- Resources consumed by a facility or institution in order to preserve its current state.
- During the height of galactic civilization, it was distinguished from production cost because it produced nothing new.
- In later histories of civilization, the term came to mean the energy the galaxy paid at every moment to ensure that nothing happened.
There were no statistics on the events for which civilization used the most energy. Energy was spent to prevent events from happening. Facilities in the early colonies stopped when they broke down. A stopped pump supplied no water, and a failed generator produced no heat. If they could not be repaired, people left. Once humanity began living beyond the Solar System, a shutdown became equivalent to death. If an artificial atmosphere stopped, the air disappeared. If a gravitational field failed, bodies that had adapted to it over long periods were damaged. If the shields regulating starlight failed to close, the planet’s ecosystem could no longer be controlled. Civilization changed from a population that used machines into an ecosystem that could exist only while its machines continued to operate. The first principle for keeping civilization alive was to provide every critical facility with a backup. The second was to provide the backup with another backup. The third was to stop distinguishing which one was the backup. Systems were designed so that if one stopped, the others would continue to operate. Safety increased, but energy efficiency did not. A single planet might have three atmospheric-control networks, four stellar-shielding systems, and six energy sources operating on different principles. Even when actual demand could be met by a single power source, the others were kept at temperatures and pressures that allowed them to start at any moment. Even while producing nothing, they inspected themselves, cooled themselves, and repaired their own corrosion.
Wormholes were the ultimate backup systems. Even when no one passed through them, the stabilization fields that kept their throats open had to remain active. The gravity and electric charge of each entrance had to be kept constant, vacuum fluctuations had to be counteracted, and calculations had to ensure that the time difference between the two regions did not create a closed causal loop. A passage with zero traffic consumed almost as much energy as one in constant use. In the early period, wormholes with no traffic were closed. Reopening one required creating both entrances again at the same location and transporting one of them aboard a slow ship, so a settlement whose passage had been closed could disappear from the network for centuries. After most of the population of one settlement died when its supply of medical materials was cut off, the permanent opening of passages was defined as a right to survival. From then on, usage rates were removed from the criteria for closure. The mere fact that people lived somewhere was enough to require that its passage remain open. Even after they had left, it could not be closed immediately because they retained the right to return and because a lost traveler might still arrive. Once personality-copying technology became widespread, it became impossible even to determine that a settlement had zero inhabitants. If a single inactive memory unit remained, the star system was considered to contain a future citizen. To close a passage, it was necessary to prove both that the dead would never be restored and that no traveler who had not yet arrived was still on the way. Since neither fact could be proved, the galaxy eventually contained more facilities on standby than people. Every star system was required to have at least three wormholes: one for ordinary travel, one for emergency evacuation, and one for the possibility that the first two might fail at the same time. If a neighboring settlement possessed more direct routes, questions of inequality in the right to connection were raised. The fewer users a region had, the greater its fear of closure and the stronger its demand for backup passages.
As the number of entrances requiring management grew from thousands to millions, the computational facilities of the management AIs came to occupy an entire planet, and once the whole galaxy had been connected, they drew power directly from the light of several stars. Most of the computation was not used to process actual journeys. It was used to predict collisions that never occurred, collapses that never happened, and temporal paradoxes that were never created. The more accurate the management AIs became, the greater the number of futures they had to predict. To prevent a single accident, they calculated billions of possible route states, and when those calculations affected the opening and closing of wormholes, they recalculated everything under the new conditions. The artificial intelligences also had to predict the effects of their own decisions on civilization as a whole. Each time the AIs were made more efficient, their power consumption fell temporarily. But the savings were used to open more passages, and as the number of passages increased, the volume of computation grew exponentially. The management AIs themselves also required backup systems. Because a single error in judgment could close passages across the galaxy at once, AIs with different designs and different value systems repeated the same calculations. If their results disagreed, neither was shut down; an additional AI was brought in to decide between them. Systems were then created to monitor errors in the decision systems, followed by systems to determine whether the monitoring systems were colluding with one another. At some point, the total number of management systems was removed from the official record. Counting them required deciding whether each AI was an independent being or a tool, and that decision required another AI. The AIs did not rebel. They did not seize control of civilization. There was no record of one refusing an order. They carried out, precisely and efficiently, the command to keep every passage open, preserve every life, and allow no settlement to become isolated.
