they exposed to the wood or glass. But they seldom learned that trick until they had been in the maze for several hours and through several reward cycles. Felka reached for one of the cages attached to her waist, flipping open the catch so that the contents — three white mice — spilled into the maze. Away they went, momentarily gleeful to have escaped the metal prisons. Felka waited. Sooner or later one of the mice would run into a trapdoor or flap that was connected to a delicate system of spring-loaded wooden levers. When the mouse pushed past the flap, the movement caused the levers to shift. The movement would often be transmitted across the maze, causing a shutter to open or close one or two metres away from the original trigger point. Another mouse, working its way through a remote stretch of the maze, might suddenly find its way blocked where previously it had been clear. Or the mouse might be forced to make a decision where previously none had been required, anxieties of possibility momentarily clouding its tiny rodent brain. It was quite probable that the choices of the second mouse would activate another trigger system, causing a distant reconfiguration of another part of the maze. Floating in the middle, Felka would watch it happen, the wood shifting through endless permutations, running a blind program whose agents were the mice themselves. It was fascinating enough to watch, after a fashion. But Felka was easily bored. The maze, for her, was just the start of things. She would run the maze in semi-darkness, with the UV lamp burning. The mice had genes that expressed a set of proteins that caused them to fluoresce under ultraviolet illumination. She could see them clearly through the glass, moving smudges of bright purple. Felka watched them with ardent, but perceptibly waning, fascination. The maze was entirely her invention. She had designed it and fashioned its wooden mechanisms herself. She had even tinkered with the mice to make them glow, though that had been the easy bit compared with all the fettling and filing that had been needed to get the traps and levers to work properly. For a while she even thought it had been worth it. One of the few things that could still interest Felka was emergence. On Diadem, the first world they had visited after leaving Mars in the very first near-light ship, Clavain, Galiana and she had studied a vast crystalline organism which took years to express anything resembling a single ‘thought’. Its synaptic messengers were mindless black worms, burrowing through a shifting neural network of capillary ice channels threading an ageless glacier. Clavain and Galiana had wrenched her away from the proper study of the Diadem glacier, and she had never quite forgiven them for it. Ever since, she had been drawn to similar systems, anything in which complexity emerged in an unpredictable fashion from simple elements. She had assembled countless simulations in software, but had never convinced herself that she was really capturing the essence of the problem. If complexity sprung from her systems — and it often had — she could never quite shake the sense that she had unwittingly built it in from the outset. The mice were a different approach. She had discarded the digital and embraced the analogue. The first machine she had tried building had run on water. She had been inspired by details of a prototype that she had discovered in the Mother Nest’s cybernetics archive. Centuries earlier, long before the Transenlightenment, someone had made an analogue computer which was designed to model the flow of money within an economy. The machine was all glass retorts and valves and delicately balanced see-saws. Tinted fluids represented different market pressures and financial parameters: interest rates, inflation, trade deficits. The machine sloshed and gurgled, computing ferociously difficult integral equations by the power of applied fluid mechanics. It had enchanted her. She had remade the prototype, adding a few sly refinements of her own. But though the machine had provided some amusement, she had seen only glimpses of emergent behaviour. The machine was too ruthlessly deterministic to throw up any genuine surprises. Hence the mice. They were random agents, chaos on legs. She had concocted the new machine to exploit them, using their unpredictable scurrying to nudge it from state to state. The complex systems of levers and switches, trapdoors and junctions ensured that the maze was constantly mutating, squirming through phase-space — the mind-wrenching higher-dimensional mathematical space of all possible configurations that the maze could be in. There were attractors in that phase-space, like planets and stars dimpling a sheet of space-time. When the maze fell towards one of them it would often go into a kind of orbit, oscillating around one state until something, either a build-up of instability or an external kick, sent it careering elsewhere. Usually all that was needed was to tip a new mouse into the maze. Occasionally, the maze would fall towards an attractor that caused the mice to be rewarded with more than the usual amount of food. She had been curious as to whether the mice — acting blindly, unable to knowingly co-operate with each other — would nonetheless find a way to steer the maze into the vicinity of one of those attractors. That, if it happened, would surely be a sign of emergence. It had happened, once. But that batch of mice had never repeated the trick since. Felka had tipped more mice into the system, but they had only clogged up the maze, locking it near another attractor where nothing very interesting happened. She had not completely given up on it. There were still subtleties of the maze that she did not fully understand, and until she did it would not begin to bore her. But at the back of her mind the fear was already there. She knew, beyond any doubt, that the maze could not fascinate her for very much longer. The maze clicked and clunked, like a grandfather clock winding up to strike the hour. She heard the shutterlike clicking of doors opening and closing. The details of the maze were difficult to see behind the glass, but the flow of the mice betrayed its shifting geometry well enough. ‘Felka?’ A man forced his way through the connecting throat. He floated into the room, arresting his drift with a press of fingertips against polished wood. She could see his face faintly. His bald skull was not quite the right shape. It seemed even odder in the gloom, like an elongated grey egg. She stared at it, knowing that, by rights, she should always have been able to associate that face with Remontoire. But had six or seven men of about the same physiological age entered the room, possessing the same childlike or neotenous facial features, she would not have been able to pick Remontoire out from them. It was only the fact that he had visited her recently that made her so certain it was him. ‘Hello, Remontoire.’ ‘Could we have some light, please? Or shall we talk in the other chamber?’ ‘Here will do nicely. I’m in the middle of running an experiment.’ He glanced at the glass wall. ‘Will light spoil it?’ ‘No, but then I wouldn’t be able to see the mice, would I?’ I suppose not,‘ Remontoire said thoughtfully. ’Clavain’s with me. He’ll be here in a moment.‘ ‘Oh.’ She fumbled one of the lanterns on. Turquoise light wavered uncertainly and then settled down. She studied Remontoire’s expression, doing her best to read it. Even now that she knew his identity, it was not as if his face had become a model of clarity. Its text remained hazy, full of shifting ambiguities. Even reading the commonest of expressions required an intense effort of will, like picking out constellations in a sprinkling of faint stars. Now and then, admittedly, there were occasions when her odd neural machinery managed to grasp patterns that normal people missed entirely. But for the most part she could never trust her own judgement when it came to faces.
