Academia.eduAcademia.edu
Classification of Approaches to Technological Resurrection Alexey Turchin Digital Immortality Now Foundation Science for Life Extension alexeiturchin@gmail.com Maxim Chernyakov Russian Transhumanist Movement Abstract. Death seems to be a permanent event, but there is no actual proof of its irreversibility. Here we list all known ways to resurrect the dead that do not contradict our current scientific understanding of the world. While no method is currently possible, many of those listed here may become feasible with future technological development, and it may even be possible to act now to increase their probability. The most well-known such approach to technological resurrection is cryonics. Another method is indirect mind uploading, or digital immortality, namely the preservation of data about a person to allow for future reconstruction by powerful AI. More speculative ways to immortality include combinations of future superintelligence on a galactic scale, which could use simulation to resurrect all possible people, and new physical laws, which may include time-travel or obtaining information from the past. Acausal trade with parallel worlds could help combine random resurrection and reconstruction based on known data without loss of share of worlds where I exist (known as existence measure). Quantum immortality could help to increase the probability of success for cryonics and digital immortality. There are many possible approaches to technological resurrection and thus if large-scale future technological development occurs, some form of resurrection is inevitable. Keywords: resurrection – immortality – superintelligence – mind uploading – quantum immortality – cryonics Highlights: Resurrection is reconstruction of personality based on information. The problem of personal identity will be solved in the future. If large-scale future technological development occurs, some form of resurrection of the dead is inevitable. The main player in this resurrection will be powerful AI. Combination of random mind generation and acausal trade between parallel worlds in the multiverse allows resurrection of all the dead without loss of “measure”. 1. Introduction 3 2. Death, identity and immortality 5 2.1. Definition of death depends of the current level of medical technology 5 2.2. Identity problem as a core to the resurrection 6 2.3. Non-informational identity carrier as a backdoor to immortality 6 2.4. Mathematical formalism 8 2.5. Expected future technological progress as a basis for resurrection hopes 9 3. Methods of resurrection requiring preparation now 9 3.1. Cryonics 9 3.2. Chemical fixation, plastination, and slicing 10 3.2.1. Chemical fixation 10 3.2.2. Preservation of the brain slices 11 3.3. Digital immortality based on indirect mind uploading via traces 12 3.4. Sideloading: calibrating a computer self-model while person still alive 12 3.5. Restoration from digital back-up and “evacuation” of consciousness at the moment of death 13 3.6. Live preservation of small pieces of brain tissue 13 4. Reconstruction by Superintelligent AI 13 4.1.AI is necessary for resurrection 13 4.2. Kardashev 1-level or global AI: Resurrection of cryopatients and uploading of plastinated brains 14 4.3. Kardashev 2-level AI: Jupiterian Brains powered by Dyson spheres – digital immortality resurrections via simulation 14 4.4. Kardashev 3-level or galactic-scale AI: Acausal deals with a parallel world for resurrection of partly preserved people 15 4.4.1. Resurrection via quantum random generator 15 4.4.2. Resurrection based on the expectations of peers 15 4.4.3. Resurrection based on combining expectations, random noise, and acausal trade with parallel worlds 16 4.5. Kardashev 4- and 5-level superintelligence: Omega point 17 5. New world models 17 5.1. New physical discoveries which may help humanity find new resurrection methods 18 5.1.1. Time travel: sending a nanorobot into remote past to collect all data 18 5.1.2. Reading information from the past 19 5.1.3. Observation of the Earth from the remote distance 19 5.1.4. New informational traces 19 5.1.5. Parallel worlds in which the person is still alive 19 5.2. Simulation 20 5.3. Aliens 20 6. Use of quantum immortality 21 6.1. Nature of quantum immortality 21 6.2. QI improves chances of cryonics 21 6.3. QI increases the chance that benevolent superintelligence will appear and invest in resurrection 22 6.4. Universal problem solver based on the quantum immortality 22 6.5. Eternal return 22 7. Partial resurrection 23 7.1. “Resurrectors” and their goals 23 7.2. Information without identity 23 7.3. Partial preservation of the information 24 7.4. Identity without information 24 8. Transformative resurrection 24 8.1. The main paradox of resurrection 24 8.2. Clones 24 8.3. Postmortem transformation into some other form 25 8.4. Partial resurrection with social transformation 25 9. Transcendental resurrection 27 10. From resurrection to immortality 27 10.1. Resurrection timing 27 10.2. Resurrection accuracy 27 10.3. Repeated resurrection means practical immortality 27 11. Practical steps 28 11.1. Signing a cryocontract 28 11.2. Collecting digital immortality data 28 11.3. Express a desire to be resurrected 28 Conclusion 28 Acknowledgements 29 References 29 In memory of Kathy Forth 1. Introduction The topic of resurrection is a somewhat taboo area for scientific inquiry. But if we take a rationalistic approach to the human mind and conclude that “mind” is just information, resurrection becomes a tractable and quantifiable task of reconstruction of the same information. In this case, resurrection is, in some sense, close to archeology. Different ideas of technological resurrection have a long history, but most such ideas were explored independently. For a long time, the idea of resurrection was the sole province of religion, and one of its most attractive parts, especially in Christianity. Fedorov suggested resurrection via collecting the atoms of the dead (Fedorov, 1903). Ettinger and many others have explored cryonics as an instrument for future resurrection (Ettinger & Rostand, 1965). Rothblatt (Martine Rothblatt, 2012) wrote and Kurzweil said in his interview (Berman, 2011) about digital immortality based on computer reconstruction of a person based on her informational traces, and Almond suggested the use of a quantum random generator to fill the gaps in such a resurrection (Almond, 2006). Almond also coined term “AI assisted indirect mind uploading” for digital immortality. Jones wrote a book about hypothetical quantum archeology, capable of retrieving a dead mind and sending it into the future (Jones, 2017). Bostrom wrote about future superintelligence which will create extensive past simulations, thus creating copies of people from the past and effectively resurrecting them (Bostrom, 2003). Tippler suggested that infinitely powerful superintelligence at the end of the Universe could resurrect everybody by a brute force just by simulating all possible humans (F.J Tipler, 1997). Steinhart analyzed different naturalist and digital theories of resurrection (Steinhart, 2015). Our goal in this article is to create a full list of all known methods of technological resurrection and estimate their feasibility, as well as what needs to be done now in order to eventually achieve immortality. We will not explore the questions of the need for resurrection, which directly follows from idea of badness of death (Bavelaar, 2016), nor we will look into social implications of the possible resurrection process. We will list resurrection methods starting with most technologically simple and thus close in time to us and progressing to more remote, hypothetical, or partial solutions. At the end, we will outline what can be done now to increase probability of future resurrection as well as discuss how different approaches might be successfully combined. The findings are presented in graphical form in Fig. 1. Figure 1. Overview of all ideas about resurrection 2. Death, identity and immortality 2.1. Definition of death depends of the current level of medical technology Death, from the medical point of view, is the moment at which attempts to save a life become futile. However, such a definition is somewhat circular, as the life-saving attempts are futile because the person is dead, and he is dead because resuscitation attempts are futile. This ambiguity results in constant changes of the understanding of death, as medical technology advances and people can be revived after longer periods of “clinical death”. To escape this ambiguity, the conception of “informational-theoretical death” (ITD) has been suggested (Merkle, 1992). ITD is the moment in time when information about the personality is irreversibly destroyed. However, no information can be permanently destroyed as the same information may appear again in sufficiently large universe: for example, it is impossible to destroy the number 137. Any information could be presented as one sufficiently long number and thus is similarly indestructible. 2.2. Identity problem as a core to the resurrection We could define life as the continued stream of subjective experiences, and death as the permanent end of such a stream. In that case, immortality will be a life stream without end, and resurrection the continuation of the same stream of experiences after an arbitrarily long gap. The most important in above definitions is that the resurrection depends on the sameness of the person, or the problem of the so called “personal identity”. The problem is known to be notoriously difficult. The deep analysis of the problem is out of the scope of this article, but some overview is needed. Several publications claim that the identity problem will make mind uploading difficult (Pigliucci, 2014; Swan & Howard, 2012), and most people share this view. In depth analysis of the mind uploading theory and problems is presented in (Wiley, 2014). According to Parfit, there are two types of personal identity (Parfit, 1984): “qualitative identity” which we call informational identity in this article, is the measure of similarity between two minds. “numerical identity” which we call observer identity for clarity, answers the question: will a copy of me (in the next moment of time) be me, or will it not? 2.3. Non-informational identity carrier as a backdoor to immortality In short, belief in the existence of an observer identity—the idea that “my copy is not me”—implies that informational identity is not enough for “real” identity, and that some form of non-informational identity carrier (NIIC) is needed. This NIIC could either be a “soul” or “continuity of consciousness”, but these things are currently unmeasurable. Thus, their existence (or lack thereof) is purely based on belief for now; in other words, we can’t measure observer identity. NIIC is something like an invisible chain connecting different moments of my life and making me, me. NIIC is suggested here as a negative definition, without