Notes from the author

1) Yes, it is hard SF, even if it doesn't appear that way in the opening. Much will become clear by the end of the first chapter; most by the middle of the book.

2) I very much appreciate constructive criticism in the comments. I want to make it as good as I can.

3) I also very much appreciate comments where people say what they think is going on at various points. There are many foreshadowed things that eventually get revealed (and a big one that doesn't until the sequel), and I would like to know if I am making them blatantly obvious or hopelessly obscure.

4) Many of the characters argue things that I personally don't agree with. There are real dangers with AI research; I am not claiming to have the answers.

I watched the assassin drive his blade down towards Cicero's unsuspecting back. The blade was the embodiment of killing magic, and all of the assassin's considerable power was behind it. There was no time for me to shout a warning, and anyway, why should I bother? Cicero's smug interviewer was surely aware of the drama, but Sara had remained silent while the assassin had snuck up behind him. It would bring her great fame if Cicero were to be killed live during her interview.

The blade froze with a jolt just before piercing Cicero's light shirt, making the assassin stumble. Cicero stepped aside, remarked quietly to Sara that it was impolite to set him up, and clicked his fingers. The assassin disappeared, leaving just the blade, which resumed its deadly path through the space formerly occupied by Cicero and into the interviewer. She had just enough time for a look of resignation to pass over her face before she too vanished as the blade hit.

Cicero smiled with one side of his mouth at the camera operator who had courageously kept recording. "I apologise to my audience for the untimely end of this interview due to Sara's... suicide. Assassination attempts are such conversation stoppers. I do wish people would give up trying." His eyes lost focus for a moment, and then he grinned with unaccustomed mischief.

"I guess I owe it to you all to answer Sara's last question." Cicero concentrated on the spot next to him where she had been. A body was appearing; the interviewer was coming back. Surely even Cicero was not capable of so quickly bringing back someone who had just been killed? No, it was not Sara after all, it was just a lifeless caricature of her, but it was superb. Sara's green eyes had become smaller; her nose was not quite symmetric; her shiny hair was matt and somehow gave an impression of a tacky wig. It was clear whom it represented, but it was subtly ugly and radiated petty maliciousness. Whenever people saw her henceforth, they would think of this repulsive image. It was a possible career ender for Sara.

Cicero addressed the caricature. "Yes, Sara, I do defend the right of anyone to try to assassinate me, as a specific instance of my general policy of promoting freedom in the domain. But that does not imply that I appreciate it, and I equally defend my right to retaliate in whatever manner I consider most effective." He looked back at the camera man again, as the mock-Sara faded into nothingness. "I should provide proper attribution. I do not have the artistic talent to do that caricature. It was my daughter." Cicero then strode away to the nearest teleportal, leaving behind an audience buzzing with the revelation that Cicero, always reticent on his private life, had a daughter.

I absentmindedly looked up his public statistics. Over four thousand battles and he had only been killed once. That loss, of course, was before he was famous. Before he was the invincible, two thousand thaum celebrity. Hero. Thug. Statesman. Right wing loony. Freedom fighter. Left wing loony. Anarchist. Intellectual. Target. Opinions on Cicero varied, but no one denied his preeminent status as a target. The assassin had glowed with a seventeen hundred thaum radiance... higher than my quite respectable twelve hundred public glow, but no match for Cicero, even with the element of surprise. Nobody could match Cicero.

I fantasized idly about attempting it myself. I would lose, of course, despite my secret. I was no fighter, and had little practice in the pointless duels of the domain. The question was moot; I had no particular wish to harm Cicero, even if he was an arrogant strutter with simplistic views. Besides, I certainly did not want the inevitable publicity should I happen to somehow win.

I returned my attention to my friends. Cassandra was saying that he looked rather annoyed. "I would have thought he would be used to assassination gambits by now," I replied.

