Reviving Alcor’s patients from cryopreservation is fundamental to its mission. As the revival technology begins to come into focus, the process of planning patient revival can begin. At an abstract level, the revival process might use any of three primary means: in situ repair, scan-and-restore, or scan-to-WBE (Whole Brain Emulation). In situ repair will require the development of medical nanorobots capable of operating in cryopreserved tissues (“cryobots”), while scan-and-restore and scan-to-WBE could benefit from this technology. The cryonics community will likely have to play a major role in the development of cryobots. In addition, while it might not seem immediately obvious, the need to test any cryopreservation revival protocol on human subjects before it is used to revive cryopreserved patients, combined with the need to comply with basic ethical principles, will force the extensive use of computer simulations, WBEs, neurobots to monitor nerve impulses, and technologies to scan cryopreserved brains. WBEs, neurobots and scanning technologies are, therefore, of broad interest to all members of the cryonics community who seek to ethically evaluate cryopreservation revival protocols before they are used to revive cryopreserved patients.
Alcor’s mission has three major elements:
1. Maintain the current patients in biostasis.
2. Place current and future members into biostasis (when and if needed).
3. Eventually restore to health and reintegrate into society all patients in Alcor’s care.
While acknowledging its importance, we have historically ignored the third major element of our mission: reviving our patients.
The technologies that will allow us to carry out this component of our mission are finally becoming clear, and we can now begin the process of planning for the revival of Alcor’s patients. In the following discussion, we will distinguish between those technologies that are likely to have such broad societal value that they will probably be developed without substantial input from the cryonics community, such as molecular nanotechnology and nanomedicine, and those technologies that might require support from the cryonics community, such as medical nanorobots capable of operating at cryogenic temperatures in cryopreserved tissues (i.e., “cryobots”).
Identifying the critical tasks that will not happen unless we make them happen is crucial if the cryonics community is to revive our cryopreserved friends and loved ones as rapidly and reliably as possible.
What are these critical tasks? Horizon Mission Methodology  is a method for making long term plans to accomplish major objectives that appear, upon initial examination, to be either very difficult or even impossible. We can apply Horizon Mission Methodology to the problem of reviving Alcor’s patients.
The core concept is to look back at the present from the perspective of a future in which the objective has already been successfully achieved. Reorienting one’s thinking to this new conceptual framework greatly facilitates the search for a solution and allows a clearer reexamination of previous assumptions that might otherwise have inhibited a clear understanding of possible answers.
The assumption that the objective has been accomplished also significantly reduces the size of the search space, simplifying the search for a solution. Paradoxically, the more difficult the objective appears, the greater the reduction in the size of the search space and the more effective Horizon Mission Methodology becomes.
Our conceptual framework is that we are looking back from the year 20xx (with the particular value of “xx” left unspecified), the year in which Alcor’s patients were revived. In looking back from the perspective of those who successfully revived Alcor’s patients, the first thing we realize is that we almost certainly had to carry out repairs at cryogenic temperatures, at least in the early stages of the repair process, even if some less major components of the process were deferred until after the patient was warmed. We now can examine what these early stages of the repair process must have looked like.
Three Methods for Revival
The three primary methods by which revival from a state of cryopreservation might take place include in situ repair, molecular scan-and-restore, and scan-to-WBE.
In Situ Repair
In situ repair uses the minimum repair methodology necessary for any given region of tissue. In this approach, any functional tissue, or tissue that can be restored to a state from which it can restore itself to a functional state, will be retained and repaired.
In situ repair scenarios typically involve medical nanorobots called “cryobots” that enter the cryopreserved tissue at liquid nitrogen temperature by “tunneling” through the circulatory system, bringing them to within approximately 20 microns of every point in the patient’s brain and approximately 40 microns of every point in other tissue.
Cryobots might be about 2 microns in diameter and have robotic arms designed using rotary joints. Molecular rotary joints can have very low energy dissipation. Small onboard computers guide their actions, while more substantial computational power is provided externally. Onboard power dissipation must be limited to prevent undesired elevation of tissue temperature. External computation does not suffer from this constraint and guides overall repair activity.