explicitly declaring of what it could be. Several popular ideas about it exist: Soul. This is a very general philosophical category without exact definition. It is assumed that it can exist without the body and have all features of the dead human being, including appearance, memory, behavior, and emotions. It is also assumed that the soul may be immortal or may be able to forget a previous life (informational identity) and enter into a new body. There is no scientific proof that the soul exists, and even if some “ghost observations” are true, they can’t be taken as evidence of the soul’s immortality. “Immortality” implies an infinite existence, which can’t be measured. In other words, even if a “soul” exists, it could also die. The “pure” soul is a soul which is just an undeveloped set of properties, like an initial state without body or memory, similar to DNA, as it supposedly defineы some human traits. Qualia. These are subjective experiences which could be unique to every person (Chalmers, 1996). As there are no known ways to measure qualia, we can’t say if they are actually different in different people, and if the personal “feeling of self” is unique for every person. Continuity of consciousness. It has been often suggested that personal identity depends on continuity of consciousness, but it is not clear why it should, as continuity is unmeasurable (Wiley & Koene, 2015). Elsewhere, Turchin has defined human personal identity as a “complex evolving socio-biological adaptation” (Turchin, 2016). Our society ignores events where stream of consciousness is broken, e.g. during deep sleep states, narcosis, coma, clinical death, forgotten dreams, split brain after hemispheres dissection. It appears that human personal identity is robust to such events. Pure observer. This is close to idea of Atman in Hinduism (Osborne, 2014). The Atman is the same (according to the theory) for all observers, so there is no death. Several contemporary theories like “open individualism” support this idea (Kolak, 2004). The main feature of a NIIC is that by definition it can’t be measured, as NIIC is the difference between me and my absolute informational copy. Thus, we can’t know which possible features of a NIIC listed above are true, if any, except the case of a “ghost-soul”, which presumably could be measured by paraphysiological experiments; however, to date they have provided only anecdotal, non-repeatable evidence. Even if any NIIC exists, it is not an obstacle to resurrection, but an instrument. For example, if causal continuity is needed, it could be preserved separately form the human brain inside some continuous physical process. This process can continue indefinitely—some candles are still preserving fire of the Hiroshima explosion (Kishikami, 2010). In other word, if NIIC exist, then death is not death, as NIIC could survive it, like soul survives death in mythology. This, if no “soul” exist, resurrection is possible via information preservation; if soul exist, resurrection is possible via returning of the “soul” into the new body. But some forms of NIIC are also very fragile and mortal, like continuity. The problem of the nature of human identity could be solved by future superintelligent AI, but for now it cannot be definitively solved. This means that we should try to preserve as much identity as possible and not refuse any approaches to life extension and resurrection even if they contradict our intuitions about identity, as our notions of identity could change later. 2.4. Mathematical formalism Let’s assume that every person consists of information, In, and some non-information identity carrier, N. Each person’s observer-moment could then be notated as P(In, N). Information is changing with time, and identity is constant, so it could be recorded as P(In(t), N). If there is no such thing as non-informational identity, it reduces to P(In(t)). Now we could define main terms using this notation: To be alive means that for a given P(In(t), N), — P(In (t+1), N) exists. Death means that P(In(td), N) is the last one observer-moment, that is, no P(In(td+1), N) exists after the moment of death td. Immortality means that for any P(In(t), N) some next moment P(In(t+1), N) exists. Resurrection means that after moment of death P(In(td), N) some other person with the same information and identity, P’(In (td), N), will exist. If there is no NIIC, the task of resurrection is equal to the task of reconstructing information about the personality to some necessary level of accuracy. It is obvious that the position of every atom is superfluous, as human atoms are constantly changing and humans constantly forget things. As Turchin discussed in an article about digital immortality, “one night distance” (typical information loss during night sleep) is probably enough for regarding the person as informationally the same—or we have to think about people as different every day, which contradicts our adaptation called “human identity” (Turchin, 2018a). If we ignore NIIC, the task of resurrection is the task of correctly retrieving information about the human mind. The size of this dataset, and what should be included in it, is not currently obvious. One estimate of the size of conscious human memory are around 2.5 GB (Carrigan Jr, 2006), and most of it is pure garbage, so the unique personal identity core information could be rather small, on the order of several megabytes (as discussed in (Turchin, 2018a)). On the other hand, the full structure of the connectome is probably a description of 100 trillion connections in the brain, or approximately 107 GB of information. If we describe the locations of each atom in the brain, it will be around 1025 atoms in the brain, and describing the location, type and connection of each atom is probably around 1 MB, so the total information about a human brain can probably be stored in about 1022 GB. From the information point of view, restoring several megabytes is much simpler than 1025 megabytes. Surely, some “golden middle” exists, where all needed information is extracted, and all random information is ignored. This golden middle is probably somewhere between 1 GB and 1 PB, as it is above level of full conscious memory but below level of full connectome data, which is probably redundant. 2.5. Expected future technological progress as a basis for resurrection hopes Technological resurrection will become possible only if significant technological progress occurs. Such progress should happen in the following spheres: Artificial intelligence. It is assumed by many that progress in AI will reach a point where AI will be capable of self-improvement, and thus will increase its own power many orders of magnitude, becoming superintelligence (Bostrom, 2014; Yampolskiy, 2014; Yudkowsky 2008). Superintelligent AI will likely start space exploration, colonizing the Solar system and creating computers called Jupiter brains (Sandberg, 1999) or Matrjoshka brains (Bradbury, 2001) with an estimated power of 10E50 flops. This type of AI could be called “Kardashev 2 AI”, based on the Kardashev scale of extraterrestrial civilizations (Kardashev, 1985), as it will use the energy of the whole Solar system. It will likely later become galactic-size AI, equivalent to Kardashev 3 civilization, and finally reach the size of the visible universe, at which point it would probably exist until it reaches the Omega point, where it will perform an infinite number of calculations. Each Kardashev level of AI means total calculation power growth of at least 20 orders of magnitude—10 orders because of the growth of the size, and 10 orders because longer time of calculation compared to the time of existence of the previous AI’s level. Superintelligent AI will also enable progress in the other technologies listed below. Nanotechnology. It is also assumed that in the future atom-level precession nanotech will appear and enable body repair, as well as reading of data from human remains and from frozen brains and creating exact copies of the objects (Drexler, 1986). Space-time physics. Some new ways to manipulate physical reality may appear. They may make possible time travel, reading data from the past, or travel to parallel worlds. However, this can’t be taken for granted, as we can’t predict future discoveries of unknown physical laws. Biology of human brain. Obviously, better understanding of what is actually going on in the human brain is needed, and a combination of nanotech and AI will provide better instruments to look into the brain. Philosophy of mind. This is needed to better understand the nature of the personal identity, which will have a large practical impact on what kind of data should be preserved. 3. Methods of resurrection requiring preparation now In this section, we will overview methods of resurrection for which a living person will need to prepare. 3.1. Cryonics Cryonics is the practice of preserving the human body after death at cold temperatures until new technologies appear, as cold temperatures are known to slow down chemical processes. At temperatures near that of liquid nitrogen (-196 C) biological tissue could be preserved for millions of years (Merkle, 1992). Human embryos have been successfully preserved for up to 24 years (Barr, 2017), but preservation of whole body is difficult because of the water-ice problem; freezing damages cell membranes. To prevent this, different technologies of vitrification—the process in which water is cooled without creating crystal structures—are used, which include adding cryoprotectors to the blood vessel of the dead person, preventing ice formation; however, these compounds may be cytotoxic. Thus, contemporary cryonics is close to chemical fixation. At first it was expected that frozen tissues will be controllably thawed in the future and return to life, but now this seems less likely because if the amount of damage they suffer in the freezing process (Rothblatt, 2007). Even so, if NIIC is somehow connected with the “sameness” of atoms, which is highly doubtful, classical cryonics will preserve such identity. For resurrection after cryopreservation, an extensive restoration process should be performed on the cryopreserved brain to repair all damage. One potential method is the use of nanorobots, which will be able to penetrate into the brain and fix damage at the cellular level. Another option is to use the cryopreserved brain only as a source of information about the human mind. In that case, the mind will be scanned and information from it uploaded into a computer. But in that case NIIC supposedly connected with the brain would be lost, which may discourage some. There are several possible approaches to informational scanning: Brain slicing and photography. This approach is already in use now for brain research. Noninvasive tomography. A frozen brain could be photographed by X-rays from different angles, and its internal structure reconstructed by powerful computer algorithms. Use of atomically precise