"Well yes," explained Cassandra. "I don't think it was the attempt to assassinate him per se that annoyed him. I think that he wants to be taken seriously as a statesman, and he was offended that Sara got him talking to her under false pretences. His fighting prowess has gained him fame that opens doors as a politician, but he tends to be stereotyped, and that makes it hard to sell his views. All most people care about is his invincibility."

"It does seem a bit of overkill though, that caricature," I contributed. "All she did was try to have him killed, whereas he may have destroyed her entire career. It was a bit out of proportion." There was general agreement from the coterie. Edmund demurred, asserting that you had to set an example to scare off other attempts.

Cassandra had a hypothesis. "I suspect the daughter is the key to his uncharacteristically boorish behaviour this time. I reckon that she sent him a vis of the caricature a few seconds after the attack. Young children would not like people who tried to kill their daddy; her natural perception of Sara would be warped by this dislike. She would not need to add any extra caricature to her impression – so the whole thing could be done very quickly."

"You would expect speed," said Edmund pompously, "from Cicero's daughter."

Cassandra continued, "I suspect that the whole humiliating show was more about pleasing his daughter than anything else. You have to cut dads a little slack."

"I think he is a dangerous lunatic." Roger was not one to mince his words. "Poor Sara will have trouble living down that humiliating death and disfigurement. The only reason anyone pays him any attention is his battle record, and his political platform is totally irresponsible! Besides, politicians have no business bringing their families into their public lives."

"You sound almost as critical of him as Cassy there, Roger," I said. There was no way I could resist emphasising Cassandra's rare disapproval of Cicero.

"Actually, I have some respect for his politics. I cannot say that I agree unreservedly with his platform, but at least his opinions are clearly thought out, consistent, and not just designed to be popular. But it is certainly true that I am not as impressed by his martial skills as most people." Cassandra was not into fighting, at least not in the physical sense.

"I grant you that he does at least have his heart in the right place," I grudgingly assented. "In a little safe box somewhere where no assassin will get it."

Cassandra laughed along with the others, and I felt better. Not that Cassy and I were an item, or anything... but... well to a, let's face it, rather nerdy twenty year old like me, Cicero's panache, power, charisma and reputation were rather threatening. Furthermore the attention Cassy in particular paid me was more gratifying that I wanted to admit, even if only as a friend in the domain.

Fortunately, Cicero was married. At least so he had said offhandedly a year or two ago, breaking the hearts of a million young women. However, he never actually publicly produced his wife, and after a few months the tabloids restarted their speculation. Maybe his new revelation about having a daughter would give them something else to go on about. The only other thing he had ever said about his family was that his sole loss in battle was to his younger sister when they were both kids.

Roger, the jerk, declared that I was jealous of Cicero. Naturally, he was correct, which just made the accusation more offensive. The rest of our group were mercifully quiet. Roger's standard level of tact was slightly less than that of an average seven year old, but the universe had cursed us by giving him the second highest thaumic rating in the known world. For the last three years he had been Top DOG in the Domain Open Gladiatorial contests, which Cicero did not enter. Not surprisingly, given the nature of magic in the domain, this power came with an amazing intellect and an even more impressive serve of arrogance. The psychs would have said high functioning autistic. We had other names, although we kept them to ourselves – you know what they say about people in glass houses. Cicero's mere existence, keeping Roger from the universally acknowledged first spot, was a constant irritant to him. Roger, in turn, was a constant irritant to everyone else.

"You should talk," I replied, slightly more hotly than I intended.

"I never denied my jealousy."

Roger had started our coterie, so we couldn't really just throw him out. Furthermore, he was brilliant, with the occasional insightful statement that made us willing to put up with a lot. We were that sort of group. Roger could also be quite pleasant when he put effort into trying. He did not bother very often.

If Roger was the most respected elder, I was the class clown. At least that was what I aspired to, and on my better days the others did laugh at some of my jokes. Cassy was the worldly journalist, knowing something about everything, and making deductions that were invariably accurate in retrospect. She was probably right about Cicero's reasons for annoyance and retribution.

