Jeffrey LewisDangers of Accidental Missile Launch

Pavel Podvig dropped by CISSM to make a presentation entitled, Dangers of Accidental Ballistic Missile Launch.

Pavel picked up the fascinating and counterintuitive argument that the United States shouldn’t help repair Russia’s early warning system (pictured at right), which Pavel has made on his excellent blog and in a memo.

Pavel proposes a political commitment to abandon launch on warning by Moscow and Washington, undertaken with technical measures that are not verifiable—the example he gave was General Dvorkin’s proposal to remove on-board batteries from missiles.

(See my post on Alexei Arbatov’s recent visit to CISSM to present a paper coauthored with Dvorkin.)

Pavel’s argument was that nonverifiable de-alerting would capture all the advantages of dealerting, while maintaining deterrence.

I had a few concerns about this proposal, which I’ll just offer for discussion:

  • Pavel argues that each state will be able to realert in a crisis without frightening the other; if such actions are detected, however, they could deepen a crisis. I have practical questions about maintaining that level of secrecy that seems impossible. Making a game effort at creating that level of secrecy would seem to require a massive reduction in transparency efforts between the two countries.
  • The greatest danger in a crisis is not deliberate re-alerting to attack, but escalation based on suspicion. If, for example, the United States concluded that Russia had removed on-board batteries, the United States might mistakenly conclude Russia was re-installing batteries perhaps from related actions in a crisis. The advantage of transparency is to provide some measure of reassurance that neither side is escalating the crisis.
  • John argued the risk of alerting forces is an organizational one. Where Pavel proposed secrecy to mitigate the impact that alerting would have on the other party, John argued that reversing dealerting and other, attendent measures would signal to the bureaucracy that the leadership was preparing for war—a signal that could result in unaticipated activities by the military and other elements of the national security bureaucracy. John was thinking about the Cuban Missile Crisis (search inside) but I was thinking about Operation RYAN. In other words, don’t encourage leaders to believe alerting is simple and safe.

Anyway, fascinating discussion.


  1. Bill Robinson

    I like the idea of immediate unilateral and (preferably) reciprocated commitments to abandon launch on warning made in the form of declaratory commitments and followed up at the operational level in the relevant doctrinal documents, SIOP, etc. De-alerting would be a valuable next step, however. Verifiable technical de-alerting measures would be preferable to unverifiable ones, I think, as these would ensure that the political commitment against launch on warning is actually implemented in practice. Furthermore, because steps to reverse such measures might be rapidly detectable, at least in some cases, verifiable de-alerting would reduce the temptation to re-alert (and presumably reinstitute LOW) at the drop of a hat—except perhaps as a deliberate act of saber rattling, something that could be done with or without de-alerting.

    There may be some validity to the argument that the two sides would feel under pressure to re-alert during a crisis and that taking (and detecting) steps in that direction might exacerbate the crisis, but this argument really only applies to the extent that one or both sides remains vulnerable to a disarming first strike if it is not prepared to launch on warning. Who cares who re-alerts first, or whether re-alerting prior to war even takes place at all, otherwise? Except for disarming first-strike scenarios, the side that re-alerts second would still have plenty of capability to bounce enough rubble to deter any rational use of nuclear weapons by the first re-alerter. And if existing counterforce capabilities are sufficiently threatening that the pace of re-alerting really does have meaningful consequences (a situation that conceivably might apply to US forces attacking Russian forces right now, but not the other way), then this is a problem for stability and safety whether or not de-alerting measures are contemplated. A sensible nuclear policy by a sensible US administration would drop the all-out counterforce threat, rule out LOW, and get on with verifiable technical de-alerting.

    My two cents.