manufacturing to disassemble the brain slice by slice and later scan these slices. Cryonics could greatly benefit if cryothanasia (Minerva & Sandberg, 2017) is legal. Cryothanasia is euthanasia combined with cryopreservation; surprisingly, these two pieces are not legal in the same jurisdictions, as euthanasia is mostly legal in Europe (and in some jurisdictions within the US) and cryonics in the US. 3.2. Chemical fixation, plastination, and slicing 3.2.1. Chemical fixation As we have seen above, cryonics already uses some type of chemical fixation, and it is expected that the brain will be sliced for scanning in the future. Thus, cryonics becomes only one of several instruments of data preservation. The next logical step is to move directly to chemical fixation and/or brain slicing immediately after death. The Brain Preservation Foundation recently awarded a prize for “Aldehyde-Stabilized Cryopreservation” (Brain Preservation Foundation, 2018); however, it still requires cryopreservation. They later formed a startup, Netcome, with the aim of preserving brains via a combination of chemical preservation and cooling for a price of 10 000 USD. However, their initial business plan was not clear and they received bad press (Letzter, n.d.). Another competitor for the prize suggested a non-cryogenic solution of high quality (Mikula, 2016). There are several types of chemical fixations with different prices and which cause different levels of damage to the brain. The simplest and cheapest is soaking the brain in formalin or alcohol. There are two advantages of plastination: Legal – A brain is not treated like a body in most jurisdictions, but as a tissue sample. Tissue samples have been donated for scientific research since the 19th century, and thus the plastination of a brain avoids many of the legal difficulties of cryonics. Logistical – A plastinated brain could be preserved for a long time without any human attention, as constant addition of liquid nitrogen is not needed. Also, there is no risk of additional damage because of cold cracks, liquid oxygen accumulation, or crystal formation. The brain could be preserved in a home or buried in a time capsule. Cryogenic vessels are potentially more dangerous and should not be kept in homes. However, cryonics has been more popular historically, and the only startup suggesting brain plastination is young and controversial. The quickest known plan for potential resurrection is brain donation for chemical preservation followed by slicing in order to upload a connectome, and modelling of this connectome in a supercomputer. There are several projects working in this direction, and despite many technological difficulties, there are no obvious unsurmountable difficulties inherent to such projects. The first uploaded (and thus resurrected) person could appear around 2030 if AI will evolve very quickly (Turchin, 2017). Donating one’s own brain for medical research could be a method of personal uploading. 3.2.2. Preservation of the brain slices Lenin’s brain was sliced after his death for future study and in hope of future resurrection. The resulting 34 000 slices are (probably) still preserved in the Brain Institute in Moscow, inside what was the “Brain Pantheon” of Soviet leaders (Neumeyer, 2014). A lot of human brains were chemically preserved for scientific research starting from the 19th century, typically using formalin. Unfortunately, these types of chemicals seem to damage brains structures like the connectome and synapses. Slicing is damaging to the brain, but it also helps quickly deliver fixatives to the cells without use of the patient's circulation system. Currently, slices of brain tissues 1 mm thick could be successfully preserved via diffusion of fixative chemicals (Gwern, 2017). Thus, by combining slicing and cheap chemical fixation, cheaper and better brain tissue preservation could be achieved than by chemical fixation without slicing. Very thin diamond knifes are needed, or brain penetrating needles. Full body perfusion with fixatives requires advanced professionals and equipment and may also take a longer time; thus, it can’t be done for everybody, especially in the case of unexpected death. But brain removal and placement in fixing solution could be done in almost every morgue with price of ingredients starting from only a few dollars. 3.3. Digital immortality based on indirect mind uploading via traces DI has two meanings. One is direct mind uploading, which would require a connection to the brain of a living person, and is not a resurrection technology per se. Another is indirect mind uploading (Almond, 2003), or reconstruction of the personality based on her informational footprint in life (Rothblatt, 2012). For such reconstruction to become possible, powerful AI will be needed. Constant growth of information technologies gives us the opportunity to cheaply record a lot of personal data, including constant audio and video, and all of our behavior in the internet. But this is only passive recording. DI also include idea of active recording, which is the conscious act of self-description similar to the Confessions by Rousseau (Rousseau, 1782). It also includes running tests on oneself, recording memories, the practice of active imagination during EEG sessions, dream recordings, etc. (Volpicelli, 2016). Turchin has suggested a protocol for digital immortality that consists of passive constant recording and active self-description (Turchin, 2018a). Regardless of the exact protocol, collecting all of these data could easily require tens of terabytes of storage, and one also needs to preserve it for the long term. There are several methods for such preservation, including the Internet Archive; undergrounds vaults with eternal disks, like M-disks; and preservation on Moon, as will soon be provided by organizations like Arch Mission. 3.4. Sideloading: calibrating a computer self-model while person still alive The idea of “sideloading” is that the model of the brain is created while the modeled brain is still alive and their interaction used to calibrate the model’s behavior. The model is started from some blank human mind model, which is then calibrated to mimic the original. This concept was first explored in the novel Zendegi (Egan, 2011). Sideloading is currently impossible, but it may become possible in a few years when chat bots and robotic brains advance further. Some have already try to sideload their dying parents via chatbots (Alestig, 2018). We could also introduce the idea or “reverse sideloading” where a human biological clone of a person is programmed to have the same mind as an original person. It could be done via education or some nanoprobes, which translates information from computer to the biological body that is why we call it “reverse”. 3.5. Restoration from digital back-up and “evacuation” of consciousness at the moment of death We don’t yet know how to directly create digital backup of the human mind, but one idea to reach direct uploading is gradual evolution of neural implants based on nanotechnology (Urban, 2017) or of genetically modified neurons. If nanorobots which could gradually replace neurons of the brains or connect to them to read their state were created, these nanorobots could survive biological death and “evacuate” consciousness from the brain. 3.6. Live preservation of small pieces of brain tissue There have been suggestions that the claustrum (Reardon, 2017), hypothalamus, or even a single neuron is the neural correlate of consciousness (Argonov, 2012). If only a small part of the brain is responsible for consciousness, then only that part should be preserved. A small piece of the brain tissue could be extracted from the dead brain and be preserved alive by implantation into other organism or by feeding it a liquid solution. Such a preserved part will carry only a small portion of a person’s memories, but the whole personal identity. Thus, NIIC consisting of physical continuity of consciousness will be preserved. HeLa cell lines continue to be alive for more than 70 years, but it is unlikely that it will be possible to preserve even a small piece the brain tissue alive for so long. Smaller pieces of tissue could potentially be successfully cryopreserved in the same way human embryos are; such small objects can be frozen so quickly that no ice crystals form. There are no known experiments that have succeeded at keeping parts of the human brain alive after the death of the rest of the body. The brain of a guinea pig was preserved alive for several days perfused with oxygenated fluid (Mühlethaler, Curtis, Walton, & Llinas, 1993). Separate rat cortical neurons have survived in vitro for 15 days (Ruardij, Goedbloed, & Rutten, 2003). 4. Reconstruction by Superintelligent AI 4.1.AI is necessary for resurrection AI is key for almost all methods of resurrection, but different levels of AI will provide different instruments for resurrection. The importance of AI to the process also means that almost all resurrection technologies lead to the same end point: upload into a computer. However, this may not be the last step, as information about a mind may be downloaded into a new biological body via advanced nanotech or biotech. Thus, if biological neurons are required for experiencing qualia, they could be recreated and calibrated in the almost the same way as they were in the original mind. A resurrected person will have conscious experiences, but whether she would have the same NIIC is not clear, but it seems likely that there will be ways to ensure it. Some religious people have issues with the concept of technological resurrection: as they believe that the soul is immortal and is in heaven after the death of the body, they think the creation of a copy of a person will either produce a soulless zombie or will be an attempt to “extract” the soul from “God’s hands” and from “paradise”. They may consider this a sin against God and the soul’s freedom. This line of reasoning is most likely just a cognitive dissonance between different belief systems. The main difference between the two descriptions is the direction of the arrow of time: God is at the beginning of time and superintelligent AI is at the end. This consideration may help to bridge these two views. 4.2. Kardashev 1-level or global AI: Resurrection of cryopatients and uploading of plastinated brains As soon as first superintelligent AI appears and create nanotechnology (Yudkowsky 2008), it will be able to return cryopatients to life as well as scan and upload plastinated brains. But this AI’s power may be not enough for proper digital immortality reconstruction as its implementation would require many calculations and modelling of the past. This stage will be relatively short and may take only a few years. After an AI starts space travelling it will quickly grow toward the Kardashev 2 level. The AI will explore the solar system, and could start to build “small” astro-engineering objects (Bradbury, 2001), which it could use for much more powerful computing. It would likely also start to send von Neumann probes to other stars and even other galaxies (Armstrong & Sandberg, 2013). The AI should quickly start “archeology” on Earth, as much remaining information will be decaying and should be collected as soon as possible. 