For the umpteenth time I wondered who Cassandra really was. Like Cicero, she was a privacy freak – almost certainly operating under a pseudonym. She wore the body of a twenty year old – and a rather unremarkable one by the why-be-less-than-stunning standards of the domain - but who knew how old she was in reality? I guessed she was at least sixty. Her magic power was far too weak for her intelligence – it was implausible that she had grown up with the domain.

Tanya the quiet took advantage of the pause in the conversation to say, "I think that the assassin was from northern California."

I took the bait and asked why, even though I had a good idea what she was about to say. Unusually, I had my reasons for seeming to be ignorant on this topic.

"Look at the composition of the advertisements for the interview. It is carefully biased towards people in northern California."

"A clever use of the spectator effect," said Roger.

I checked, and indeed that was true. The only plausible explanation was that Sara had set it up that way on purpose to provide a home ground advantage to the assassin. The spectators themselves, by the mere act of observing, had weakened Cicero, unless he also was local.

"It appears that Cicero has no trouble with a fraction of a second handicap," said Edmund with satisfaction.

"Did anyone notice whether he demonstrated any latency artefacts?" asked Roger. "At least we could use it to rule out one location."

"I don't think it will be all that much use. I did notice a latency effect." It was Cassandra, of course. Some other people may conceivably have paid attention, but Cassandra was the only one of us who could have resisted the opportunity to show off by mentioning it before Tanya had brought the topic up. "The size of the delay was consistent with Cicero being physically present in almost any major population centre, worldwide."

"Even northern California," said Tanya, "if he were to realise that he had been set up, and had decided to pretend that he was further away than he actually was."

Cassandra nodded agreement, and flashed a knowing smile at me. I flushed with embarrassment. Only a few weeks ago I had tried a similar trick on Cassandra, to try to work out her real location. She had demonstrated a similar latency from several geographically dispersed locations, giving me no information. I had not even been sure whether it was a reflex camouflage on her part, or if she had guessed that I had arranged things on purpose. She had not said anything at the time, and I had hoped that she had not realised what I was up to. She did not seem angry; maybe she enjoyed the chase.

"Sorry guys" said Edmund, who had suggested we meet here and watch the interview. "I had heard a rumour that it was going to be particularly interesting. I wouldn't have called you if I knew it was just going to be some loser taking a pot-shot at the big C." Edmund was actually an interesting person when the conversation did not involve Cicero.

A few of us said our goodbyes and walked off or vanished, returning to real life. I was about to do likewise, when I remembered I had some nanotech ideas that I wanted to discuss with Cassandra.

"Got a minute, Cassandra? I've had an idea for limited self replication I want to run by you."

One eyebrow rose, quite attractively. "Sure. Here or in a room?"

"A room, if you don't mind."

She waved towards a public room with the door open, and we walked out of the plaza, away from the others. It is a shame that it is impossible to teleport freely in the domain. I do not see why one should not be able to do so, although some people go on about intuitive plausibility and suspension of disbelief and the need for a recognisable metaphor for navigating the unimaginably huge amount of data that was the domain. Recognisable? Ha! Why allow other forms of magic, then?

Some people believe that rooms were the reason that the domain was built in the first place. In the twentieth century, so I am told, communication of complex ideas was unbelievably cumbersome. People with no artistic talent would draw – slowly – pictures that in some way represented the ideas they were trying to express. Often they gave up and just spoke, hoping that waving hands wildly would convey thoughts that were too hard to express. Even that only worked if they were in physical propinquity. Computers were no help either – you interacted with them with your fingers – mechanically - as if you weren't interfaced to the domain.

I just postdate that, and I cannot imagine life without the rooms. Even just working by myself, being able to have concrete pictures appear as fast as I can think them, helps me understand my own problems better. When talking to others it is an amazing boost to productivity, which is why eighty percent of white-collar workers in the United West spend most of their working time in the domain.

