NASA

According to the latest numbers, there are nearly 2,000 working satellites circling Earth right now. Round and round they go, quietly weaving a web of technology that helps power the world. Sometimes they hit a snag. A satellite can find itself on a dangerous collision course with another object—usually a piece of space junk, sometimes another satellite.

This is the scenario that played out between the European Space Agency and SpaceX earlier this month, on the Labor Day holiday in the United States. Orbital data showed a probability of collision that was becoming worrisomely high. Unless someone made a move, the satellites could smash into each other within days. A crash could be disastrous. So ESA sent SpaceX an email.

Email, believe it or not, is a common form of communication between satellite operators, whether they work for an intergovernmental agency of 22 nations or a company founded by Elon Musk. Sometimes they might use the phone. There is no international network of notifications for satellite traffic. When the data suggest trouble on the horizon, one satellite operator usually fires off an email and then waits for the other to reply.

Usually, the operators discuss who will dodge whom, tweaking a satellite’s altitude just enough to avoid danger. But slow responses—or none at all—can make the looming encounter seem like a game of chicken.

ESA was the first to reach out. SpaceX responded, saying it had no plans to adjust the position of the satellite. The probability of a collision was about one in 50,000 then, and satellite operators usually don’t make moves until the likelihood reaches more than one in 10,000.

The satellites hit that threshold within a day of the email exchange. ESA sent SpaceX another email. Then another. But the agency didn’t hear back.

Had the operator seen the messages, a SpaceX spokesperson later said, “we would have coordinated with ESA to determine best approach with their continuing with their maneuver or our performing a maneuver.” But SpaceX, it turned out, didn’t see those emails, thanks to a computer bug in their on-call paging system.

By the time they did, the ESA satellite had already ignited its thrusters and dodged. SpaceX had stumbled into a painfully relatable circumstance: missing an important email and then scrambling to explain why with a sorry, just seeing this!

The two satellites avoided a crash, which can produce thousands of pieces of debris, increasing the chance of other collisions.

But the scenario raised concerns about how satellite operators manage space traffic as it becomes even more congested. The SpaceX satellite, launched in May, was part of the company’s Starlink project, which is meant to provide internet around the world, especially in isolated spots. Musk says he wants to launch about a thousand satellites each year to eventually build a fleet of 12,000. Jeff Bezos wants to launch thousands of satellites of his own, for a similar effort under Amazon. The internet-satellite company OneWeb plans to deploy hundreds.

As these constellations of satellites grow, so does the risk of collision. And more risk means more emails and phone calls.

“If you have to deal with one or two close approaches a month, picking up the phone might be fine,” says Brian Weeden, a space-policy expert at the Secure World Foundation, a nonprofit group that promotes peaceful uses of space. “But if you now have to deal with dozens or hundreds, it’s not going to scale.”

Collision avoidance is a knotty process from start to finish. ESA, SpaceX, and virtually every satellite operator rely on orbital data collected by a unit of the U.S. military. The Air Force’s 18th Space Control Squadron tracks about 23,000 artificial objects bigger than a softball, a tiny slice of which are functioning satellites. If a collision starts looking possible, the unit issues warnings to the appropriate operators.

But the data are limited, which Weeden attributes to aging computer systems and information-sharing policies, aimed at protecting national-security interests, that can leave operators in the dark. In practice, that means satellite trackers don’t know exactly where something is in space, so they cannot say with certainty whether two objects will collide. When operators prepare for potential maneuvers, they combine the military data with their own knowledge of the satellite’s position, taking into account “uncertainties in orbit information,” as ESA put it.

In most cases, one half of the standoff is a piece of inactive junk, a dead satellite or a shard left over from a previous collision. In these cases, satellite operators take unilateral action. Two working satellites, and things can get awkward. Earlier this year, Sarah Scoles at Wired described the reactions of Spire and Capella, two commercial companies, to a collision warning. Spire has more than 80 satellites and “wasn’t that worried about this warning,” Scoles wrote. “Capella, which has just a single satellite, was.” Imagine that email chain.