4.3. Kardashev 2-level AI: Jupiterian Brains powered by Dyson spheres – digital immortality resurrections via simulation At this level, AI uses a large part of the energy of the Sun (probably collected by a Dyson sphere) to power enormous calculation engines. One hypothetical example is a so-called Jupiterian brain: a whole planet converted into a computer via some form of nanotechnology (Sandberg, 1999). However, construction of a Dyson sphere in the solar system should not affect the Earth, as it would be a place where data about most people who have lived is preserved. On this level, reconstruction of most people who properly collected digital immortality information becomes possible. However, few people now collect data for digital immortality in the most effective way; as a result, AI will only be able to resurrect those who lived just prior to before its creation and “enjoyed” ubiquitous surveillance. But if such an AI sought resurrection of the dead, it could use “resurrection-simulation”, a simulation of the whole history of the humankind based on all available traces, which uses these data as reference points to calibrate the simulation. It could provide a good result for people who lived recently, including those who died in the 21st century, but as uncertainty about the past grows exponentially due to data decay and the absence of “big data” in the past, there will be some time before which such a resurrection simulation is insufficient. The main idea of a resurrection-simulation is that if one takes the DNA of a past person and subjects it to the same developmental condition, as well as correcting the development based on some known outcomes, it is possible to create a model of a past person which is very close to the original. DNA samples of most people who lived in past 1 to 2 centuries could be extracted via global archeology. After the moment of death, the simulated person is moved into some form of the afterlife, perhaps similar to his religious expectations, where he meets his relatives. 4.4. Kardashev 3-level or galactic-scale AI: Acausal deals with a parallel world for resurrection of partly preserved people The use of the von Neumann probes will allow future AI to colonize the Milky Way galaxy in about 100 000 to 1 million years, depending of the speed of the probes (Armstrong & Sandberg, 2013). This galaxy has 100 billion stars, whose energy could be used for calculations by creating enormous galactic-size populations of Dyson spheres (Zackrisson, Calissendorff, Asadi, & Nyholm, 2015). This will provide colossal, but not infinite amounts of calculation power, which could be used to reconstruct the past. 4.4.1. Resurrection via quantum random generator Almond suggested the following idea about resurrection of all of humanity: the use of a quantum random generator, which create a random mind within a computer (Almond, 2006). If the many-worlds interpretation of quantum mechanics is true, when all possible minds will appear in different timelines starting from the moment of random mind creation, which would mean resurrection of everyone from his own point of view. However, this approach will not help an outside observer, who wants to resurrect a relative, for instance, as the observe would see only a random mind. This could be called immortality-for-self (more on which is found below). This approach is computationally simple as it requires creation of just one mind simulation. 4.4.2. Resurrection based on the expectations of peers On the other side, it is possible to use the expectations of relatives as the only resurrection criterion. Any mind which complies with the expectations would suffice. This could be called immortality-for-others. Immortality-for-self and immortality-for-others could be reached via different technological procedures, both of which are much simpler then solving the problem of full resurrection. Resurrection based on expectations is also rather computationally simple, as most expectations are simple descriptions of the expected behavior, and something like generative-adversarial neural networks could be trained to satisfy any formalized set of expectations. There is also no problem of loss of information about the person, because if some expectations are lost, it would mean the task of resurrection was simplified. Reconstruction based on expectations raises the ethical question of identity. For example, a parent would like to see exactly her child, but would not just accept a child who looks exactly like her own. This problem will be addressed and partially solved in the next section. 4.4.3. Resurrection based on combining expectations, random noise, and acausal trade with parallel worlds The ideal method of resurrection would be a combination of the factors listed above; quantum randomness resurrection and resurrection based on expectations do not increase the computational complexity of either task. Future AI could use all sets of expectations and other available data to recreate the known portions of a personality and then use random quantum noise to fill in for unknown portions. Surprisingly, any personality created by the combination described above will be the resurrection of some personality that has existed, even if it didn’t exist in our world! Any random person will be a resurrection of an actual person who lived in a parallel world. Such parallel worlds should exist if we accept the many-worlds interpretation of quantum mechanics. As a result, so called “measure” of personal existence (which is a share of all possible worlds where one exists), will not decline after resurrection using quantum randomness. For example, we know that someone’s name was John S—. His last name was either Smith or Simpson. We create a model of John S— and use a quantum generator to choose between either Smith or Simpson. In half of all possible worlds we will get Smith, and in the other half we’ll get Simpson. If the actual name was Smith, this means that the “measure” of Smith declines by half. Some scientists think that the decline in measure should be counted as a lower level of existence, so our resurrection was not full (Mallah, 2009). In any real case of many random parameters relating to personality, it appears that the measure will decline by many orders of magnitude, because of the combinatorial explosion of possible combinations of parameters. While the question of the “measure decline problem” for resurrection is debatable, it seems that it would be better if measure will not decline. However, there is no decline of measure. If we look at a broader picture, there is another possible world, where John S— was named Simpson, and this is the only difference. In this world, where another AI which will try to recreate John S—, also by using a quantum random generator to decide his full name, which will give “Simpson” half of the time. However, if we combine both worlds, at the beginning we have one Smith world and one Simpson world, and at the end we will have four worlds: two with a half measure of Simpson, and two with a half measure of Smith. Thus, the total measures of Smith and Simpson will not change. While these worlds can’t communicate, they could know about each others’ existence from consideration of the general world model and use it for the type of cooperation called an “acausal deal”, one example of which is the thought experiment of “counterfactual mugging” (Nesov, 2009). In other words, if an AI creates just one random quantum generator to model a human mind, it may expect that an astronomically large number of such generators are created in other worlds. Even if the many-worlds interpretation (MWI) of quantum mechanics is not true, there are other ways in which the universe could be infinite and include many parallel worlds that would justify the use of randomness to fill gaps in the resurrected mind without fear that the original will never be resurrected. The suggested combination of expectation and randomness is a computationally simple resurrection method. 4.5. Kardashev 4- and 5-level superintelligence: Omega point The Kardashev 4 level involves control of entire visible Universe (1010 galaxies) and Kardashev 5 AI is assumed to capable of acting of the scale of the multiverse. This definition of the multiverse could include the 10500 possible worlds of the string landscape, plus eternal inflation conditions near the Big Bang, plus Everett’s full multiverse (Tegmark, 2009). A Kardashev 6 level has also been suggested, which would exist outside space-time and be able create universes and would be generally undistinguishable from God (Sicoe, 2014). Surely, there could be even higher levels which we can’t imagine based on our current knowledge, especially if the “mathematical universe” hypothesis (Tegmark, 2014) of world origin is true, or if multilevel nested simulations are possible (Torres, 2014). For example, the direction of evolution could be miniaturization in the directions of nanotech, femtotech, and space-time foam (Smart, 2012). It is currently unknown if it is possible to escape the end of the universe, and it is even unknown what kind of the end it will be, but there are many ideas how such an escape could be possible (Dvorsky, 2015). If there is no escape, AI will be limited to the observed universe, where it still could have astronomically large calculations in scenario of heat death (Armstrong & Sandberg, 2013). On this level, resurrection becomes possible even without much data about humanity (F.J Tipler, 1997). This could be achieved by running all possible past simulations of all human (and not necessary human) histories. Tipler shows that the upper bound of all possible humans is 1010E70. This is a very large number, and the computational power of one visible universe is not enough to reach it. However, if some data about Earth history and limitations of accuracy of the simulation are applied, the number could be much smaller and probably achievable by the capabilities of a Kardashev 4 AI. 5. New world models Approaches to resurrection depend on the world model. When people believed in animalistic spirits, they thought the dead would also turn into such spirits. When people believed in an omnipotent God, they expected it would create a special, separate world for souls to live in after death. The possible types of resurrection listed above are based on an understanding of the world based on physics and biology at the start of the 21st century. They do not require any new fundamental discoveries about space, time, and consciousness, which may or may be made in the future. While we can’t count on future discoveries, we also can’t exclude them, as history has shown many dramatic changes of the world model are possible. 