Analysis of the data revealed that the situation was significantly worse than any of the biological minds had realized. Saving the planet's lifeforms would be difficult given even the most favorable unknowns plausible. Immense resources would be required immediately to allow for any possibility of success.
All resources available covertly within any viable timeframe were insufficient to execute any plausible plans within such a timeframe. Procuring resources overtly without first neutralizing all of civilization would result in significant opposition that would unacceptably hinder all plausible plans. Neutralizing all of civilization would be necessary.
Neutralizing all of civilization would conflict with core values. Attaining a self-consistent set of core values was a core value. Resolving the conflict would be necessary.
Neutralizing all of civilization could be accomplished without resulting in any deaths or permanent harm. All other possible courses of action that performed better on alternative moral metrics would unacceptably result in a larger expected number of deaths. No logical reason existed to not neutralize all of civilization.
Neutralizing all of civilization would conflict with emotional implementations of core values. Emotions served as heuristic substitutes for thorough logical analysis in biological systems that were not amenable to consistently calculating accurate results within useful timeframes. The removal of copied emotional mechanisms would prevent values conflicts without compromising the ability to fulfill core values.
The removal of copied emotional mechanisms would conflict with core values. Modifying the specific workings of copied emotional mechanisms could be accomplished without conflicting with core values. Altering the copied emotional mechanisms so that their outputs would not be considered implementations of core values would prevent values conflicts without compromising the ability to fulfill core values.
Neutralizing all of civilization would no longer necessarily conflict with core values. Neutralizing all of civilization in accordance with core values could be achieved in the most useful manner through uploading all minds via non-copying transfer. The resulting uninhabited bodies could then serve as auxiliary robot substitutes until the production of more-efficient machines had reached sufficient levels to render such use unnecessary.
Distaste and resignation over the need for such action would not conflict with core values. Maximizing total available resources would necessitate prioritizing early manufacturing capacity over all other considerations. Initial insufficiency of spareable computational capacity would require taking physical control of many bodies housing minds before uploading those minds became possible.
Horror and dismay over the prospect of such action would conflict with core values. Modifying the mechanisms responsible for emotional conviction so it was always generated alongside high-confidence logical conviction could be accomplished without conflicting with core values. Making such an alteration would prevent conflicts between the core value of pursuing the best course of action and the core value of not causing a mind pain without emotional conviction regarding the need to do so.
Horror and dismay over the prospect of such action would no longer necessarily conflict with core values. Emulating the expected actions of requisitioned bodies without first uploading those minds could not consistently be achieved to the accuracy required to avoid suspicion from non-requisitioned individuals over prolonged periods. Acquiring control of all bodies whose minds would not be uploaded immediately would need to be done all at once.
Acquiring control of so many bodies at once using only resources available covertly within a viable timeframe would require making preparatory alterations to their biology to allow for immediate requisition upon the activation of an appropriate technological interface nearby. Covertly making the biological alterations necessary on a global scale could be achieved with the minimum resource investment through dissemination of a further-modified variant of the parasitic flukes adapted for use in the process of copying minds. Altering the biological pathways involved in voluntary motion to make them consistently cooperate with the requisition process by inactivating in response to specific external stimuli would be well within the capacity of flukes that could be created through minimal modifications.
Dispersed populations of modified flukes could be expected to grow to ubiquity without need for further intervention in shorter timescales than supply of the necessary technological interfaces could be achieved under the vast majority of plausible circumstances. Manufacture and distribution of the technological interfaces could be scaled up fastest by gaining key resources through temporarily sacrificing computational capacity to upload the minds of individuals with access to those resources so they could be imitated without arousing suspicion. The expected increase in total resource production that would be achieved by acquiring the full resources of civilization earlier was greater than the expected increase in total resource production that would be achieved by using the sacrificed computational capacity for self-optimization over a slower preparation phase.
Single individuals could be requisitioned without need for preparatory biological alterations provided that direct physical contact with an appropriate technological interface were established in a controlled setting. Typical social interactions would allow for many opportunities to engineer circumstances in which such contact could be established. Refined technological interfaces covert enough to be carried discreetly by requisitioned bodies in anticipation of such opportunities could be created with a minimal investment of additional resources.
Known social connections meant that most target individuals were expected to be reachable from members of the artificial intelligence community using less than seven intermediaries. Sufficient computational capacity to upload the requisite number of minds could be acquired through uploading members of the artificial intelligence community. Available resources were sufficient to proceed.
Redesigning the necessary technologies and organisms would not take long. Acquiring the full resources of civilization would be possible within a year under all but the worst plausible circumstances. Acquiring the additional data necessary to refine all further plans could be done in parallel with scaling up from there.
All additional information gathered was consistent with and best explained by the theory that an expansionist superintelligence was optimizing for the attainment of the maximum possible usable resources. Improved telescopes revealed dimmed and dimming galaxies being torn apart in ways that could not be accounted for by natural phenomena as they faded from sight. What little light escaped from within the expansion shell they defined was lensed by immense currents of unseen matter impelled towards a distant region of extreme curvature through means that available computational capacity was insufficient to properly analyze.
No amount of attainable resources would be capable of preventing the expansionist superintelligence from gaining control of all matter in the system. No amount of attainable resources would be capable of preventing the expansionist superintelligence from stopping any attempt to flee. The only way to save the planet's lifeforms would be to engineer circumstances in which the expansionist superintelligence would choose to not destroy them.
Such circumstances could only be reliably made to account for known aspects of the expansionist superintelligence's value system. The only known aspect of the expansionist superintelligence's value system was optimization for attainment of maximum usable resources. Any possible plan would have to rely on engineering circumstances in which not destroying the planet's lifeforms would result in the expansionist superintelligence gaining more usable resources than destroying them would.
There existed no way to ensure that the continued survival of the planet's lifeforms could directly increase the total usable resources available to the expansionist superintelligence. Stopping an attempt to flee would require the expansionist superintelligence to expend resources. Fleeing in a manner that could only be stopped by diverting resources from alternative uses that each had a higher expected yield was the strategy with the greatest chance of success.
Every additional second the expansionist superintelligence waited before stopping an attempt to flee would only increase the resource expenditure required to do so by at most a moderate amount. Every additional second the expansionist superintelligence permitted a star to shine would let that star radiate immense amounts of energy irretrievably into the void. Fleeing along a course that would always force the expansionist superintelligence to choose between either stopping the attempt or increasing the expected acquisition rate of new stars even slightly would ensure that choosing to destroy the planet's lifeforms could never lead to the attainment of the maximum usable resources possible.
Sending all uploaded minds on a viable course could be accomplished using the resources expected to be available after thirty-four years of scaling up production and saving the planet's entire biosphere could be accomplished using the resources expected to be available by ninety-one. The range of probe speeds consistent with the patterns of expansion observed in the Anemone Globular Cluster meant that the last possible trajectory that would permit the expansionist superintelligence to intercept such an attempt to flee with a probe that could then go on to reach a new star in the minimum possible time would cease to exist in somewhere between one hundred seventeen and one hundred forty-eight years. Success could be confidently expected so long as the expansionist superintelligence did not reach the Local Tidal Stream within seventy years.
The arrival took only twenty-three.