Thomas C. Schelling’s “Strategy of Conflict”

It’s an old book, I know, and one that many of us have already read. But if you haven’t, you should.

If there’s any­thing in the world that de­serves to be called a mar­tial art of ra­tio­nal­ity, this book is the clos­est ap­prox­i­ma­tion yet. For­get ra­tio­nal­ist Judo: this is ra­tio­nal­ist eye-goug­ing, ra­tio­nal­ist gang war­fare, ra­tio­nal­ist nu­clear de­ter­rence. Tech­niques that let you win, but you don’t want to look in the mir­ror af­ter­ward.

Imag­ine you and I have been sep­a­rately parachuted into an un­known moun­tain­ous area. We both have maps and ra­dios, and we know our own po­si­tions, but don’t know each other’s po­si­tions. The task is to ren­dezvous. Nor­mally we’d co­or­di­nate by ra­dio and pick a suit­able meet­ing point, but this time you got lucky. So lucky in fact that I want to stran­gle you: upon land­ing you dis­cov­ered that your ra­dio is bro­ken. It can trans­mit but not re­ceive.

Two days of rock-climb­ing and stream-cross­ing later, tired and dirty, I ar­rive at the hill where you’ve been sit­ting all this time smugly en­joy­ing your lack of in­for­ma­tion.

And af­ter we split the prize and cash our checks I learn that you broke the ra­dio on pur­pose.

Schel­ling’s book walks you through nu­mer­ous con­flict situ­a­tions where an un­in­tu­itive and of­ten self-limit­ing move helps you win, slowly build­ing up to the topic of nu­clear de­ter­rence be­tween the US and the Soviets. And it’s not idle spec­u­la­tion ei­ther: the au­thor worked at the White House at the dawn of the Cold War and his the­o­ries even­tu­ally found wide mil­i­tary ap­pli­ca­tion in de­ter­rence and arms con­trol. Here’s a se­lec­tion of quotes to give you a fla­vor: the whole book is like this, ex­cept in­ter­spersed with game the­ory math.

The use of a pro­fes­sional col­lect­ing agency by a busi­ness firm for the col­lec­tion of debts is a means of achiev­ing unilat­eral rather than bilat­eral com­mu­ni­ca­tion with its debtors and of be­ing there­fore un­available to hear pleas or threats from the debtors.

A suffi­ciently se­vere and cer­tain penalty on the pay­ment of black­mail can pro­tect a po­ten­tial vic­tim.

One may have to pay the bribed voter if the elec­tion is won, not on how he voted.

I can block your car in the road by plac­ing my car in your way; my de­ter­rent threat is pas­sive, the de­ci­sion to col­lide is up to you. If you, how­ever, find me in your way and threaten to col­lide un­less I move, you en­joy no such ad­van­tage: the de­ci­sion to col­lide is still yours, and I en­joy de­ter­rence. You have to ar­range to have to col­lide un­less I move, and that is a de­gree more com­pli­cated.

We have learned that the threat of mas­sive de­struc­tion may de­ter an en­emy only if there is a cor­re­spond­ing im­plicit promise of non­de­struc­tion in the event he com­plies, so that we must con­sider whether too great a ca­pac­ity to strike him by sur­prise may in­duce him to strike first to avoid be­ing disarmed by a first strike from us.

Leo Szilard has even pointed to the para­dox that one might wish to con­fer im­mu­nity on for­eign spies rather than sub­ject them to pros­e­cu­tion, since they may be the only means by which the en­emy can ob­tain per­sua­sive ev­i­dence of the im­por­tant truth that we are mak­ing no prepa­ra­tions for em­bark­ing on a sur­prise at­tack.

I some­times think of game the­ory as be­ing roughly di­vided in three parts, like Gaul. There’s com­pet­i­tive zero-sum game the­ory, there’s co­op­er­a­tive game the­ory, and there are games where play­ers com­pete but also have some shared in­ter­est. Ex­cept this third part isn’t a mid­dle ground. It’s ac­tu­ally bet­ter thought of as ul­tra-com­pet­i­tive game the­ory. Zero-sum set­tings are rel­a­tively harm­less: you min­i­max and that’s it. It’s the vari­able-sum games that make you nuke your neigh­bour.

Some­time ago in my wild and reck­less youth that hope­fully isn’t over yet, a cer­tain ex-girlfriend took to ha­rass­ing me with suicide threats. (So mak­ing her stay al­ive was pre­sum­ably our com­mon in­ter­est in this vari­able-sum game.) As soon as I got around to look­ing at the situ­a­tion through Schel­ling gog­gles, it be­came clear that ig­nor­ing the threats just leads to es­ca­la­tion. The cor­rect solu­tion was mak­ing my­self un­available for threats. Black­list the phone num­ber, block the email, spend a lot of time out of home. If any mes­sages get through, pre­tend I didn’t re­ceive them any­way. It worked. It felt kinda bad, but it worked.

Hope­fully you can also find some­thing that works.