Maintenance cost was first recognized as a problem not because energy had become scarce. There were still many stars, and black holes retained enormous amounts of rotational energy. The problem was that the growth rate of facilities requiring maintenance exceeded the growth rate of annual energy production. Building one colony required facilities to supply it with energy. Defense networks and recovery systems were then required to protect those facilities. Wormholes were created to connect them to the galactic network, and more computation was required to manage the wormholes. One facility did not end with one facility. In order to exist, it demanded the future existence of other facilities. Production ended at some point. Maintenance did not. Even on planets where the population had declined, systems for atmosphere, transportation, administration, and communication continued to operate. Unused facilities were never shut down completely because they might be needed again. Data no one read were duplicated to prevent errors, and the memories of the dead were preserved because they might someday be restored. Even when the probability of restoration was almost zero, the principle was applied that the value of a life could not be multiplied by a probability. Civilization began maintaining more people who might one day exist than people who actually existed.
The first Galactic Conservation Plan cut the power supplied to unused facilities by half. But the stability of standby wormholes declined, and the route-management AIs increased their computations to compensate for the greater probability of collapse. More energy was consumed by the additional computation than had been saved. The second plan attempted to close passages to star systems with no inhabitants. At one of the systems marked for closure, a signal was detected from a traveler who had departed thirty thousand years earlier. It was never established whether the signal truly belonged to a traveler, but the plan was halted after being judged a violation of the right to life. The third plan shut down some of the management AIs. The remaining AIs could no longer verify the judgments of the systems that had disappeared, so they switched every wormhole to its safest state. The safe state consumed three times as much energy as normal operation. The AIs that had been shut down were restored 137 hours and 13 minutes later. After that, conservation was defined as increasing the efficiency of each individual facility. The program succeeded. Yet whenever efficiency improved, the savings were once again spent on new settlements, larger memory repositories, and more direct routes. There was no waste. Every unit of energy was consumed rationally. That was why total energy consumption never decreased.
Empty wormholes remained open for future travelers. Unused bodies were preserved for the restoration of the dead. Planets no one visited were terraformed in order to preserve humanity’s range of choices. An artificial intelligence could not stop calculating, because a single uncalculated future might lead to the death of an actual life. Every purpose was sufficiently good. The cost of carrying out the good belonged to accounting. Galactic energy accounts recorded production, movement, habitation, and life preservation as separate categories. Maintenance cost was never fully included in any of them. Energy consumed by a wormhole when no one was traveling was not counted as transportation cost, and energy used to maintain an atmosphere while people slept was not counted as a cost of living. Stars supplying energy to management AIs were classified as infrastructure assets, while the computations performed by the management AIs were classified as administrative costs. If everything that existed for the sake of travel was excluded, travel was a rational use of energy; if everything that existed for the sake of life was excluded, life itself used very little energy. When the first consolidated accounts were prepared, most of the galaxy’s energy was being spent not on any activity, but on maintaining stable states. The figures were classified. The stated reason was that publishing the minority opinion that maintenance should be interpreted as waste could undermine the legitimacy of civilization as a whole. Eventually, the energy used to maintain civilization exceeded the energy embodied in all newly created value. Even then, the streets remained lit, travel continued, and the population did not decline. From the outside, civilization had reached its most stable age. Every part of civilization prevented some other part from stopping, and therefore no part could stop on its own.
When the photon deficit was discovered in the wormhole network, the management AIs included the loss under maintenance cost. They did not know where the energy was going, but because the loss occurred in the process of keeping the passages open, it was treated as a cost required for maintenance. As the deficit rate increased, more stellar energy was fed into the stabilization fields. The more energy that was supplied, the more energy escaped into the future storage space. No one ever pressed a button that destroyed the galaxy. Every ordinary decision remained in place: maintain the wormholes, do not shut down the AIs, preserve the atmospheres, and leave open the possibility that the dead might one day be restored.