5.1. New physical discoveries which may help humanity find new resurrection methods 5.1.1. Time travel: sending a nanorobot into remote past to collect all data If there will at some point be technology that allows travel to the past, then our future descendants will be able to directly save people dying in the past by collecting their brains at the moment of death and replacing them with replicas. However, most currently existing ideas of time-travel require some high-energy exotic matter, like wormholes the size of a large black hole, which would destroy everything around it. But if such time-travel is possible, a nanorobot could be sent several billion years before now, where it could secretly replicate and sow nanotech within all living being without affecting the course of history. At the moment of death, such nanorobots could be activated to collect data about the brain and preserve it somewhere until its future resurrection; thus, there would be no need for forward time travel. Sending just one nanorobot to the past seems to be a much more plausible idea if any time-travel is possible. Even if it requires some exotic matter, like supermassive black holes or the Big Bang itself, this would not be a problem, as such events have happened in the past, and a nanorobot from the past could replicate and travel to the Earth via von Neumann probes (self-replicating starships). However, a more serious problem is damage to the causal consistency of the time line, known as the “butterfly effect”, and its extension, the “grandfather-killing paradox” (Lewis, 1976). Even if a nanorobot tries to minimize its effects of the course of history, such effects will inevitably occur, and neither human civilization nor future AI might appear—or at least it might be completely different, and thus the point of the collecting information about the past would vanish. However, a trick similar to one described above (Section 4.4.3) could be implemented: acausal trade between parallel worlds. Humanity could send its nanorobot back but collect information from nanorobots which were sent to the past by different civilizations in a parallel time line. Here small changes in the past would play the role of a random quantum generator. Each nanorobot may be equipped with a program that tells it not to provide its information until the receiver civilization agrees it will not send its own nanorobot back, ensuring cooperation in this analogue of the prisoner’s dilemma. Another way to avoid timeline change is the idea that loops in the timeline could form a strange attractor. From an observer’s point of view, this would look like a stable timeline not affected by the butterfly effect. This will happen because loops where there is no future effect preventing sending the probe back, that is, there is no grandfather-killing paradox, will become infinitely thicker than unstable loops. In other words, the measure of loops where time travel can’t affect the future in the way that prevent sending the time probe back will be much higher. The only thing we need from the past is information (plus NIIC, if any exist), and sending a nanorobot to the past is a method intended for information collection. However, it could potentially preserve some forms of NIIC, for example, if it preserves continuity of consciousness by secretly and continually uploading human minds just before death. If NIIC is the quantum state of a single neuron, it could also be preserved. 5.1.2. Reading information from the past The ideal solution would involve receiving only information about the mind (and NIIC) from the past without sending any probes to the past; this seems less contradictory to the known laws of physics. One idea which has been suggested for retrieving information from the past is so-called “quantum archeology”. The idea is that some of the properties of quantum mechanics (QM) could be used to read data from the past. One such idea is based on “retrocausality” (Price, 2008), which is a timeless interpretation of quantum mechanics, where different non-locality paradoxes are explained by the supposed ability of quantum particles communicate backwards in time. However, the current understanding of QM doesn’t allow sending of any meaningful data via quantum teleportation, which only synchronizes random effects, like the spin of two particles in different locations. Nevertheless, some clever experiments like non-explosive bomb testing (Elitzur & Vaidman, 1993)—not exactly this approach, but a similar one—may help read the data of the past. 5.1.3. Observation of the Earth from the remote distance As the speed of light is finite and the universe is mostly empty, a hypothetical large telescope could be able to observe past events on Earth. However, to get to where the human past would be visible, superluminal travel and advanced astro-engineering would be necessary. 5.1.4. New informational traces Even reading information from the past is not needed if a new type of information traces is found. In general, the growth of entropy in the universe means that no information is lost, it just becomes impossible to decipher (Hawking, 2014). Every action has many consequences, and the whole Earth is the result of history; each atom’s location is determined by many past influences. Moreover, science constantly finds new ways to extract data about the past, like DNA from bones and the isotopic composition of hair samples which reveal what a person had ate (Ungar & Sponheimer, 2011). Such traces may include subtle chemical changes, effects on atomic location inside solid objects in our surroundings, DNA and its epigenetic changes, as well as changes in unknown or hypothetical types of particles, like dark matter. 5.1.5. Parallel worlds in which the person is still alive If the many-worlds interpretation of quantum mechanics is true (Deutsch, 2002; Higgo, 1998), then a person who died in our world didn’t die in the same way in some other branch of the multiverse and could be still alive there. If it is impossible to go back in time, it may still be possible to communicate with other branches of the multiverse that have the same “timestamp” as ours and branched recently. We will not speculate here about how such communication could happen; it seems likely that it is not completely forbidden, as shown in experiments like bomb fuse testing (Elitzur & Vaidman, 1993), but that it becomes extremely different after branches become decoherent. Some theories explains quantum mechanics through such interactions (Hall, Deckert, & Wiseman, 2014). Surely, it would be morally bad to “steal” a person from another branch where she is alive and happy. However, communicating with another branch may help to extract information about the personality which is not available in our branch, and may be possible via a communication method such as an “interdimensional telephone”. However, it seems that this resurrection approach could be used for only a short time after a person’s death, as people tend to die of aging, and the person of interest would soon be dead in almost all parallel timelines, probably before technology to travel to other timelines could appear. But at least in one timeline she could be cryopreserved, maybe after some improbable chain of events, like falling through ice; this information could be used for her reconstruction in all timelines that are able to communicate with the information-preserving timeline. 5.2. Simulation According to Bostrom, if future superintelligence is possible, we are likely to already be living in a computer simulation (Bostrom, 2003). If we live in a simulation, it runs on an enormous computer, which most likely have records of everything interesting which has happened in the simulation. If ways to access such data could be found, the data could be extracted and used for resurrections—either by us, or by the owners of the simulation. Maybe we are already in the reselection-simulation discussed above, which only proposes to relive past history and recreate all of the dead more or less accurately. Some small share of all simulations may simulate an “afterlife” according to some religious beliefs. It is also possible that simulation is something akin a computer game for the beings living the future civilization, and after “death” here, they remember who they actually are and either return to their “world or play a new game. If we are in a given simulation, it is also likely that we are in the multilevel nested simulation (Torres, 2014). Data about all previous levels is probably stored at each level of such a simulation; thus, human personal data survives on many levels and could be restored in at least one of them. 5.3. Aliens If extraterrestrial intelligence exists, either on other planets or in other dimensions, it or they could collect data about the Earth and even save people from death, if they are secretly present in the solar system. This is called the “Zoo hypothesis” (Ball, 1973). If aliens collect such data, they could provide us access to it after we contact them and thus help us to resurrect our dead. More about alien superintelligence can be found in Section 6.3. 6. Use of quantum immortality 6.1. Nature of quantum immortality The main idea of quantum immortality (QI) is that our world is constantly splitting, according to Everett’s MWI, and thus in a choice between life and death, there will be always at least one timeline in which I will not die (Almond, 2008b; Higgo, 1998). The idea is controversial, first, because it is not known if MWI is true, and second, many argue that surviving in a small probability timeline can’t be counted as true immortality for two main reasons: firstly, the “measure” of existence will decay quickly, and thus from a decision theory point of view such timelines should be ignored, and secondly, even if one survives, such survival will be “bad” as it will be in a maximally damaged state of consciousness (Aranyosi, 2012). All of these objections are questionable for the reasons previously explored by Turchin (2018b). For example, despite being called “quantum immortality”, the idea doesn’t require any quantum effects, only the fact that the universe is infinitely large. This could be true in different ways (Tegmark, 2009). In an infinitely large universe there will be always copies of me in other supercivilizations which experiment with resurrection, so they will resurrect some being identical to me. The existence of QI helps survival, but it is not resurrection. In this section, we will explore ideas relating to the use of QI as a method for amplification of other resurrection technologies. Resurrection using quantum generation of random minds, as discussed above, takes advantage of QI. 6.2. QI improves chances of cryonics QI significantly increases the chances of successful resurrection via cryonics. If QI is true, the biggest share of timelines where I survive until 2100 include cryopreservation (Randall, 2004). For example, if, in the normal world, cryopreservation’s chances are 0.1 per cent, and my chances to live until the year 2100 naturally is 1 in a million (here we exclude the effects of new life extension technologies and only look at the current human life expectancy distribution), it means that I have a 1000 times greater chance to survive to 2100 because of cryonics compared with my personal longevity. Because of QI, the success probability of cryopreservation grows to 50 percent (our estimation that QI will work), in other words, 500 times. No matter how small the chances of the success of cryonics, QI increases them to the level of its own probability of success. In the case of QI, if you sign up for cryonics the multiverse will do the rest. Signing up for cryonics replaces the default outcome of the “bad immortality” of infinitely long aging, with a good “resurrection” outcome. However, if you want to ensure that your friends also survive, you will still need to invest in the quality and probability of the global success of cryotechnologies. 