What's called nuclear weapons policy is scary and complex, but bear with me: the situation can be much safer than it is now and has been for decades. A single change could reduce the danger of war, increase security, and restore deliberation.
Under the old Eisenhower policy of "massive retaliation" and its successors, the U.S. has been planning to "launch on warning," whether the data come from radar stations, satellites, or other sources. Since intercontinental missiles based in Russia take around a half hour to "deliver" hydrogen bombs, and detection time of a missile launch is sometimes even shorter, and there must be time for analysis before a decision and for firing procedures afterwards, the President, possibly woken from sleep, would have no more than a few minutes to consider the evidence and possibly order an attack.
But why launch on warning? What if the warning if false, as has happened repeatedly and so far been discovered at the last moment?
We have been ensorcelled by a false analogy. It's natural to keep thinking in familiar terms even as technology changes radically. For example, we have "desktops" on the computer screen. In a similar way, we speak of fusion devices as "weapons." Hydrogen bombs can be orders of magnitude more powerful than the explosions at Hiroshima and Nagasaki, just as the devices of 1945 were orders of magnitude more powerful than each of the bombs dropped on London or Hamburg. Nuclear devices are not "weapons" but, in a phrase borrowed from the movie "Dr. Strangelove," "doomsday machines." "Exchanging" them would constitute not war but mass slaughter on an unprecedented scale. As Einstein said, everything has changed except our modes of thinking.
The perceived need to launch U.S. missiles before some of them are wrecked by incoming enemy fusion devices has meant that the decision to fire is restricted to one person, the "commander in chief." In the U.S. governing system of checks and balances, no single person should have this authority, no matter how wise, well-informed, and emotionally mature.
Is it possible to design a system of "minimal sufficient deterrence" based on devices known to be survivable after an enemy attack had been completed? If so, we might think less in terms of weapons that might be destroyed by an attack, more in terms of weapons that are elusive and thus survivable.
Under a doctrine of "minimal sufficient deterrence," everyone in the executive branch would know that any order to use nuclear weapons would be illegal and thus not to be followed unless specifically and newly authorized by Congress (including official alternatives outside the capitol).
One kind of elusive weapon is a nuclear submarine, such as we already have. Another is bombers that are at fail-safe points. A third is mobile missiles on land, the location of which is not known to the enemy. There are challenges with any of these systems, but the challenges are smaller than the risk of rash nuclear attacks.
It is a natural human impulse, not restricted to some in the military, to want to use "weapons" before some of them are destroyed. Tremendous ingenuity has gone into warning systems to make this possible. Nonetheless, there have been false warnings. It does not matter whether a false warning is caused by sunlight glinting off clouds in the Midwest, a flock of geese, a weather balloon launched by Norway, the paranoia of an enemy leader, or a training tape accidentally mistaken for an actual attack.
If nuclear devices are survivable, perhaps by elusiveness, then a much smaller number would be necessary to assure retaliation and thus to deter an attack. Could such a sharp reduction be negotiated? That would have several advantages: (a) the system would be cheaper, (b) false warnings wouldn't be acted upon, rendering the system much more stable, (c) if nuclear devices were ever "exchanged," the severity of "nuclear winter" would be minimized, (d) the need for a quick decision would be eliminated and Congress could reassert its right, under the Constitution, to declare war instead of only to review, after 60 days, what the executive had done under the defective and outmoded War Powers Act of 1973.
Why not ask people of good will, without an economic or ideological interest in the present system, to meet the challenge of designing a system of minimal sufficient deterrence?
One problem is known in the literature as "decapitation," or the killing of officials with the authority to order retaliation. The Soviets have tried to deal with this by a retaliatory system with the colloquial name "the dead hand." In the U.S., going back many decades, the President has delegated the power to order a nuclear attack, a practice discovered when the Kennedy administration was young.
Another challenge is that leaders have "used" nuclear devices not by "delivering" them, but by threatening to do so. Under a policy of "minimal sufficient deterrence," and a further extensive diminution of the stockpiles in both the U.S. and Russia, these threats would become even less credible than they have been.
.A third challenge is how to create a rigorous system of inspection without reducing the elusiveness of the nuclear devices that would be kept.
A policy of "mninimal sufficient deterrence" would be a compromise between the present system of massive overkill and total nuclear disarmament, as advocated by Jonathan Schell in 1983 and by others, and discussed between Gorbachev and Regan in their Reykjavik summit in 1986.