Alcor need not develop the required molecular computers, nor the medical nanorobots that operate in liquid water. However, it is less clear that operation of cryobots at cryogenic temperatures for repair of cryopreserved tissue will be developed by mainstream organizations. It seems more likely that the organization that develops cryobots will be founded by cryonicists, possibly with Alcor’s help. In any event, we assume that cryobots are developed by an organization called Cryobotics, whose precise nature (for profit, non-profit, jurisdiction, source of funding, etc.) is left open.
In situ repair will use cryobots. Cryobots operate in cryopreserved patients at cryogenic temperatures. Their first mission is to tunnel out the patients’ circulatory system. Their second mission is to assess the state of each region of tissue. The primary purpose of this assessment is to determine if the local tissue can be warmed to enable further repairs to take place in the liquid state. The expectation is that well cryoprotected regions that are minimally damaged (“good regions”) could be rewarmed until they are liquid, and repairs would continue in the liquid state. Regions where cryoprotectant did not penetrate, or which were otherwise subjected to significant damage (“bad regions”), would be processed at low temperature using in situ molecular scan-and-restore. Repair of fractures proceeds by taking a local molecular scan of the region near the fracture, and enough of the surrounding region to enable entry of cryobots into the extended fracture region, followed by rebuilding of the extended fracture region. On-site cryobots will report out to off-site computers, which will analyze the results of any regional molecular scans and develop a regional rebuilding plan for the bad regions compatible with the boundary conditions defined by the adjacent good regions.
Cryobots will need to be able to communicate with external computational resources, as well as take local molecular scans.
The primary requirement for correct functioning of cryobots is the ability to identify and tunnel through the circulatory system. For this to be possible, the circulatory system in the cryopreserved patient must be relatively intact and identifiable, especially the capillaries. If the circulatory system cannot be identified, or if it is not possible to create an appropriate network of tunnels without damaging the tissue, then it might be necessary to take a molecular scan of the cryopreserved patient as a whole. If, upon assessment, most of the tissue cannot be warmed to a liquid state but must be scanned in place, then a molecular scan of the cryopreserved patient as a whole would seem more appropriate. The level of damage that would prevent correct identification of the circulatory system, followed by tunneling into it by cryobots, would likely be severe.
We assume that the organization that develops cryobots is called Cryobotics. The primary charge of Cryobotics will be to develop those components of the technology that mainstream companies have not developed. We can safely assume that mainstream companies will have developed medical nanorobots able to operate in patients at liquid-water temperatures as well as the nanofactories necessary to build these nanorobots. Cryobotics will have to develop the cryobots that carry out the cryogenic component of in situ repair. Once the tissue has been restored to a liquid state it is reasonable to expect that more conventional medical nanorobots will be able to deal with the patient. The specific questions raised by cryonics will be operation of cryobots at cryogenic temperatures: tunneling through the circulatory system, setting up the communications and power infrastructure, taking local molecular scans, restoring tissue at cryogenic temperature, and rapidly warming cryopreserved tissue to a liquid state. Cryobotics might also play a major role in the development of molecular scan technology, as local molecular scans are an integral component of in situ repair. Local molecular scans might be necessary in patients with imperfect cryopreservation in narrowly confined areas, even when the overall cryopreservation is excellent.
Who Funds Cryobotics?
A few questions of great practical importance will include:
1. Who will fund Cryobotics and why?
2. Does funding have to come entirely from the cryonics community?
3. Must funding be in the form of donations, or are there commercial applications of cryobots?
4. Might Cryobotics be funded because cryobots could perform useful surgical procedures on patients who are beyond the ability of conventional surgical techniques? Might a conventional medical patient ever decide to be cryopreserved because certain treatment options are only available to cryopreserved patients?
5. Are there medical conditions that can be treated by cryobots, but whose treatment would be difficult or impossible by other means?
If cryobots have applications in conventional medicine, then Alcor and the cryonics community would not have to fund some or all of their development. A high-leverage activity would be to envision such applications, find those who would benefit from them, and explain to them the benefits thereof. Such beneficiaries might then be induced to provide significant funding for Cryobotics. However, it seems almost certain that the task of envisioning and identifying high-value applications of cryobots—and very likely the early developmental work as well—will fall to the cryonics community.