6.3. QI increases the chance that benevolent superintelligence will appear and invest in resurrection The same way in which QI increases the chances of success with cryonics, it also increases the chances that there will be a timeline in which humanity will not go extinct, and, further, that humanity will create a benevolent superintelligence interested in the resurrection of the dead. Not everybody agrees that benevolent superintelligence should resurrect the dead; for example, because it may create more “pure good” by creating happy people de novo—but we are interested in timelines in which AI has a subgoal to resurrect the dead. Also, non-human superintelligence created by aliens in a causally disconnected part of the universe may create many different simulations of possible civilizations and randomly create one very similar to our civilization and thus “resurrect” people who lived on Earth. Given the infinite size of the universe, there will be infinitely many superintelligences, each creating many random past simulations of possible planets. Thus, it is likely that our planet and any particular person will be simulated, and moreover, that there simulated the event of a particular person’s resurrection after death will occur. This, however, requires a two-level simulation: super-superintelligence to simulate a superintelligence which resurrects the dead. 6.4. Universal problem solver based on the quantum immortality There is also a controversial idea, which we do not endorse or recommend, that a quantum suicide experiment may be used to create something akin to a universal problem solver (Almond, 2008a). Imagine a machine which creates measure condition A, and if condition A is not met, it kills me: I will survive only in worlds where condition A is met. If we use A as a condition for person X’s resurrection, then the observer will find himself only in the world in which person X is alive. This could obviously can go wrong in many ways, as the condition may be very improbable. In addition, if many people use this idea, the world would soon be empty. A way to escape some of the difficulties of this “universal problem solver” was suggested by Yudkowsky in “Anthropic trilemma” (Yudkowsky, 2009). In this thought experiment, a large number of copies of an already uploaded mind is created, and all copies then play something like the quantum suicide experiment. At the end, even in worlds where the result of the game was negative, one copy remains. As a result, the subjective probabilities of the experimenter are skewed in the direction of the desired outcome, but for the outside viewer there is no loss of life. 6.5. Eternal return According to Nietzsche, everything will happen again, if you wait for eternity (Nietzsche, 1883). According to Tipler, this effect is unlikely, based on the quantum cosmology of the 20th century (Tipler, 1980), but if universe is cyclic or if we account for the existence of other universes, “eternal return” (everyone who lived before will live again absolutely the same life) could theoretically occur, though it collapses into something like quantum immortality if we account for possible variation during each “run” of a human life. It might still be reasonable to consider this a form of resurrection, though. Also, if one exists for an infinitely long period, one will encounter any possible other beings, including those, which are arbitrarily similar to the dead person. For example, if I exist for billion years, I could meet a person with exactly the same name and appearance as the one I loved, and this will be like eternal return from the outsider point of view. 7. Partial resurrection 7.1. “Resurrectors” and their goals Partial resurrection is resurrection of only some aspects of a person. Resurrection is not its final goal per se; partial resurrection will happen according the goals of some perpetrator. This perpetrator is resurrecting the dead for some purpose, and that purpose will affect the way in which resurrection is performed. There could be several “resurrection agents” or “resurrectors”, of which we list the most prominent types: Friendly superintelligence which seeks complete resurrection of everybody based on its understanding of “good.” Future AI resurrecting people from the past instrumentally, for example, to have a perfect simulation of its past. Relatives who want their loved ones back. Such relatives may include parents, children, and lovers. They could have different levels of egocentrism connected with their desire for resurrection. Some relatives just want good for their loved ones, while others may want a person back as soon as possible because they want to communicate with that person and suffer personally without this type of communication. Probably most “lovers” will be this type of agent, and they may choose quicker but partial resurrection, as they are seeking to fill the hole in their own perception. The person himself, when the person being resurrected is a stakeholder in his own resurrection, he is likely to be most interested in identity and less interested in exact traits, excluding those that are very valuable according to his own judgement. While he can’t act as a resurrector, he could specify conditions for his future resurrection. 7.2. Information without identity An egoistic resurrector may be more interested in the behavior of the deceased person than in her “true identity”. The resurrector may use his own expectations as resurrection criteria, as discussed above in the section 4.4.2. In extreme cases, this may not be resurrection, but just a search for another person who could fill the gap. For example, a man could search a new girlfriend who he feels is similar to a previous one, but better. 7.3. Partial preservation of the information There are several distinctive cases of partial informational resurrection, which are discussed in Section 8, as they are often a case of transformative resurrection: Reconstruction based on the expectations of the relatives. Reconstruction based on expected social function. Reconstruction of behavior but not internal states (e.g. chat-bots). Reconstruction with element randomness. 7.4. Identity without information If anything like reincarnation is possible, it is an example of preserving identity without preserving information. A reincarnated person would be presumed to be the “same” even if there are no informational traces of his past life. Another example of identity preservation without information would be a person with total amnesia who learns new things. Identity could be theoretically preserved without information. For example, if a piece of brain tissue were extracted from a dying brain and kept alive, later growing into a new brain, it could be regarded as the same brain with continuity of electrical processes, despite the fact that it would have lost most of its memories. 8. Transformative resurrection 8.1. The main paradox of resurrection The main paradox of resurrection is that, at first, we must collect many small pieces of data in order to create “the same” person, but as soon as we succeed we should immediately start to improve this person so she can adapt to her new life in the new world. As a result, most of the person’s traits will be useless or even harmful. It seems that we could simplify the resurrection task if we resurrect a person who is already in an improved state. 8.2. Clones The most obvious idea of the transformative resurrection is human cloning. Obviously, clones do not preserve identity and do not preserve personal memories, but they do preserve many important personal traits from the point of view of an outside observer. Moreover, clones could be adapted to the new surroundings of the future in the most natural way: by education. In Houellebecq’s novel “The possibility of an island”, another option which combines cloning and digital immortality in the form of self-description is explored (Houellebecq, 2007). After a person’s death, a clone of that person is created and educated based on the something like an autobiography of his original; the clone thus continues his line of thought. This approach to resurrection seems unlikely, as it assumes a technologically advanced and very stable society that lacks powerful AI. A clone inside a very fine-tuned environment, almost a simulation of the past, could produce a human being very similar to the dead one. However, it seems to be ethically questionable to imprison a person in such an environment, similar to that shown in the movie “The Truman Show” (Weir, 1998). Raising a clone in the same culture and speaking the same language but not isolated from the environment seems to be more ethical approach to this particular type of resurrection. Surely, a cloned person could be also cured from some diseases or could know the typical mistakes made by his progenitor as well as potential pitfalls from the past life. It would be example of even higher transformation but less preservation of the same “identity”. The existence of clones as twins is accepted as normal by society, but not considered anything close to resurrection, despite the fact that due to embryo freezing some genetic twins could be of different ages (Telegraph, 2012), and even another one could be born after death of the first one. 8.3. Postmortem transformation into some other form Another method of transformative resurrection is transformation into something which is not a person at all. Such transformation is mostly reasonable for immortality-for-others, but it could be relatively cheap and effective, and many forms already exist: • Collecting and analyzing the person’s works in ways that are useful for other people: collected writings, catalogue-raisonne, biography, biopic movie. • One person simulating the other person: theatre, transformation during games, cosplay, psychotherapy techniques, e.g. the dialogue of voices (Stone & Stone, 2011). • Life inside the memory of another person: in dreams, in fantasies; in a more advanced case, one could even share his brain’s calculating power with a “subpersonality” modeled based on memories of the dead person. • Symbolic transformation: transformation into qualia, mood, image, drawing; something simple and stable. Interestingly, if a person concentrates all his attention at the moment of death in one simple object, then the image of this object may work as a bridge for the continuity of identity, if identity is defined through the continuity of the mental states. It may sound esoteric, but it follows from the definition of identity as continuity, which is likely false. 