“There are quite a few little crater pits to each side of the break here. Like that one…” Genette pointed with a forefinger, and a green laser dot appeared on the rim of a small pit, quickly circled the rim, then fixed at its center. “See that? And then that?” Circling another one. They were very small. “These are fresh enough that they may have happened during or after the break.”

“So, ejecta?”

“No. Gravity here is so slight, the ejecta seldom come back. If anything did, it would almost dock. These pits are deeper.”

Swan nodded. The asteroid’s lumpy surface had many rocks lying loosely on it. “So what did the accident report call these craters?”

“Anomalies. They speculated they might be pit ruptures, where ice deposits melted at the heat of impact. Could be. But I take it you have looked at the accident report for Terminator?”

“Yes.”

“Do you remember there were anomalies there too? Whatever struck the tracks didn’t hit cleanly. There are outlier craters, very small, that were not there before the event. Now, on Mercury they could be ejecta coming back down, I grant you that—”

“Couldn’t the impactor have broken up coming in?”

“But that usually happens where there’s an atmosphere heating and slowing it.”

“Couldn’t Mercury’s gravity do it?”

“That effect would be negligible.”

“I don’t know, so maybe it didn’t break up.”

The little figure nodded. “Yes, that’s right.”

“What do you mean?”

“It didn’t break up. In fact, it came together.”

“What do you mean?”

“I mean it was never comglomerated, until the very last moment. That’s why none of the detection systems on Mercury saw it coming. They should have seen it, it had to come from somewhere, and yet it was not detected by the surveillance systems. So to me this indicates an MDL problem. Minimum detection limit. Because there is always a minimum limit of detection, either inherent to the detection method, or else artificially set higher than the actual minimum.”

“Why do that?”

“Usually to keep warnings from going off all the time when there isn’t really any danger.”

“Ah.”

“So, each system is different, but in the Mercurial defense apparatus, what they call the method reporting level is almost equivalent to the system’s method detection limit. In other words they set their reporting level at twice the detection level, which is six or seven times the standard deviation in their measurement variability. It’s a typical setting to make people comfortable they’ll generate both the fewest false negatives and the fewest false positives.

“So, but consider what then lies below that reporting level. Basically, only very little rocks—pebbles, well less than a kilogram each. But if there were a lot of them, and they converged only at the last second, with each one coming in from a different quadrant of the sky, and at a different speed, but timed such that they all arrived at the same spot, at the same time… Then they would just be little pebbles, until the last second. They could have been tossed from the far side of the solar system, maybe, and over a number of years, maybe. And yet even so, if thrown correctly, eventually they make their rendezvous. Many thousands of them, let us say.”

“So, a kind of smart mob.”

“But not even smart. Just rocks.”

“Could that work? I mean, could anything calculate how hard to throw them, and on what trajectory?”

“A qube could. With enough of the solar system’s masses identified as to locations and trajectory, and enough calculating power, it can be done. I asked Passepartout to do it—to calculate an orbit for something like a ball bearing or a boccino, thrown from the asteroid belt to hit a particular target on Mercury. It didn’t take long.”

“But could the throws be made? I mean, would it be possible to build a launcher that would launch them with the necessary precision?”

“Passepartout said there are machines in existence with tolerances two or three magnitudes more precise than would be necessary. One would only need a steady launch platform. The stabler the better, in creating consistency.”

“That’s quite a shot,” Swan said. “How many masses get included in the trajectory calculation?”

“I think Passepartout included the heaviest ten million objects in the solar system.”

“And we know where all those are?”

“Yes. Which is to say the AIs know where they are. And all the biggest terraria and spaceships conform to itineraries set years in advance. As for the calculations, it takes a qube to be able to do it in a reasonable amount of time, meaning fast enough to use it for real-time launch instructions.”

“How long does it take?”

“For a qube similar to Passepartout, three seconds. For conventional AIs, about a year per pebble, which of course would render the method inoperable. You have to have quantum computing to be able to do it.”

Swan was feeling sick to her stomach, as if she were back in the utilidor. “So ten thousand little rocks thrown downsystem, over a matter of months or years, with such directions and velocities that they all arrive at one spot at the same time.”

“Yes. And a few stochastic gravitational fluctuations no doubt cause a little bit of scatter at the end. Indeed when that happens, those pebbles must usually miss entirely.”

“But some just barely miss.”

“Exactly. Like these little pits we see. Caused perhaps by a spaceship that changed flight plans, or the like. So maybe one or two percent of the pebbles experience a clinamen of this sort, or so Passepartout guesses.”

Now the wrench in her gut was getting severe. “So someone is doing this on purpose.” She waved at the abandoned terrarium.

“That’s right. And also, a qube has to be involved.”

“Shit.” She put an arm across her stomach. “But how… how could someone…”

The inspector put a little hand to her arm. Ygassdrilfloated under them, cold and dead. A gray potato. “Let’s get back to the Justice.”

Back inside the Interplan hopper, after they had eaten a meal, Swan stayed up late in the galley, and again the inspector did too.

Swan, who had not been able to stop thinking about the day’s revelations, said, “So, all this means that whoever—”

Genette raised both hands and stopped her. “Qubes off again, please.”

After they had both turned off the devices, she continued: “That means whoever did this could have done it years ago.”

“Or at least quite some time ago, yes. Some stretch of time.”

“And there wasn’t a single launch site.”

“No. But maybe there is still the launch mechanism. Their gun, or catapult, or whatever it was, would have to be a very precise instrument. A particularly fine bit of manufacturing. The tolerances Passepartout suggested were really quite fine, requiring molecular printers and so forth. We might be able to find the factory that made something so particular—we’re looking into that. And then, who might have ordered it.”

“What else?” Swan asked.

“We are looking for the program for the factory, and the design of the instrument. Its printing instructions. Also the orbital program needed to make the calculations. Qubes don’t make that kind of thing up without being asked to do it—or so we have been assuming until now. The qube that did it would have that action recorded in it, as I understand it. And so the program is likely to still exist somewhere. And there are still only a finite number of factories making qubes.”

“Couldn’t they have destroyed their qube when they were done using it?”

“Yes. But there’s no reason to assume they’re done.”

This was a chilling thought.

“We must look for the qube, the orbit program, the factory program, also the factory, and the launcher itself, and whatever the launch platform was.”

Swan frowned. “All those could be destroyed, or cleaned pretty clean.”