated the distinction between splitting a resource and integrating the underlying interests — the idea this experiment tests.