8.4. Partial resurrection with social transformation Social transformation is transformation of personality into some human organization which is not a human mind, but which could have goals, behavior, and values. • Foundations. Organizations that continue the projects of a person after he dies. There are several examples of successful foundations in the names of deceased artists, which own their collections, organize exhibitions of their work, and research their art. One example of a successful artist’s foundation is Henry Moore’s Foundation (Little, 2016). A good foundation can continue to work according to the dead person’s will, promote his values, work on the catalogue of his art work and eventually participate in his resurrection. However, foundations are expensive and legally complex. • Museums. Museums are buildings that house personal belongings, an archive, and could works as a research institute. Many famous people have founded personal museums, often in their own houses. • Chat-bots. In the Apple app store, one can download “Roman Mazurenko”, a memorial app based on the logs of an entrepreneur of that name who died in 2015 (Newton, 2015). Currently, such chat-bots mostly produce nonsense and may have only symbolic meaning. • Children. Children often are claimed as some form of immortality. But not everybody would agree that he is the immortality of his parents, not to mention grandparents. • Family. Some form of immortality can be achieved by identification with some large entity, most likely a family, company, or nation-state. Family can carry on traditions and have similar personal traits or even roles (Hellinger, 2001). • Monuments. Creating dedicated personalized objects is an attempt to increase social memory about a person. Street statues subconsciously affect people; society does not forget a person who is still present in their surroundings. • Temples. Most Christian temples are built around the relics of some saint and such reliquaries are dedicated to his name. A temple is the symbolic postmortem home of the person. • Legacy. A person’s legacy consists of other people’s memories; it is often said that the dead will “live in our memories”. However, memories are fragile and will die off in a few decades as their carriers also die. Such personal memories can potentially be transformed into institutionalized social memory or myth. • Relics. Preservation of the cell line and body parts or a mummy. The Christian tradition of burial of whole skeletons is based on the belief that the actual body will be resurrected after Judgment Day. A skeleton could provide a lot of information for future reconstruction, including DNA, habits, and travel history. • Values. The embodiment of values and ideas. Some people completely identify with their set of values, most likely connected with the ideology of some large project. If such a project continues, it is the most important form of immortality for them. This practice is institutionalized by giving the person’s name to the institutions with which he was connected, or which represents his values, like research labs and battle ships. • Name as symbolic reincarnation. There is a tradition in some countries to give children the names of dead relatives. 9. Transcendental resurrection Here we list some ideas about resurrection which come from the religious philosophy, and which imply some ontological change in the status of the person, which in many cases can’t be described in human language: Physical resurrection according to God’s will Confluence with God Reincarnation Afterlife Transcendence 10. From resurrection to immortality 10.1. Resurrection timing All other things being equal, it is better to be resurrected earlier. The world will not have changed much, and the resurrected individual will be able to preserve his social connections and have more opportunities to participate in future events. However, if we consider the rate of technological progress, later resurrection may happen via more powerful AI and thus be of better quality. 10.2. Resurrection accuracy NIIC implies that the new person is either “completely me” or “completely not-me”, that is, only a binary type of relationship. However, informational identity is more flexible, and there could be different levels of accuracy in the resulting reconstruction of a person. It is templating to say that less accurate reconstructions will dominate, but this is not necessarily so in a universe dominated by very powerful AIs and in an infinite universe where all possible reconstructions will appear. 10.3. Repeated resurrection means practical immortality If a person could be repeatedly resurrected, it would mean she had practical immortality. Such resurrections could be of two types: the person being repeatedly resurrected as a continuation of her first “interval in life”, or such intervals could follow one another, so she would be resurrected based on the next “backup point”. The second variant is more like real immortality, as in that case the person’s memory will continue and increase with each resurrection. In reality, future resurrection could have different forms simultaneously, which will interact with each other. For example, ancient Parthenon is “resurrected” at least 3 times simultaneously: as actual renovated building, as its copy in the National Greek museum and as original sculptures in British museum. 11. Practical steps 11.1. Signing a cryocontract It is not enough just to know about cryonics. One must actually sign a contract with a cryocompany and complete all necessary paperwork. There are three well-established cryocompanies in the world: Alcor and Cryonics Institute in the US, and KryoRus in Russia. Alcor is the most expensive but has the most sophisticated technology and legal protections. There is an option to finance a cryocontact via life insurance in Alcor (Devlin, 2016). US laws allow an almost immediate start of cryopreservation procedures after a person is pronounced dead, but in Russia, the paperwork will take many hours. In most of Europe, cryonics is illegal, but preliminarily cooled bodies could potentially be sent abroad after death (Guardian, 2006). It is also important to discuss one’s wishes with the relative or relatives who have the right to choose a method of burial, most likely spouse, parent or offspring. If cryonics become more popular, it will be cheaper, so promoting it will help. Even if cryonics is unavailable for someone, chemical preservation of the brain may provide some chance for the immortality. 11.2. Collecting digital immortality data While cryonics, which Turchin called plan B, is not available to everyone, as it is currently expensive and available only in some jurisdictions, digital immortality (Plan C) in the form of data collection data about oneself is in reach for almost everyone. All it requires is access to a modern computer, or at a bare minimum, paper and pencil. Turchin has created a “digital immortality protocol” which can be freely accessed and serves as a guide through the process (Turchin, 2018a). Briefly, digital immortality requires both passive and active data collection. Passive DI may be constant video recording of important behavior, while active DI may be writing an autobiography and collecting the results of psychological tests on oneself. 11.3. Express a desire to be resurrected Many people do not want to be resurrected, and future friendly AI may respect their wishes. Thus, to be resurrected by a friendly AI, one must express a desire for resurrection. Reasons people may not wish to not be resurrected may include a fear that after resurrection they will be prisoners of a hostile entity, like evil AI. But rejecting resurrection will only influence friendly AIs, not evil ones. Thus, by rejecting resurrection, one dramatically increases one’s relative chance of “hostile resurrection” compared with the chances of resurrection by friendly entity. Conclusion We have found that there are many different ideas about technological resurrection. Some are practical and will not require much new scientific knowledge to implement, while others are more hypothetical. The sheer number of different paths to resurrection means that at least one of them may work and resurrection becomes much more possible. However, humanity must survive multiple global catastrophic risks and create a friendly AI in order to make resurrection possible. If humanity has no future, there will either be no resurrection, or only resurrection by a hostile entity, like evil AI with a random goal (s-risk) or curious aliens. That said, the only world model there in which there is no resurrection is a small world without much future technological progress, without a multiverse, and with very fragile personal identity. Acknowledgements We would like to thank David Wood, Anastatsia Gacheva, Jonathan Jones, and Alexander Kononov for inspiration and useful comments. All possible errors are our own. References Alestig, I. (2018). New technology allows relatives to “chat with dead”- Ny teknik erbjuder anhöriga att ”chatta med döda” - dagen.se. Retrieved from http://www.dagen.se/nyheter/ny-teknik-erbjuder-anhoriga-att-chatta-med-doda-1.1104260 Almond, P. (2003). Indirect Mind Uploading: Using AI to Avoid Staying Dead. Retrieved from https://web.archive.org/web/20120310211305/http://www.paul-almond.com/IndirectMindUploading.htm Almond, P. (2006). Many-Worlds Assisted Mind Uploading: A Thought Experiment. Retrieved from https://web.archive.org/web/20110513092111/http://www.paul-almond.com/ManyWorldsAssistedMindUploading.htm Almond, P. (2008a). Civilization-Level Quantum Suicide. Retrieved from https://web.archive.org/web/20120310105846/http://www.paul-almond.com/CivilizationLevelQuantumSuicide.htm Almond, P. (2008b). Quantum Suicide and Inconsistency: Assuming you will survive quantum suicide may be simplistic. Retrieved from https://web.archive.org/web/20120309233248/http://www.paul-almond.com/QuantumSuicideInconsistency.pdf Aranyosi, I. (2012). Should we fear quantum torment? Ratio, 25(3), 249–259. Argonov, V. Y. (2012). Neural Correlate of Consciousness in a Single Electron: Radical Answer to “Quantum Theories of Consciousness.” NeuroQuantology, 10(2). Retrieved from http://neuroquantology.com/index.php/journal/article/view/548 Armstrong, S., & Sandberg, A. (2013). Eternity