If you are good at communicating your imagination to the computer, which takes practice, you can make sophisticated pictures. If you are also good at communicating your imagination to a human, you can make the right pictures. A well-recorded explanation in the domain - a "vis" - beats a book so overwhelmingly that almost all teaching and professional pre-recorded communication is now done by vis.

I am a vis author by trade. I can make quite good ones, fast. I had a vis to show Cassandra to demonstrate my new, and, in my not so humble opinion, impressive ideas for a self replicating nanobot.

A nanobot is a small robot. It is a very small robot, often too small to see with an optical microscope. Why is this useful? Well, suppose you have a bucket full of these robots, communicating with each other. You could consider the collection to be a single robot, with trillions of hands. Each hand can perform sub-microscopic manipulations, giving you spectacular precision. Furthermore, if you have a nanobot small enough to manipulate individual atoms, then you don't have to worry about purity of your input materials in many circumstances – the nanobots choose exactly the atoms they wish. The nanobot's hands never have to move far, so they get where they are going quickly, giving you great speed. Of course, each nanobot can only process a tiny quantity of material at once, but you have trillions of them in your bucket, all working quickly and precisely. You end up with industrial scale manufacturing throughput, capable of turning pretty much anything into pretty much anything else.

A generation ago they were the technology of the future. As is often the case, they still remain so. There are two main problems with nanobots.

Firstly, it is hard enough to make one to do anything. If you were clever enough to come up with a working design, there do exist machines that will manoeuvre the individual atoms to build your nanobot. At a price of a few dollars per atom, they are rather expensive to construct. This brings up the second problem. You need so many of them, that having made one, you really need to make another trillion or even more. Probably vastly more. If the entire world's gross product were to be devoted to building nanobots, we would not even have made a thimbleful in the last decade.

On the other hand, if you could make your nanobot replicate itself, then you only have to build one of the expensive critters. That seems like an advance. One trouble of course is that you still have to build the first. But I had a design...

Cassandra, like her historical namesake, was a doom crier. She particularly worried about the second, downstream problem. Suppose you have built a self replicating robot. That is fantastic. You proudly watch it reproduce. Now how do you stop it? What you have done is create life, and if it is not stopped, then your nanobot will soon turn into a very large number of nanobots, and eventually rearrange the entire earth into copies of itself. The fancy technical name for this doomsday scenario is "the grey goo catastrophe." It is not that this issue was obscure, or that it doesn't worry me – terrify me in fact - but we are so far away from building anything that could conceivably be a threat that I am more concerned about getting something that can do as much as start. Still, in deference to Cassandra's perfectly valid concerns, I have made fail-safe measures important parts of my designs, even ones that I do not intend to ever physically build.

Actually, the situation is not quite as bad as I made out there. The nanobots generally need specific elements to work with, and an energy source. This gives a good way to limit their growth, particularly the energy source. Unfortunately it is hard to make them both general enough to self replicate, and specific enough so that they can't use any sources of energy other than the ones you had designed.

We stepped into a room, and I turned on the vis. At the surface level, an enlarged nanobot appeared and started moving giant model atoms around to make a replica of itself. I could feel Cassandra studying it at another level – asking the computer for details of individual parts, investigating the conditions, the model parameters, all the notes associated with the vis. I could tell that she was seeing the practical problems. My design was not very reliable. Ninety percent of the time it would make an accurate self replica, but the other ten percent of the time something would be wrong.

There were two 'bots now, and they continued working, both duplicating themselves. They took silicon or carbon from their surroundings for construction, and a few other atoms for specialised purposes. Energy came from rearranging other carbon atoms bound in alcohols and oxygen in the air into carbon dioxide. Appropriate alcohol and silicon were easy to come by – beer and sand would do the trick. By now there were four 'bots, well on the way to eight, but one had malfunctioned. Oh well, three on the way to six. Who cares if some fail when you have exponential growth?

