4 ms·
I read an interesting book a while ago by Thomas Schelling [1] that said in nuclear launch systems, being unreliable is a feature not a bug. Let's say you want
by michaelt 10y ago
I read an interesting book a while ago by Thomas Schelling [1] that said in nuclear launch systems, being unreliable is a feature not a bug.
Let's say you want to threaten your enemy with nuclear war so they'll do what you want. Maybe you want them to withdraw their missiles from Cuba. The threat you make actually has two components - the claim you'll launch, and the credibility of that claim. The enemy is only scared of the /product/ of threat and credibility, so if either is zero your enemy has nothing to fear.
If you say "we'll certainly launch" you have no credibility (you'd have launched already, and everyone can see you haven't) and if you say "we certainly won't launch" you have no threat.
In fact, Schelling points out, the only credible threats are /randomized/ threats, that fall somewhere between 100% and 0% chance of launching. You threaten your enemy by showing him you have a system that has a 0.1% chance of launching per day, and you'll roll the dice every day until they withdraw their missiles from cuba (or whatever you want them to do).
Except rather than rolling dice, the random threat is from systems failures and mad generals.
This is what the DEFCON level is really about - increasing the 'readiness level' amounts to reducing the controls that prevent accidental launches, increasing the threat of a random launch.
'Doomsday machines' can be explained by a similar calculation: Nobody will believe you if you claim you'll trigger nuclear war if a single soldier took a single step over a border. So why would you for two soldiers or two steps - or indeed the capture of some trifling town near one of your allies' borders? It's not like mutually assured destruction will become a more palatable concept when the enemy tanks are ten miles over the border instead of one.
However, if you build a doomsday machine with tank detectors ten miles past the border, then you give enemy diplomats and spies a tour and show them that it works, the claim that you'll launch has credibility /despite/ the fact the ten mile line is arbitrary. The invader's calculation is no longer "If our conventional forces do X, will /they/ trigger a launch?" but instead "Do /we/ want to trigger a launch?" and the answer to that is always 'no'.
So Dr Strangelove is actually a pretty well informed film :)
[1] https://en.wikipedia.org/wiki/Thomas_Schelling#The_Strategy_of_Conflict_.281960.29 https://en.wikipedia.org/wiki/Thomas_Schelling#The_Strategy_...
- spangry 10y agoI can't remember where I got this from, but wasn't this 'instability' strategy also one that Nixon (personally) employed? As in, he deliberately behaved in a way to make the Soviet leadership think he was a bit of a madman? Also, I've always been fascinated by the Doomsday machine/credible second-strike capability doctrine, and what their ultimate consequences might be. One I find particularly interesting is this: https://en.wikipedia.org/wiki/Nuclear_peace https://en.wikipedia.org/wiki/Nuclear_peace
- twoodfin 10y agoYes, Nixon and Kissinger were believers in a "Madman Strategy": http://nsarchive.gwu.edu/nukevault/ebb517-Nixon-Kissinger-and-the-Madman-Strategy-during-Vietnam-War/ http://nsarchive.gwu.edu/nukevault/ebb517-Nixon-Kissinger-an...
- Phlarp 10y agoNixon and Kissinger derived and expanded on this from ideas originally employed by Eisenhower.
- jballanc 10y agoVery interesting idea. It certainly jives with the British tradition of Letter's of Last Resort (https://en.wikipedia.org/wiki/Letters_of_last_resort https://en.wikipedia.org/wiki/Letters_of_last_resort ; "This American Life" also did an interesting piece on these letters: https://www.thisamericanlife.org/radio-archives/episode/399/contents-unknown https://www.thisamericanlife.org/radio-archives/episode/399/... ). Essentially, the Brits have nuclear subs positioned around the world ready to launch. If, for some reason, the UK gets wiped out and one of these subs becomes disconnected from the chain of command, then a letter locked in the Captain's safe contains instructions on what to do: launch, or don't launch. One would think the very existence of these letters, and the non-zero probability that they would say "don't launch", reduces the effectiveness of the UK's nuclear deterrent under a MAD strategy, but as you point out it is actually the non-zero probability that one of these subs becomes accidentally disconnected and the letter in their safe says "launch" that makes their deterrent work in the first place. Essentially, in game theory terms you're turning a prisoner's dilemma into a Nash equilibrium, which while it sounds straight-forward enough in theory is not always so simple to achieve in practice.
- michaelt 10y agoWell, the UK is a bit of a special case - the main time a threat to launch is unconvincing is if (a) you're trying to change the status quo or (b) you're protecting an ally and it wouldn't really hurt you to cut them loose. If Britain claims they'll launch if London is attacked, that's pretty believable on its own. If Britain (say) claims they'll launch if Russia invades the Crimean Peninsula, that's kind of hard to believe. As Britain's cold war allies were also America's cold war allies, Britain didn't really /need/ to scare the soviets with the chance of a random launch as America pretty much had that taken care of. And as Britain is an island, they didn't have the 'soldiers crossed the border by accident' problem either.
- sgt101 10y agoBritain and France (and other small countries in Europe) faced a strategic issue in that "The Soviets" could escalate a nuclear war to a tactical /medium range missile level, avoid the use of strategic missiles and pulverise western europe. In the meantime the US would probably use tactical weapons to kill many soviet troops and smash up eastern europe and bits of places like Ukraine, but to strike at Moscow from Germany would have been a strategic escalation and Soviet strategists might have argued that the US wouldn't have made it. On the other hand a British PM faced with 100 or more H-bomb detonations + fall out from the 500 or more that the conflict in Europe would have required (probably closer to 800 given that the red army plan called for 450 +) probably would have. And it would have been difficult for a strategist to argue that they wouldn't, much less "The French" who were all drunk all the time as any good Soviet well knew.
- colanderman 10y agoI wonder how much of DEFCON could be applied to parenting. When kids are misbehaving, parents often respond with ultimatums -- "you have until the count of 10 to stop doing X". Of course the kid keeps doing X until just before the count has expired. Maybe parents ought to instead roll a die once per second, acting on a 1, with fewer sides to express greater discontent.
- toast0 10y agoIn my experience, counting down from three works pretty well. If it's frequent enough, you don't have to announce the rules, you just say and sign three to start. Compliance in three seconds is actually pretty good: It takes time for the child to stop a task, and gives time for the adult to plan a forceful intervention, if needed.