10 ideas
Brute force takes what it wants and needs no cooperation from the enemy. Coercion instead exploits the latent capacity to inflict pain, and it works best when that pain is held in reserve. The power to hurt is bargaining power: its value lies in the damage not yet done, which the adversary can avoid by complying.
Deterrence threatens punishment if the adversary acts, then waits, so the burden of action falls on the other side. Compellence demands that the adversary do something, usually by starting a pain that stops only when they comply. Compellence is harder because it needs a deadline and a visible act of submission, and it makes compliance look like humiliation.
A threat you would never deliberately carry out can still be credible if you create a real risk that events slip out of control. You don't threaten to go to war. You threaten to stay in a situation where war might happen whether or not anyone chooses it. Brinkmanship works by sharing a risk neither side fully controls.
You can make a threat credible by removing your own ability to back down, so the only remaining choice is the one you threatened. Troops stationed as a tripwire, public pledges and automatic mechanisms all shift the last clear chance to avoid disaster onto the opponent. Being weaker in choices becomes being stronger in bargaining.
Schelling argues that how a state handles one commitment can shape what adversaries expect of it elsewhere, since they may read a concession as a sign of its future resolve. For that reason a state may judge it worth defending a stake that is small in itself, because part of what is at stake is its reputation for keeping promises.
Adversaries who cannot or will not negotiate explicitly still coordinate on limits in war. These include rivers, national borders, the distinction between nuclear and conventional weapons, and no gas. Such limits hold because they are qualitative, obvious and salient, not because they are fair. A clean line is stable, while a limit set by degree invites erosion because there is no natural stopping point.
Even during combat, violence functions as communication. Each action signals intent, resolve and what the adversary could still lose. Seen this way, the question in war shifts from who wins on the battlefield to what each side is trying to get the other to do, and how the pain being inflicted, or held back, conveys that.
Before nuclear weapons, a country usually had to defeat an enemy's army before it could hurt the enemy's population. Nuclear weapons let a state inflict catastrophic civilian pain without first winning militarily. This turns strategy from a contest of strength into a contest of nerve, risk-taking and endurance.
When striking first confers an advantage, each side fears that the other will strike. It may then attack preemptively because it expects to be attacked, even though neither side wants war. The danger compounds as expectations feed on each other. Stability therefore requires forces that can survive a first strike, so that neither side gains by hitting first.
Coercion only works if the threat to punish noncompliance comes with a credible promise not to punish compliance. If the adversary expects to be hurt anyway, it has no incentive to yield. Effective threats must therefore leave the target a clear, acceptable way out and a believable assurance of restraint.