And yet the core was there, and was thought to consist of a ball of solid iron surrounded by a somewhat hotter jacket of molten iron mixed with various other elements. All of this had been stripped bare and exposed to space by the Agent. For the first few hours, Peach Pit had literally glowed with radiant heat. Or so they guessed, since the dust kicked up by the cataclysm had cloaked it for a while. Some of the core’s outer jacket of molten metal must have been torn away, dispersed into the rubble cloud as gobbets and slugs and droplets of melt that soon cooled and hardened. As much was proved by metal-rich bolides that had since plowed into the Earth and been dug up and analyzed. By the time the dust had literally settled to the point where Peach Pit and its siblings were clearly observable, an outer crust had formed over it, consisting of melt that had cooled swiftly as it radiated its heat into space. The cooling had continued ever since. Now, the better part of a year later, Peach Pit, or PP1 as it was now designated, was still warmer than the other parts of the moon. It had shown greater resistance to fragmentation. Other rocks bounced off it, or dashed themselves to pieces on its gleaming surface. A few significant chunks—PP2, PP3, and so on—had been ripped off in the early days when it had still been soft, but now it was clad in a mile-thick armor of solidified iron that was proof against just about any calamity short of a second Agent.

Doob became so absorbed in such thoughts that he almost missed the transit of Izzy across the sky. It angled directly over the rubble cloud, seeming to weave among the giant tumbling boulders, though this was of course an illusion. It had long been the brightest man-made object in the sky, and it was brighter now that so many pieces had been added onto it. The effort had been impressive. Stirring, even. But seeing it against the scale of the disaster behind it forced him to ask himself what was the point. What was the longer-term plan for the Cloud Ark? The swarm concept was a nice architecture, much more survivable than One Big Ship, but where was it going to go?

No one seemed to be talking about that. He understood why. Survival was the first imperative. Long-term strategy came next.

The amount of iron in PP1 was for all practical purposes infinite. It would take humans many thousands of years to find uses for that much metal.

But it was way up high. Hard to reach.

And yet they had to reach it.

And it was closer, easier to reach, than the Arjuna asteroids that Sean Probst was so excited about.

Feeling an idea take shape in his head, like an iron core congealing deep in a moon, he put it on hold and forced himself to turn his attention to more immediate questions. A few days ago in the Oval Office he had formed a resolve to get his ass into space and begin making things happen up there. Which was fine. But he had three months left on terra firma. He couldn’t neglect his responsibilities here. Some of which—the most important—were to his kids, to Amelia, and to their frozen embryo. But on top of that he had been given other jobs, and if he screwed them up badly enough, e.g., because he was standing on hotel balconies in the middle of the night thinking about how much iron was in PP1, then they might not send him up to the Cloud Ark at all. He hadn’t wanted to go, but once he had assented to the idea, he had begun wanting it more than anything, and he now feared that they would take it away from him. And if they sensed that fear, they could use it to control him. Better to overperform, to exceed expectations, to act like it was nothing at all.

SEVENTY-TWO HOURS LATER HE WAS LOOKING OUT THE WINDOW OF a U.S. Navy helicopter banking through a misty Himalayan valley as it lined up its final approach to a runway in Bhutan. Or perhaps the runway in Bhutan was the more correct phrasing.

There were about 750,000 people in this country, which meant that they were entitled to supply two candidates for the Cloud Ark. The arithmetic was a little fuzzy; if the same ratio were applied consistently all over the world, something like twenty thousand candidates would be gathered in. If an arklet could accommodate five people, then four thousand arklets would be needed in the swarm. Each arklet required a heavy-lift rocket to get it into orbit, and some assembly and prep work once it had reached Izzy.

Could it be done? If the entire industrial capacity of the world were thrown into the production of rockets, arklets, space suits, and the other goods needed? Perhaps. But probably not. Doob was privy to some recent estimates that put the numbers at closer to one-quarter of that figure.

And anyway, could the arklets really support five humans each? Without a doubt they were large enough for five people to bang around in, but it was not at all clear that each could be self-sufficient in food production. Building a sustainable ecosystem in a tube the size of a railway tank car was no small task. Biosphere 2, a well-known experiment in the Arizona desert, had attempted to support eight people on an ecosystem the size of a couple of football fields, and been unable to make it work long term. But its mission had been clouded by political strife and odd quasi-spiritual factors. A more down-to-earth project run by the Soviets had determined that eight square meters of algae—an expanse of pond scum about the size of two ping-pong tables—was needed to keep a single human supplied with oxygen. In the space between the hard inner hull and the inflated outer hull of a single arklet there was more than enough room. But much more real estate would be needed if the arklet were also to produce food. And those calculations didn’t even begin to address the real complications of keeping thousands of people alive in space for many years. It wasn’t enough just not to asphyxiate and not to starve. People would need medicine, micronutrients, recreation, stimulation. Ecosystems would get out of whack and need to be repaired with pesticides, antibiotics, and other hard-to-make chemicals. The thrusters that kept the arklets out of trouble would need to be refueled, and not only that but they would need maintenance and repair. The idea of a completely decentralized Cloud Ark was a chimera; it was not sustainable without a mother ship, a central supply dump and repair depot. The only plausible candidate for that was Izzy. But Izzy wasn’t designed for anything like that purpose. They’d been trying to make it over by cramming it with vitamins, but that only delayed the moment when they’d run out of all the goods they didn’t know how to produce in space, and people would begin dying in quantity.

From the fact that he had gotten nowhere raising awkward questions about this, Doob inferred that the Arkitects knew about it, and were on it, and just didn’t want to talk about it because public doubt and controversy were not going to help. Doob’s job, clearly, was to act like everything was okay. Today, that meant scooping up two young people from the Himalayan kingdom of Bhutan.

Did the little performance he was about to put on really mean that twenty thousand people from all over the world were going to end up living happily in the Cloud Ark? He just had to shut down the little Rain Man in his head—“Doob, as in dubious”—and not even think about it.

They had taken off two hours ago from the George H. W. Bush, a supercarrier keeping station in the Bay of Bengal. Doob had viewed the ship through the eyes of a man who, in a few months, would be making a permanent move to its orbiting equivalent. She was a completely artificial island, thousands of people densely packed into a wad of pure technology. The professionalism of the crew and the efficiency with which she ran were amazing. Could something like that be duplicated in space, with people chosen by lot from all over the world, and trained in camps over the course of a single year?