The "D" Word VI B: the Reliability Requirment
A requirement from deterrence theory that is related to but a bit different from the “certainty of harm” requirement we looked at in the last posting is that
3. Your means of inflicting harm must be seen to be reliable.
As in, will your “nuclear deterrent” work when you want it to? That is, not work to deter but work when it fails to deter. When it’s no longer really a deterrent but something else.
Here more technical questions may arise.
We learned in our tests how to make nuclear bombs very reliable indeed. If we wanted them to go off, they would. The “fizzles” we got (not fully realized detonations, which still had much larger yields than any conventional bomb) happened mostly because we were pushing the limits and trying something we weren’t sure would work. When we converted a “test device” into a bomb, we had something we could be very sure would work.
The fact that our nuclear bombs were so reliable brought up two important questions: 1. How do we keep them from going off accidentally, and 2. how do we keep them from working reliably if they should fall into the wrong hands.
How we addressed these matters was explored in the entries linked above.
After we stopped doing explosive nuclear tests in 1992 during President George H. W. Bush’s administration and also stopped designing new nuclear weapons, worries emerged that the components of the weapons we were keeping in our stockpile might degrade over time, making them less reliable. In 1993, during President Clinton’s administration, Congress authorized a “Stockpile Stewardship Program.” Every year since then, our weapons labs and the Secretaries of Defense and Energy have had to certify to the President that our nuclear weapons are “safe and reliable.”
They have to come to this conclusion without doing any explosive testing. The tests must be, that is, “sub-critical.” Computers can help with simulations. Which are, of course, only simulations.
I visited Bradbury Museum at the Los Alamos National Laboratory in 1992, just after the conversion from weapons design to Stockpile Stewardship had begun. I could tell some change was afoot in the mission of the laboratory, but not exactly what it was.
Your nuclear weapons may be safe and reliable, but can you reliably deliver them to where you want them to explode? Will your bombers get through? Your ICBMs will be much more likely to get through but are they accurate? By the end of the 1960’s and still, the short answer is--probably. Some bombers. Most of the missiles. The missiles’ warheads may not come down on their targets, but close enough. With nukes, close enough is close enough. Close enough to have some capability to deter, you have to think.
Neither the Hiroshima bomb nor the Nagasaki bomb, both very small bombs by today’s standards, were detonated all that close to their targets. You probably know what they did to those cities.
Another way you might be able to keep an enemy from thinking they might get away with it would be to build a Doomsday Machine. That’s a machine that would be able to execute a second strike automatically, even if a surprise first strike had left nothing working in your country and no one alive.
During the Cold War, we and the Soviet Union both built Doomsday Machines. We got our system into service first, in 1963, the year I graduated from college. I didn’t know when I graduated that we had a Doomsday Machine in place. The Cuban Missile Crisis had happened that Fall. I was pretty nervous anyway.
Did the Soviets know we had our Doomsday Machine in place? I hope so. If they didn’t, our Doomsday Machine couldn’t have had a deterrent effect. Deterrence was, I think, the idea here.
The Soviets’ Doomsday Machine was called Perimeter, or Dead Hand. I don’t think we know even now exactly when they put it into service. It was during the Cold War but a while after our system--which was called the Emergency Rocket Communications System--had become operational.
At the end of the Cold War, we shut down the ERCS. Perimeter may not yet be shut down. In 2009, almost twenty years after the end of the Cold War, a Russian general said Perimeter was still operational.
If you had built a Doomsday Machine and it proved to be unreliable—meaning that after you’d been attacked in a surprise first strike it didn’t work—, I don’t suppose you would get complaints. From either side. For different reasons.
In any case, deterrence theory says that to be effective as a deterrent, your nuclear deterrent needs at least to be believed by the other side to be reliable.
Here’s another aspect of reliability. If you get warned that a first strike is in progress, will you have learned about it in time to launch a retaliatory strike? Not in time to defend yourself against the incoming ICBMs, which is still not possible, but in time to be able to launch your bombers and missiles before they are destroyed in the attack?