"So you've got replication to work, at least in your model which is a full quantum simulation." The words didn't sound very positive, but her tone was all a nervous show-off could ask for. "Wow," she added as a postscript. The timescale on the vis was speeding up, and there were over twenty functioning 'bots now.

We continued watching for another minute. There were billions of them now, enough to perceive through a small microscope. The scale had changed, and you could no longer see the individual 'bots, just a rapidly spreading iridescent goo. An electrical signal passed through the writhing mass, and they stopped replicating. Every single 'bot in the image had realised that it was supposed to stop replication.

Getting the nanobots to stop replicating was not that difficult, compared to the tour de force that made them able to replicate in the first place. The hard part was making the cut off circuitry reliable enough in the face of imperfect replication. That took skill. I grinned proudly, like a father showing off a first baby.

"May I?" Cassandra asked, staring at the vis. As I nodded, she changed some parameters in the model embedded in the vis, changing the scenario, modifying the elemental composition of the raw material the nanobot was working with.

"Triple safety stops, and that doesn't even count the reliance on alcohol for energy. You're improving, Victor." Cassandra's voice carried approval, but not total. She believed in much tougher safety measures even for virtual simulations. She probably never travelled in a car as it would be too dangerous.

She turned quiet for a while, and the vis flew. If I had not seen her use a vis this fast before I would not have believed it. I stifled a pang of envy. If I got upset when my friends were better than me at what I was best at, then I would end up with boring friends. "You have a backup?" she asked finally.

"Of course."

The vis adjusted before my eyes. A nanobot changed slightly. Five imperfections, each individually unlikely, appeared. I followed the hint of details, and was given a probability calculation. One in ten to the twentieth power or so would end up like this. It does not sound like a significant proportion, but you get a lot of nanobots per gram of goo. After a few hours, you would probably get one similar to the mutant 'bot Cassandra had designed. The vis continued. The call went out to stop replicating. Cassandra's modified 'bot ignored it. The fail safe mechanisms had all failed. Soon that one 'bot would be two, four, and eventually numbers we do not have familiar names for.

That was bad, but not fatal. They would eventually run out of alcohol. Cassandra showed another fault, this time in the energy use system. A nanobot with this fault could use other carbon sources for energy. After the other nanobots ran out of alcohol, it kept going, consuming all nearby carbon sources. Plastics, plants, animals, people, even most topsoil would contain enough carbon to let the mutant nanobot spread. The entire biosphere would be threatened. Having never encountered silicon based life, nothing living would have had a chance to evolve defences against it.

Cassandra rubbed it in with a picture of a sphere of goo the size of the earth, with a little subscript "T plus three days." It was not the entire world, of course. If you looked closer it was actually a thin covering – only a couple of centimetres average - over most of the continental plates with low concentrations in the oceans, but as far as humanity was concerned it may as well be the entire world. Equally irrelevant, it was not grey. It was an iridescent purple.

Damn she was fast and smart. The idea of looking for failure modes was of course obvious – I had spent months on it - but actually finding them was too hard for me. I had sometimes wondered whether she were a spy, and all my plans were going to some other company or, worse, political entity. But she gave at least as many ideas as she got, so that seemed unlikely. Besides, if some hypothetical competitor wanted to get ahead of me in the nanotech business, why not just hire Cassandra as an engineer rather than as a spy? She always demolished my ideas, in the nicest possible way.

"So you wouldn't recommend shares in Victor's Mini Motors to your friends for a short term gain?" I tried to make it a joke. Not very effectively.

"Crowe Engineering is my bet for short term gains right now," she said. I was not paying much attention, staring glumly at the mutant 'bot. "But I would delightedly invest in Victor Mini Motors for the long term, should you ever incorporate." My concentration returned. "Your current plan is nowhere near practical, of course, but no-one else has any better designs. The best I have seen from someone else only had double fail-safes." She shuddered just to think about it.

