"Two reasons. The first is that no one else is careful enough and their civilizations keep destroying themselves and I could do better. The second is that at the center of the Radch there is a shell built around a star. No light escapes it. No outsiders can enter it. Protecting what's in it is very important to me. It was necessary to have the industrial capacity, and the command of its physical surroundings, to protect it against anything that might threaten it."

Isabella doesn't have an immediate reply to that.

"Neither of those things changed recently, exactly, but - I am slow to change my mind on human timescales and considerations that might more wisely have caught up with me long ago became compelling a century ago, and straightforward to act on a few decades ago."

"At about what rate do other civilizations destroy themselves?"

"Once they're at the requisite tech levels then on average they'll make it two more centuries. Gets likelier for the first century while everything gets cheaper and easier and then after that it's approximately a fixed chance every year."

"Why would you say no one is solving this problem in a satisfactory way besides you?"

"No capacity to build artificial intelligences sophisticated enough to do the surveillance work. If you try to do it with humans it's by far your civilization's biggest expense, and wildly unpopular, and has a worrying failure rate anyway."

"I'm curious if you can guess my next question."

"Why we haven't shared?"

"Yeah - maybe there's some detail I'm missing but AIs don't sound like the kind of thing you couldn't loan out."

"They're a little tricky, the infrastructure requirements are fairly enormous, but it wouldn't be impossible. No, the problem is that they're dangerous, too. Badly-designed artificial intelligences are a catastrophic risk themselves, not necessarily contained in the slightest to the planet that built them, and no one else has figured it out in three thousand years but if they had one to reverse-engineer from then I suspect they could."

"Would nobody take an AI and some Radchaai AI-babysitters to prevent such a high risk of catastrophe?"

"I am optimistic that once our reputation for conquest has faded there'll be takers."

"...that does seem like it'd present an obstacle, but there are hundreds of civilizations, and an AI and some babysitters would tend to stabilize power in the hands of whoever you gave it to and reduce your ostensible motives to conquer them next, compared to the apparently high levels of instability otherwise, uh, enjoyed, so I'm still surprised nobody'd take one."

"So logistically speaking, what you need is a orbital station at least sixty miles across, because heat dissipation and gravity both present too much of a challenge to building it planetside, and then the infrastructure to service it - a minimum staff of a couple hundred - and then planetside installations that do everything requiring lower latency, which are much much more extensive but also net cheaper because building things in space is hard. It might reduce our incentive to conquer people but it'd also make it vastly easier and lower the costs of integration, and those effects are more visible, and it's only stabilizing once it's built which is only eight or ten years but still a time horizon many politicians don't think on."

"...okay. I still wouldn't be surprised if you could find someone who'd take one but I can understand given the givens that you would not have made extending the offer a diplomatic priority."

"Why am I here," Xander asks his sister in an undertone.

"So I'm not spending any attention wondering if you are being held hostage or something," she whispers back.

"Can you translate me, like, a book or something," Xander mutters to his tablet.

His tablet can absolutely do that.

"You're probably right that there's a taker out there. I've been focused on managing the transition internally. It's been complex."

"There are thousands of you. You don't... have to focus on things. Do you?"

"Not in that sense, but a government can still only have so many priorities however unbounded my attention, and in particular messaging can generally only have one priority. 'we're switching over to exerting control in a different fashion' would be the wrong messaging."

"Okay, that makes sense.

What remains to be explained is - I keep pulling out fancy words I don't think our friendly tablet here has heard yet but so far nothing terrible has happened - the cultural hegemony?"

A back-and-forth with the tablet. "- there I think we might be stuck."