SIEMENS

What a view!” she gasped. We were standing in one of the station’s glassed-in observatories some 22,000 miles above the Earth, a bright blue and white disc about the size of a tennis ball, much of it baking under cloudless skies, half of it in darkness. “Yes,” I said. “As often as I come here, the view is always mesmerizing.”

“So what’s the story?” she asked. “You sounded concerned when you called.”

“I am,” I said. “And I didn’t want to discuss it any other way but face to face.”

Marisa and I go back decades. In fact, we grew up together in a small town near Milan and studied electromechanical engineering together. After grad school, I joined the European Space Agency, and “Risa,” as I call her, took a job at a company that specializes in industrial simulation technology — the same technology that was used to design and test every part of the space elevator complex and its sky-high “Exploratorium” where we were standing, a vast complex of laboratories and offices, residential facilities, and launch and docking systems for future missions throughout the solar system. For the last five years my job has been coordination of the space elevator project.

“Something very strange happened a few days ago,” I began. “We believe that several of our ANS — you know, those ant-like Autonomous Networked Systems that we use for every kind of maintenance work out here — well, it seems that four of them — there’s no other way of describing it — cannibalized a seventh one.

“Cannibal…What do you mean?” asked Risa. “I mean there are a few seconds of video when a small swarm of ANS approached the lone victim, who was performing routine surface analysis operations for evidence of micro impacts,” I explained. “The victim’s eye-stream video then goes blank, and the attackers’ video feeds show nothing — as if the victim had been digitally removed from the images to avoid leaving a trail. But the RFIDs embedded in each piece of the victim have turned up in parts of the six attackers — and those attackers appear to be functioning more efficiently than any other ANS. How could such a thing happen?”

Marisa looked down at the seemingly endless meter-thick carbon nanotube cables that stretched downward from the Exploratorium, already allowing a steady stream of freight and passenger modules to ferry supplies and service personnel back and forth. The elevators, which were still being tested, would radically change the economics of space travel. Until recently, it cost as much as $20,000 per kg to carry materials into geo-stationary orbit on old-fashioned rockets. Once in full operation, the space elevator was expected to slash that to about $100 per kg.

“You guys did the simulation and testing. We just provided the software to do it with,” said Risa. “That’s right,” I answered. “And you know that we performed exhaustive simulations on everything from the materials to the individual parts of every system. We simulated how those materials and parts would hold up under the temperatures, radiation levels, and stresses they would be subjected to up here. We simulated their functional interactions, and even the manufacturing procedures for all the parts we used — including the ones used in the ANS. But Risa, there was always that business of uncertainty quantification. What we studied back in college about predicting crack propagation and determining associated levels of risk, which was hard enough to figure out, has given way to a much thornier question — the uncertainty of predicting how networked systems function when they acquire intelligence, communicate, and learn from one another.”

“You said these attacker ANS have become more efficient,” said Marisa. “What did you mean by that?” “See for yourself,” I said, touching the transparent surface of the observatory. In a direct line of sight behind my finger we could see a group of ANS combing the surface of the elevator complex for micro impacts and cracks. As I touched the glass-like surface, it’s OLED interface came to life like a giant monitor, instantly recognizing my fingerprint and genetic signature thanks to an embedded electrophoretic layer, thus giving me secure access to a world of data.

I zoomed in on the area of interest so that we could observe the ANS’ appearance and behavior. “Most of our ANS work alone or in pairs,” I explained, “and these ANS used to as well. But now, look at them — all of them are moving in unison. If we interrogate their data exchanges, which include real-time compressed video, ultrasound, X-ray, thermal sensing, and of course self-diagnostics, they are all operating in perfect synchrony. They are collectively analyzing the surface for any sign of damage. It’s as if they had formed a single, mobile multi-sensor. “What’s more,” I added, “if you look at their RFID signatures,” — I made a few quick movements of my fingers to change the perspective — “you see anomalies, chips, for instance, that were originally in the victim. Each manufactured part, after all, has its own IP address. These ANS have upgraded themselves!”

“Fascinating!” Marisa said, almost to herself as she watched the ANS assiduously working their way over the surface of the complex. “To be perfectly honest, one of our newest research projects has been examining the probability of so-called spontaneous origination of behaviors among different classes of smart, networked devices. But now it’s happened! The ANS’ mission, if you think about it, is not just to identify anomalies that could turn into problems, but to optimize whatever they come in contact with. And now they’ve taken a step toward self-optimization in order to make a process more efficient. It’s swarm intelligence and more, it’s the essence of the Internet of Things. What we’re witnessing is simply a system that’s doing what it’s supposed to do.”

“So you don’t think we need to be too concerned about what’s happened?” I said. “I mean, isn’t there a possibility that this sort of thing might pose a risk — that the ANS might surprise us in ways that are not optimized for us?” “Oh come on,” said Risa, “they’re just machines. And if you really get nervous, just press the ANS System Central Reset in the control center, take the four so-called offenders out of the loop for study, and everything will go back to normal.”

“Risa,” I said, “that’s why I wanted to discuss this with you face to face rather than using any electronic communication. “I’ve already pressed the Reset. Nothing happened. "What will they think of next?"