Was that an offer of a bribe? To invest in a company I started? If so, she offered it smoothly. It could be an excuse to get my ideas without giving as much technical information, just money? Or was she just as she appeared, a loyal friend who believed in an energetic and possibly talented young engineer? Maybe she was just saying it to be nice. No. Not that. She was pedantic in her phrasing, and I had never knowingly heard her tell a white lie.

"So what do I do?" I am an adult, adults do not wail, so it could not have been a wail, and I would prefer to not even describe it as plaintive.

"Keep trying?" she said brightly. "I'm sure there are lots of solutions. For instance, you could make a self-replicating 'bot out of Francium, and then once you have a few grams of it, use them to make a different design of 'bot – non-replicating, of course, out of some other material. Your original bots would just be an efficient factory."

Francium? I asked the domain for details. Atomic number 87. A metal. Strongly radioactive... the longest living isotope having a half-life of twenty-two minutes. Almost unknown in nature. A mutant 'bot would be limited from growing by availability of Francium, and would self destruct in a few minutes anyway in a microscopic nuclear bang. Where would you get Francium? Make it atom by atom in a cyclotron? Old Bill Gates could not afford a useful quantity of it. Probably even Arthur would shy away from that cost. Ingenious idea, in a way... but it was no more practical than my ideas, really.

I thought a bit more. The initial 'bots would not last long – on average under a second before one of their atoms decayed, probably destroying them. This was not a problem; in a second that 'bot could have made several others, recycling damaged 'bots. As long as it can reproduce faster than it dies, it does not matter if it has a short life. The second stage 'bots it made would end up being frequently radioactive themselves, which would pose a threat to whatever they in turn made. This was also OK; you could have a third strain of 'bots whose job was to filter out the radioactive 'bots of the second strain. It might work, although the design difficulty was ridiculous.

"If you wanted to cut costs at a pretty minor safety risk" she continued "you could make it on the moon instead of in some other stellar system, I guess, although I would prefer it further away. Mars would be reasonably safe." It was at this point that I knew she was joking. Surely.

"I'll keep thinking." I said, pretending to be cheerful. She smiled goodbye and ambled out of the room. I waved farewell absently, and was just putting my hand down when I froze, struck.

Do you know how sometimes somebody says something amazingly important, and you miss it because your mind was on something else? The woman who was never wrong had just given me a stock tip. I got the room to play back our conversation. Crowe Engineering. That was it. The name struck a bit of a bell, and the domain told me that it was a multi-billion dollar construction machinery supplier. How could such a boring company be a short-term stock play? Still, I had never known Cassandra to be wrong about anything. How would she know? But... she was so cautious, and so... so... accurate. I would trust her with my life. I would even trust her with a self-replicating nanobot design. If I can do that, then surely I could trust her with my money. Surely.

How much money could I raise in a hurry? I wasn't really into investing, so it took me a while to work it out. I could borrow against everything, max out credit cards. I could ask customers for advances. How would I explain why? Maybe other sources of money would be better. I had a small nest egg, mainly tied up in my house. I really did not want to lose it. I decided to trust her anyway.

I sent out successful queries to banks around the United West. Replies came through the domain seconds later. All settled. My house only nominally belonged to me now. I bought shares in Crowe Engineering. I borrowed money using those shares as collateral, and bought more shares. And more. I exploited some timing issues in settlement, and bought some more shares with money I did not really have. If their price dropped three percent, I would be bankrupt. I felt a little nausea – stock speculation was something I had never done before. A low stress life, that's the life for me. But having a bit more cash would lower stress. I trusted her.

That was enough for the day. I left the domain, climbed out of my interface, and went to bed, dreaming of the possibility that Cassandra may be wrong, or, more likely, had meant something else.