Intercontinental Ballistic Missiles launched from land-based silos would arrive at their targets, or come down wherever they were going to come down, about thirty minutes after they were launched. We figured that after our systems warned us they were coming and we got the word to the President, the President would have about fifteen minutes to decide whether to launch our retaliatory attack. Unless he wanted to wait for an actual detonation to be sure it wasn’t a false alarm. He would then be implementing what was called “launch on rideout” rather than “launch on warning.”
I wonder if anyone thought the president would ever choose “launch on rideout.” Would you ever have chosen it?
He’d have less than fifteen minutes actually because he’d have to leave enough time for the orders to go out to the launch facilities and be executed.
We practiced doing speedy launches of our bombers and our missiles after getting a warning. Our Strategic Air Command practiced these things. A lot. After we got into service the Minuteman ICBMs that used solid fuel and could be launched in only five minutes or so, we were pretty sure we could get our missiles up in time. Our bombers were another matter.
I don’t know who believed we really had enough time to get them up. Even if our practice sessions showed that we did. In the game itself, it’s always different, isn’t it?
It wouldn’t matter for us in any case, would it? Any incoming missiles would still be incoming. Getting our bombers and missiles up in time would matter only for the Soviets, not for us.
By the mid-60’s, there was also this. If the incoming missiles had been launched from submarines that were closer to us, they would arrive in maybe half the time it would take an ICBM. We knew that much. As we practiced getting our land-based missiles and bombers off within fifteen minutes, maybe we had just decided not to think about it.
To increase reliability, might you also want put your ground-based missiles in “hardened” facilities that would keep them from being destroyed in a first strike? During the Cold War, we spent a lot of money “hardening” the launch facilities for our ICBMs, not as much money as we had spent on building the launch facilities in the first place, but still a lot.
As missile guidance improved, we kept thinking we had to get harder and harder. Engineers were involved, of course. They would sign off on the designs but my sense is that no one really believed the launch facilities could be hardened enough to withstand a direct hit with a strategic nuclear weapon.
Something that is not easy to test.
Certainly the above-ground communication facilities—antennas, transmitters—would be unlikely to make it through. That meant that after the attack there might be no one to hear from or no way of hearing from them. Any survivors down in the hardened silos wouldn’t know which it was.
The missiles we had in our submarines didn’t have to think about hardening. They didn’t have to launch on warning either. They could wait to see what was up.
Then it might be a question of whether they could still hear from home.
At the beginning of the 1960’s, very early on in the missile era, the Chief of Staff for the U.S. Navy proposed that the submarine-launched ballistic missiles we had developed would suffice for deterrence. The Air Force, you may not be surprised to learn, strongly disagreed.
Our Air Force still has today over 400 ICBMs on alert in hardened silos in fixed sites in three big missile fields inside the U.S., one at Minot, North Dakota, one at Malmstrom Air Force Base in Montana and one at Warren Air Force Base in Wyoming. (I was able to visit an active launch control center at Warren. Here’s the lead entry of several about that visit.)
Because the locations of the fixed sites would certainly be known at this point to anyone who might want to launch ICBMs or SLBMs against them, the missiles are still on Cold War—a.k.a. “hair trigger”-- alert. In an alarm, or a false alarm, they could be launched within five minutes or so of receiving the authenticated launch order.
ICBMs, unlike bombers, cannot be called back. We could put something in the missiles that would allow us to destroy them in flight, but we didn’t want to because the Soviets might also figure out how to do that.
Today, with the very great increase in accuracy and power of our Submarine Launched Ballistic Missiles, the Navy’s argument would seem to be even stronger.
Except that now, in 2022, Russia claims to have developed drone torpedoes that can be left on the sea floor somewhere until needed, and when activated, travel thousands of miles guided by artificial intelligence before detonating a 100 megaton warhead on, say, Britain’s western shoreline. One of those would finish off Britain. It would also finish off any submarines in a very wide area, I’m sure we don’t know how wide.
Something else that would be hard to test.
Next: More Notes on Certainty and Reliability