Since Russian President Vladimir Putin announced a heightened nuclear alert level, much of the world has worried that he might go nuclear in his war against Ukraine. But there is another concern: a false alarm from Russian early-warning systems, which experts believe are dangerously vulnerable to errors.
The risk of a catastrophic mistake has been a threat since the outset of the nuclear age, but miscalculation becomes more likely in a period of heightened Russian-American tension. Leaders would have only minutes to make fateful decisions, so each side needs to be able to “see” clearly whether the other has launched missiles before retaliating. Ambiguity in a moment of “crisis perception,” the Rand Corporation has noted, can spark “conflict when one nation misinterprets an event (such as a training exercise, a weather phenomenon, or a malfunction) as an indicator of a nuclear attack.”
Russia and the United States are the most heavily armed of the nine nuclear powers, which include China, France, the United Kingdom, North Korea, Pakistan, India, and Israel. (Iran is poised to join the club.) But only the U.S. has comprehensive surveillance of the globe, provided by three active geosynchronous satellites, with two in reserve, whose infrared receptors can spot missile plumes. That data is supplemented by radar, which gives the U.S. the capacity to double-check that a launch has actually occurred.
Specialists call this verification by both satellite and radar “dual phenomenology.” The Russians don’t have it, at least not reliably. They lack adequate space-based monitoring to supplement their radar.
What they have is a “terrible and dangerous technology shortfall,” according to Theodore Postol, a professor of science, technology, and national security policy at MIT and a former scientific adviser to the chief of naval operations.
He believes that Russian satellites are handicapped by their inability to look straight down and distinguish the infrared signature of a missile launch against the Earth’s terrain. “Imagine that you took a photograph of a complex and rocky area of ground,” he explained in an email to me. To pick out an ant, you’d need to find it in “some very small pixel in a vast array of pixels.” In the infrared part of the spectrum, you need multiple high-quality sensors, each with a small enough field of view to discriminate between the background and a rocket plume and to avoid a false detection from reflected sunlight or extraneous interference.
American satellites look down with sophisticated sensors and orbit in fixed positions relative to the Earth’s surface. By contrast, Postol says, Russian satellites have to look sideways, along a line that forms a tangent over American ICBM missile fields in Wyoming, Montana, and North Dakota. Nine Russian satellites in elliptical orbits take turns observing small areas, looking horizontally for missiles that rise above the horizon to spot their plumes against the dark background of space.
The slanted perspective is less reliable, Postol said. First, it covers only a minuscule segment of North America. Second, it won’t see missiles until they reach higher altitudes, where the rocket plumes are dimmer and harder to define. Third, “luminosity from atmospheric phenomena causes background effects that can lead to false detections.” The mirage effect can be seen in a recent image from a newer Russian satellite. Finally, he said, if a missile is not launched precisely where the satellite’s tangent of sight meets the Earth’s surface, the rocket might never rise above the horizon.
That might seem to give the U.S. an advantage of a surprise attack. But if Russian radar picked up a missile later, the time before impact would shrink dangerously—to just over 10 minutes from a U.S. submarine near Norway, about 17 minutes from Wyoming—with little chance to double-check that the launch is authentic. Without global coverage, Russian satellites cannot see submarine launches carrying ballistic missiles with multiple warheads until the missiles enter Russian radar. Knowing their system’s limitations, Moscow might strike quickly, even letting lower officials make the call.
“There is not adequate decision-making time for leaders in Moscow to understand what might be happening if it looked like there was a general attack on Russia,” Postol observed. “They would have no ability to assess the situation. If they want to have the option to assure that they can retaliate, they will have to pre-delegate launch authority to people in the field. If you do that, you run the risk of a gigantic, spasmodic launch of all your nuclear forces.”
Another specialist, Pavel Podvig, agrees on the time problem but argues that geography is the main obstacle to double verification. He is a Moscow-trained physicist and arms control specialist who heads a Geneva-based project analyzing Moscow’s nuclear forces. “The radar warning comes too late to provide a useful check of the satellite information,” he told me in an email—although radar did exactly that in 1983.
In the early morning of September 26, 1983, a Soviet satellite signaled that five American ICBMs had been launched. Officers in Serpukhov-15, a command bunker near Moscow, watched in horror as electronic maps pulsated and lights flashed. “For fifteen seconds, we were in a state of shock,” recalled Lieutenant Colonel Stanislav Petrov, who would pass his assessment to the general staff, who would convey it to Soviet leader Yuri Andropov.
