01 · The puzzle
Coordination without a compass
The chapter's opening example: A and B want to dine together, but A prefers restaurant X and B prefers restaurant Y. Each tries to predict the other's choice, who is predicting theirs — an infinite regress. Both Nash reasoning and Virtual Bargaining leave multiple equilibria standing here; neither picks one. What's missing is a coordination device — and the classic candidate is "cheap talk": a costless, non-binding message sent before anyone chooses.
02 · One-way
A spear against a shield
Give only A the message. She pushes her favorite outcome — an inducing force — while B can only resist. When her push exceeds his resistance (Φ > 1), an influence exists and play lands on her preferred equilibrium: a first-mover advantage created purely by the procedure. This isn't just theory: Cooper et al. (1989) found experimentally that one-way communication produces a higher coordination rate in Battle-of-the-Sexes than two-way communication does.
03 · Two-way
Spear and shield on both sides
Now let both speak. The chapter's image: each agent holds an identical spear and an identical shield. Formally, each side's power now counts both its inducing force and its resisting force:
αA = (fA→B + rA) / (fA→B + rA + fB→A + rB)
With identical endowments, everything offsets: Φ = 1, no influence. Worse, two-way talk invites the unlimited conversation problem (Farrell, 1987; Farrell & Rabin, 1996): each side just repeats its own favorite — "I want to enter!" — and, as the chapter puts it, one thing is for sure: they will not say "You choose."
Think of two political parties debating the same issue across a senate floor, endlessly.
(None would not yield to the other side, would they?)
04 · The hidden asymmetry
Symmetric procedure ≠ symmetric power
Even a perfectly symmetric channel needn't produce symmetric power, because senders differ in intrinsic ability with words: persuasive language (Cialdini, 2021) , tactical empathy (Voss & Raz, 2016), even a well-placed malleable lie (Turmunkh et al., 2019). Poker rewards the player who knows when and how to bluff (Friedman, 1971; Kuhn, 1951) — and a blackmail victim hires a police negotiator precisely to borrow a larger skill set.
Coordination success, the chapter conjectures, rides on how skillfully the spear is used, not just on having one.
05 · Exits from deadlock
Three ways the stalemate breaks
The chapter names three.
Grant a final say: letting one side speak last is an ultimatum — mathematically equivalent to unilateral communication, and the possessor becomes superior.
End the talk first: in hostage standoffs, takers prefer to hang up first — the act and timing of terminating a conversation signals superiority (and the less the police learn, the stronger the takers' position).
Impose a deadline: a time constraint stops the conversation being unlimited at all (Karagözoğlu & Kocher, 2019; Roth, Murnighan & Schoumaker, 1988) — real negotiations always end by some clock.
06 · Leadership
Authoritarian and Democratic as channels
The chapter models briefly leadership with exactly this distinction: a leader with a one-way channel to her followers resembles authoritarian leadership; a two-way channel resembles democratic leadership.
With identical groups, it follows directly that the authoritarian leader is the more powerful — a comparison the chapter notes extends to the societal level.
Set aside, deliberately: the chapter brackets coalitions — followers banding together to stop a controlling leader — as a separate issue. And the panel's numbers are illustrative; the relationships are the chapter's.