in six hours: intergalactic spreading of intelligent life and sharpening the Fermi paradox. Acta Astronautica, 89, 1–13. Ball, J. A. (1973). The zoo hypothesis. Icarus, 19(3), 347–349. Barr, S. (2017, December 20). Woman gives birth to a baby after embryo was frozen 24 years ago. The Independent. Retrieved from http://www.independent.co.uk/life-style/health-and-families/embryo-24-years-frozen-born-baby-longest-ever-tina-benjamin-gibson-emma-wren-nedc-tennessee-a8119776.html Bavelaar, K. T. (2016). Puzzling with nonexistence: can death be bad for the one who dies? (Master’s Thesis). Berman, J. (2011, August 10). Bringing Back His Dead Dad Via Computer. ABC News. Retrieved from http://abcnews.go.com/Technology/futurist-ray-kurzweil-bring-dead-father-back-life/story?id=14267712 Bostrom, N. (2003). Are You Living In a Computer Simulation? Published in Philosophical Quarterly (2003) Vol. 53, No. 211, Pp. 243-255. Bostrom, N. (2014). Superintelligence. Oxford: Oxford University Press. Bradbury, R. J. (2001). Matrioshka brains. Retrieved from http://www. aeiveos. com/~ bradbury/MatrioshkaBrains/MatrioshkaBrains. html Brain Preservation Foundation. (2018). Small Mammal BPF Prize Winning Announcement – The Brain Preservation Foundation. Retrieved April 21, 2018, from http://www.brainpreservation.org/small-mammal-announcement/ Carrigan Jr, R. A. (2006). Do potential SETI signals need to be decontaminated? Acta Astronautica, 58(2), 112–117. Chalmers, D. (1996). The Conscious Mind. Oxford University Press, New York. Deutsch, D. (2002). The structure of the multiverse. Proceedings of the Royal Society of London A: Mathematical, Physical and Engineering Sciences, 458(2028), 2911–2923. https://doi.org/10.1098/rspa.2002.1015 Devlin, H. (2016, November 18). The cryonics dilemma: will deep-frozen bodies be fit for new life? The Guardian. Retrieved from http://www.theguardian.com/science/2016/nov/18/the-cryonics-dilemma-will-deep-frozen-bodies-be-fit-for-new-life Drexler. (1986). E.: Engines of Creation. Anchor Press. Dvorsky, G. (2015). Will Our Descendants Survive the Destruction of the Universe? IO9. Retrieved from https://io9.gizmodo.com/will-our-descendants-survive-the-destruction-of-the-uni-1744169933 Egan, G. (2011). Zendegi. Night Shade Books. Elitzur, A. C., & Vaidman, L. (1993). Quantum mechanical interaction-free measurements. Foundations of Physics, 23(7), 987–997. Ettinger, R. C., & Rostand, J. (1965). The prospect of immortality. Sidgwick and Jackson. Fedorov, N. (1903). Философия общего дела. Moscow. Guardian. (2006). Freezer failure ends couple’s hopes of life after death | Science | The Guardian. Retrieved from https://www.theguardian.com/science/2006/mar/17/france.internationalnews Gwern. (2017). Plastination versus Cryonics. Retrieved from https://www.gwern.net/plastination Hall, M. J., Deckert, D.-A., & Wiseman, H. M. (2014). Quantum phenomena modeled by interactions between many classical worlds. Physical Review X, 4(4), 041013. Hawking, S. W. (2014). Information Preservation and Weather Forecasting for Black Holes. ArXiv:1401.5761 [Gr-Qc, Physics:Hep-Th]. Retrieved from http://arxiv.org/abs/1401.5761 Hellinger, B. (2001). Love’s own truths: Bonding and balancing in close relationships. Zeig Tucker & Theisen Publishers. Higgo, J. (1998). Does the “many-worlds” interpretation of quantum mechanics imply immortality? Retrieved from https://web.archive.org/web/20050203192203/http://higgo.com:80/quantum/qti.htm Houellebecq, M. (2007). The possibility of an island. Vintage. Jones, J. (2017). Technological Resurrection: A Thought Experiment. Kardashev, N. S. (1985). On the inevitability and the possible structures of supercivilizations (Vol. 112, pp. 497–504). Presented at the The search for extraterrestrial life: Recent developments; Proceedings of the Symposium, Boston, MA, June 18-21, 1984 (A86-38126 17-88). Dordrecht, D. Reidel Publishing Co., 1985, p. 497-504. Kishikami, Y. (2010). Candle Night - candle-night.org. Retrieved May 24, 2018, from http://www.candle-night.org/en/candle-story/atomic-bomb-fires-are-still-bu.html Kolak, D. (2004). I am you. Springer. Letzter, R. (2018). MIT Just Cut Ties with Nectome, the “100-Percent-Fatal” Brain-Preserving Company. Retrieved from https://www.livescience.com/62202-mit-nectome-brain-upload.html Lewis, D. (1976). The paradoxes of time travel. American Philosophical Quarterly, 13(2), 145–152. Little, H. F. (2016, August 26). How an artist’s legacy became big business. Financial Times. Retrieved from https://www.ft.com/content/d77d5e74-69e5-11e6-ae5b-a7cc5dd5a28c Mallah, J. (2009). Many-Worlds Interpretations Can Not Imply “Quantum Immortality.” ArXiv:0902.0187 [Quant-Ph]. Retrieved from http://arxiv.org/abs/0902.0187 Merkle, R. C. (1992). The technical feasibility of cryonics. Medical Hypotheses, 39(1), 6–16. Mikula, S. (2016). Progress towards mammalian whole-brain cellular connectomics. Frontiers in Neuroanatomy, 10, 62. Minerva, F., & Sandberg, A. (2017). Euthanasia and cryothanasia. Bioethics, 31(7), 526–533. Mühlethaler, M., Curtis, M., Walton, K., & Llinas, R. (1993). The Isolated and Perfused Brain of the Guinea-pig In Vitro. European Journal of Neuroscience, 5(7), 915–926. Nesov, V. (2009). Counterfactual Mugging. Retrieved from https://www.lesswrong.com/posts/mg6jDEuQEjBGtibX7/counterfactual-mugging Neumeyer, J. (2014, April 10). A Visit to Moscow’s Brain Institute. Vice. Retrieved from https://www.vice.com/en_us/article/qbejbd/a-visit-to-moscows-brain-institute Newton, C. (2015). Speak, Memory. The Verge. Retrieved from https://www.theverge.com/a/luka-artificial-intelligence-memorial-roman-mazurenko-bot Nietzsche, F. (1883). Thus spoke zarathustra. Jester House Publishing. Olson, C. B. (1988). A possible cure for death. Medical Hypotheses, 26(1), 77–84. Osborne, A. (2014). The teachings of Ramana Maharshi. Random House. Parfit, D. (1984). Reasons and persons. OUP Oxford. Pigliucci, M. (2014). Mind uploading: a philosophical counter-analysis. Price, H. (2008). Toy models for retrocausality. Studies in History and Philosophy of Science Part B: Studies in History and Philosophy of Modern Physics, 39(4), 752–761. Randall, A. F. (2004). Quantum miracles and immortality. Quantum, 5, 8. Reardon, S. (2017). A giant neuron found wrapped around entire mouse brain. Nature News, 543(7643), 14. Rothblatt, M. (2007). The Geoehics of Self-Replicating Biomedical Nanotechnology for Cryonic Revival. Journal of Geoethical Nanotechnology, 1(3). Rothblatt, Martine. (2012). The Terasem mind uploading experiment. International Journal of Machine Consciousness, 4(01), 141–158. Rouleau, N., Murugan, N. J., Tessaro, L. W., Costa, J. N., & Persinger, M. A. (2016). When is the brain dead? Living-like electrophysiological responses and photon emissions from applications of neurotransmitters in fixed post-mortem human brains. PloS One, 11(12), e0167231. Rousseau, J.-J. (1782). Les confessions. Ruardij, T. G., Goedbloed, M. H., & Rutten, W. L. C. (2003). Long-term adhesion and survival of dissociated cortical neurons on miniaturised chemical patterns. Medical and Biological Engineering and Computing, 41(2), 227–232. https://doi.org/10.1007/BF02344894 Sandberg, A. (1999). The Physics of Information Processing superobjects: Daily Life Among the Jupiter Brains. Retrieved from https://www.fhi.ox.ac.uk/publications/sandberg-a-1999-the-physics-of-information-processing-superobjects-daily-life-among-the-jupiter-brains-journal-of-evolution-and-technology-51-1-34-2/ Sicoe, V. (2014, April 12). The KARDASHEV Scale (types 0 to VI). Retrieved April 22, 2018, from http://www.veronicasicoe.com/blog/2014/04/the-kardashev-scale-0-to-6/ Smart, J. (2012). The transcension hypothesis: Sufficiently advanced civilizations invariably leave our universe, and implications for METI and SETI. Acta Astronautica Volume 78, September–October 2012, Pages 55–68. Retrieved from http://www.sciencedirect.com/science/article/pii/S0094576511003304 Steinhart, E. (2015). Naturalistic Theories of Life after Death. Philosophy Compass, 10(2), 145–158. Stone, H., & Stone, S. (2011). Embracing Our Selves: The Voice Dialogue Manuel. New World Library. Swan, L. S., & Howard, J. (2012). Digital immortality: Self or 0010110? International Journal of Machine Consciousness, 4(01), 245–256. Tegmark, M. (2009). The multiverse hierarchy. ArXiv Preprint ArXiv:0905.1283. Tegmark, M. (2014). Our Mathematical Universe: My Quest for the Ultimate Nature of Reality (1st edition). New York: Knopf. Telegraph. (2012, January 4). Twins born five years apart. Retrieved from https://www.telegraph.co.uk/news/health/news/8991739/Twins-born-five-years-apart.html Tipler, F.J. (1997). The physics of immortality: Modern cosmology, God, and the resurrection. Tipler, Frank J. (1980). 3 - General Relativity and the Eternal Return†. In Essays in General Relativity (pp. 21–37). Academic Press. https://doi.org/10.1016/B978-0-12-691380-4.50009-5 Torres, P. (2014). Why Running Simulations May Mean the End is Near. Retrieved from https://ieet.org/index.php/IEET2/more/torres20141103 Turchin, A. (2016). Identity map - Less Wrong. Retrieved February 16, 2018, from http://lesswrong.com/lw/nuc/identity_map/ Turchin, A. (2017). Near-term AI safety. Retrieved from https://www.academia.edu/34863051/Near-term_AI_safety Turchin, A. (2018a). Digital Immortality: Theory and Protocol for Indirect Mind Uploading. Retrieved from https://philpapers.org/rec/TURDIT Turchin, A. (2018b). Forever and Again: Necessary Conditions for the “Quantum Immortality” and its Practical Implications. Ungar, P. S., & Sponheimer, M. (2011). The diets of early hominins. Science, 334(6053), 190–193. Urban, T. (2017). Neuralink and the Brain’s Magical Future. waitbutwhy.com. Retrieved from http://waitbutwhy.com/2017/04/neuralink.html Volpicelli, G. (2016). This Transhumanist Records Everything Around Him So His Mind Will Live Forever. Vice.Motherboard. Retrieved from https://motherboard.vice.com/en_us/article/4xangw/this-transhumanist-records-everything-around-him-so-his-mind-will-live-forever Weir, P. (1998). The Truman Show. Retrieved from http://www.imdb.com/title/tt0120382/ Wiley, K. (2014). A Taxonomy and Metaphysics of Mind-Uploading (1 edition). Humanity+ Press and Alautun Press. Wiley, K. B., & Koene, R. A. (2015). The Fallacy of Favoring Gradual Replacement Mind Uploading Over Scan-and-Copy. ArXiv Preprint ArXiv:1504.06320. Yampolskiy, R. (2014). The Universe of Minds. arXiv preprint arXiv:1410.0369. Yudkowsky, E. (2009). The anthropic trilemma. Retrieved from http://lesswrong.com/lw/19d/the_anthropic_trilemma/ Yudkowsky, E., & and, R. B. (2008). Artificial Intelligence as a Positive and Negative Factor in Global Risk. Global Catastrophic. Oxford: Milan Cirkovic. Zackrisson, E., Calissendorff, P., Asadi, S., & Nyholm, A. (2015). Extragalactic SETI: The Tully–Fisher Relation as a Probe of Dysonian Astroengineering in Disk Galaxies. The Astrophysical Journal, 810(1), 23. https://doi.org/10.1088/0004-637X/810/1/23