For five minutes, a phone in one hand and an intercom in the other, Petrov tried to interpret the cascade of warnings. Two things didn’t add up. One, “when people start a war, they don’t start it with only five missiles,” he told the Washington Post reporter David Hoffman. Two, Soviet radar didn’t see any missiles flying. Through a veil of high tension, “I had a funny feeling in my gut,” Petrov recalled. Balancing the odds at 50-50, he deemed it a false alarm. An investigation found that the warnings had been triggered by sunlight reflecting off clouds.
Since then, and following the early-warning system’s decline during Russia’s economic crisis in the early 1990s, the country has surely advanced. But how far? Podvig and Postol disagree on the state of satellite technology. Podvig told me in several email exchanges that the newer Tundra satellites “appear” to have look-down capability. However, he stopped well short of backing up his assertion with hard information, much of which is probably classified and inaccessible.
For his part, Postol cited circumstantial evidence of shortfalls. “If the new Tundra satellites were able to look down at the Earth,” he said, “only two or three of such satellites would be needed.” The Russians have nine, and they orbit through staggered observation slots lasting two and a half hours each, consistent with less reliable tangential lines of observation.
So Russia relies heavily on radar, which cannot detect anything over the horizon. A missile would not be seen until it passed rapidly through the radar station’s fan of surveillance. That would add another minute for tracking a target, defining it, and calculating its trajectory. Also, radar can be blinded by ionized air from a high-altitude nuclear detonation, and radar disruption from another cause might be mistaken for a prelude to a nuclear attack.
Russia’s nuclear early-warning system poses two big questions: Who decides to launch, and on what basis? Postol imagines that Moscow might adopt a pre-delegation posture that would decentralize or automate Russia’s launch authority—raising the likelihood of error. And what would the trigger be? Podvig argued in 2016 that under Moscow’s doctrine dating from Soviet days, Russian commanders would delay a retaliatory launch until after “signs of the actual attack (such as nuclear explosions).”
If that wait-and-see doctrine existed, it appears to have been revised. The Bulletin of Atomic Scientists reported recently that in June 2020, Putin updated the nuclear deterrence procedures, which permit a launch only under certain conditions. The first of those reads: “arrival of reliable data on a launch of ballistic missiles attacking the territory of the Russian Federation and/or its allies.” That echoes the dangerous launch-on-warning doctrine that makes inadvertent war more likely—and early-warning systems more crucial.
Then, too, pre-delegation seems to contradict what is known of Moscow’s standing procedures. Leonid Ryabikhin, an expert in missile technology and arms control who taught at the Soviet Air Force Academy, wrote in 2019 that a nuclear launch would require encoded authorization from at least two of the three portable terminals controlled by the president, the defense minister, and the chief of the general staff.
Only if the government were decapitated by, say, a first strike on Moscow would a backup system kick in. Known in the West as the “Dead Hand” and in Russia as “Perimeter,” it has been understood as a computer process that would launch automatically if command-and-control connections with the leadership were destroyed.
But is it fully automated? Not necessarily. A 2003 book by Russian Colonel Valery Yarynich, who served in the central command of the Strategic Rocket Forces, reported that human intervention would still be required at “the super hardened radio command and control center,” located somewhere underground. “The [Perimeter] system has no capability for preparing a launch order automatically without the participation of the center’s crew,” he wrote. “If, with[in] a certain period of time, there is no evidence of actual nuclear explosions” from seismic shock, a spike in air pressure, or radioactivity, no missiles would be launched.
U.S.-Russia arms control treaties address the numbers and deployment of various nuclear weapons but have left Moscow and Washington capable of obliterating humanity many times over. The chance of error remains unaddressed.
The probability of nuclear war by accident would be reduced if Russia and the U.S. shared surveillance information, a proposal discussed in 1998 by President Bill Clinton and Russian President Boris Yeltsin. A brief experiment in cooperation was held from December 1999 to January 2000 to avoid Y2K computer problems at the turn of the millennium. Eighteen Russian officers spent about three weeks at Peterson Air Force Base in Colorado to ensure that nothing went awry.
The permanent Joint Data Exchange Center, as it was to be named, fell victim to U.S.-Russian tensions following Putin’s election in 2000. By the time President George W. Bush met Putin in 2001, the late John Steinbruner wrote, “the building designated to house JDEC sat abandoned in Moscow on an overgrown lot and was reportedly being used by local teenagers as a drinking hangout.”
Postol believes that to help Russia avoid error, the U.S. must provide it with infrared sensors to differentiate a missile plume from background terrain. “Possessing the sensors gives you no information about how to fabricate them,” he said, “and therefore does not release any sensitive technology secrets.”
That’s a voice crying in the wilderness. Imagine how hard it would be to achieve such cooperation today, just when it’s needed most.