In the United States, satellite operators require approval from regulators to launch, but they aren’t required to take certain actions once in orbit. “There really aren’t any agreed-upon norms on collision avoidance between two active satellites, let alone legal rules,” Weeden says. “We’ve always relied on the assumption that satellite operators will be responsible and ‘do the right thing.’”

The satellite owners can communicate through the Air Force or initiate contact themselves. Sometimes they’re neighbors who are used to making room for each other in their shared orbit. The conversation is simple, quick. “When you’re living in an orbital shell like ours, you start to develop the relationships with others that are in that area, whether that’s a constellation or single vehicles,” says Walt Everetts, the vice president for satellite operations and ground development at Iridium, a U.S. company that has more than 70 communications satellites.

When facing down an unfamiliar satellite, operators can usually find contact information in the warnings from the U.S. military or online space-tracking databases. If the information isn’t there, they’ll have to Google it, says Klaus Merz, who handles collision avoidance at ESA’s Space Debris Office.

The first emails usually strike a tentative tone. Hey, I’m seeing this. Are you seeing this? What should we do? Merz says ESA staff had never before corresponded with their SpaceX counterparts about satellite trajectories. “Dear SpaceX colleagues,” the note began.

According to The Verge, the Air Force squadron sent nine updates to both ESA and SpaceX in the 72 hours before the satellites were expected to make their closest approach. Faced with silence from SpaceX, ESA officials decided they’d be the ones to back down. They had to trust that SpaceX wouldn’t budge; if both satellites shifted positions, they could end up smacking into each other anyway. Musk has said the Starlink satellites are designed to receive warning alerts and maneuver autonomously around other objects—a procedure it has already tested more than a dozen times, according to Quartz. But it’s not clear whether this satellite was equipped with that technology, and the ESA situation suggests humans are still in the loop.

Even after SpaceX discovered the apparent email glitch, no one emailed ESA back, Merz told me. (ESA learned about the glitch in media reports, he said.) A SpaceX spokesperson says staff “implemented corrections” and reached out to ESA that week.

The Air Force’s 18th Space Control Squadron counted just over 100 dodge maneuvers in 2018, though since satellite operators aren’t required to report the dodges, the true number may be much higher. Most warnings about potential collisions don’t amount to real danger: Last year, the 18th Space Control Squadron generated nearly 9 million, according to an Air Force spokesperson, and only 1 percent met what the unit deemed “emergency criteria.” With so many notifications flying around, experts say the industry needs a more sophisticated system for managing space traffic, equipped with some automation and a bit more information-sharing.

Automated space-traffic control is a nice idea, but probably unrealistic, says Lisa Ruth Rand, a historian who studies space debris. The international space community hasn’t agreed on expansive principles since the United Nations Outer Space Treaty in the 1960s, which governs the use of space to this day. (There is even some disagreement within the U.S. government about which agencies should be in charge of overseeing space-traffic matters.)

“Having a centralized system would make sense, but you’re talking about an issue that’s global in nature, that deals with people in different places, using different proprietary systems,” Rand says. “It would require quite a bit of coordination and a much more internationally bureaucratic framework.” Faced with that kind of negotiation‚ not exactly known for its success rate, email might seem like the most painless method.

In one peculiar sense, the reliance on email fits into the norms of space exploration quite neatly. “Outer space,” the United Nations treaty states, “is not subject to national appropriation by claim of sovereignty.” Email is similarly untouchable. “Email is the last great unowned technology, and by unowned I mean there is no CEO of email,” the Harvard law professor Jonathan Zittrain once said. “It’s just a shared hallucination that works.”

Until, of course, someday, it doesn’t. People have been remodeling the skies for decades, and they’ve begun to look more like traffic lanes than sacred spaces, managed by the same mundane communication tools, subject to computer bugs and human error. Scientists and historians warn that the space above Earth could grow so crowded that it becomes impossible to function there, or that a few collisions could turn into a cascade that knocks out whole chunks of important infrastructure. We might be more comfortable working in orbit now than ever before, but the stakes haven’t changed. Up there, getting to “inbox zero” might make the difference between harmony and disaster.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.