She didn’t wait for my reply. ‘Don’t worry. I don’t have your complicated history, nor your great commitment to the revolution, though I do have my own little commitment. So there’s no need for you to worry. I’m not going to commit suicide, which would be a crime against my parents; I’m also not going to stop work, which would be a crime against the Party and the Chinese people, and unfair to Director Tie, Unit Director Luo and you. From now on I will go to work regularly. Don’t worry.’

I said abruptly, ‘I’m going to Moscow tomorrow.’

She seemed a bit surprised at that, and asked if I was going to see Anderov. I said yes, and she expressed some misgivings. ‘He wouldn’t write back to you – do you think he is going to see you?’

I said he would; if I was there in person he would certainly want to see me. She said that if I popped up suddenly like that, even if I did see him he wouldn’t say anything useful, people like that being very sensitive. I said that I had a really good excuse for going to see him, given that I wanted to perform a ‘summoning the soul’ ceremony for Xiaoyu. Xiaoyu’s soul was still lost and far from home, which meant that she could not rest in peace. It was imperative that her soul be summoned. I didn’t care if he really believed in that kind of thing or not, what mattered was that it made a viable excuse. ‘The reason I have come to see you is to ask, what kind of thing is it that you want to know from Anderov?’

That question seemed to go right to the heart of things, and she immediately got excited and said, ‘Fantastic! I’ll write you a list this evening.’

I said this evening would be too late since I was leaving first thing in the morning, and besides something like that probably shouldn’t be put into writing: she should just tell me.

She thought about it, and said: ‘If at all possible, I’d like to know what Anderov thinks of Sivincy’s cryptographic skills: whether she is likely to pull any particularly nasty or unexpected tricks, and whether she is capable of producing a super-difficult cipher. If she isn’t, then we can eliminate one of the four possibilities that I discussed with you: RECOVERY can’t possibly be a super-large-value algorithmic cipher encrypted with a medium-large-value algorithmic cipher. Finding that out would be crucial, because if RECOVERY really is that kind of code it would be bad news for our chances of decrypting it. The calculations required would be massive, and we simply don’t have the capacity: we can’t deal with it. In that case we won’t be able to crack it for at least another couple of years.’