If funding comes from donations, Cryobotics should be structured as a nonprofit. If funding comes from investors, Cryobotics should be structured as a forprofit. If funding comes from both, great care should be exercised with respect to intellectual property (IP) issues, as mixing non-profit and for-profit organizational structures can create legal issues. For-profit entities can legally donate IP to non-profit entities, although normally they would not wish to do so because this would mean lost profit opportunities. There are legal restrictions on the sale of IP developed and owned by non-profit entities to for-profit entities. Developing IP under the guise of being a non-profit and then using the fruits of that development work in a for-profit activity would violate the public purpose for which the non-profit status was granted. There is likely to be a very restricted market for IP in the early stages of development, and therefore difficulty in establishing that the sale was in fact an arm’s length transaction. The terms of any sale are likely to be subjected to intense legal scrutiny in hindsight, once the great monetary value of the IP is obvious and any early uncertainty has been forgotten.
As a consequence, if a close working relationship involving both non-profit and for-profit components is anticipated for the structure of Cryobotics, then a careful legal review of that structure should be conducted to ensure that IP issues are handled in a way that will produce satisfactory results for all parties concerned throughout the life of the project.
Identifying sources of funding for Cryobotics is critical for rapid development of the required technology. These sources might include: (1) wealthy members of the cryonics community who expect to be cryopreserved and who set up trusts or foundations able to fund Cryobotics; (2) living wealthy members of the cryonics community with cryopreserved loved ones who wish to fund Cryobotics; or (3) members of the cryonics community who can identify major value creation opportunities for Cryobotics, and then help to develop those opportunities.
Getting Funding from Outside the Cryonics Community
It would be highly desirable to obtain funding from outside the cryonics community. How to obtain this funding before successful revival of cryopreserved patients has been demonstrated is not entirely clear. Hopefully, there are reasons for pursuing research in this area that are unrelated to reviving cryopreserved patients.
The in situ scan technology used for a particular region of cryopreserved tissue might depend on the quality of its cryopreservation. The lower the quality of the cryopreservation, the more difficult it will be to accurately restore the tissue and the more important it will be to use a scan technology that provides as much information about the tissue as possible. The greatest amount of information would be provided by a molecular scan, which, by definition, produces exact information about the position and type of every atom and molecule in the scanned tissue. A molecular scan is therefore the most conservative type of scan technology, and would be preferred if there was any question about the type of scan technology that was needed.
Should the quality of an entire cryopreservation be sufficiently poor that it becomes prudent to perform local molecular scans in essentially all regions of the brain, then the use of molecular scan-and-restore throughout the entire brain would be preferred. It would also be logistically simpler and more reliable. The precise level of damage at which it becomes reasonable to do this is unclear, but given that the quality of cryopreservation can vary widely, it seems likely that this will be the appropriate course of action for at least some patients.
Molecular scan-and-restore should be effective even in cases of severe damage. It consists of three steps: (1) a molecular scan, (2) processing of the scan, and (3) physically restoring the patient from the processed scan. In this approach, the molecular scan gathers complete information about the molecular structure of the patient’s tissues, particularly including the brain. A molecular scan gives the position and type of every atom. It provides the raw information that could, after processing, serve as the basis for restoring the scanned patient. This approach should be applicable in cases that would, by any present-day criteria, be considered beyond hope.
What is a Molecular Scan?
A molecular scan is any method of scanning which provides the location, orientation and type of every atom and molecule in the cryopreserved tissue. If we assume that every molecule has one, or at most a few, stereotypical three dimensional shapes, then we can readily approximate the total number of bits required to store an exact description of the molecular structure of the scanned tissue. A molecular scan will literally give us the location and type of every atom in the cryopreserved tissue.
To give a specific example, a single hydrogen atom might be encoded by four numbers: an X coordinate, a Y coordinate, a Z coordinate, and an atom type. Each coordinate might require 40 bits to specify, so that the three coordinates together might take 120 bits to specify. The atom type might take 6 bits to specify. A single atom would then take 126 bits to specify. A water molecule, consisting of three atoms, would require 372 bits. A more compact representation for a water molecule would specify its location (120 bits), the type of the molecule (perhaps 20 bits), and its orientation (roll, pitch, and yaw, perhaps 20 bits each), for a total of 200 bits. This is a more compact representation (200 is less than 372), especially useful in cases such as water where large numbers of them are present. This method of compressing the representation becomes more effective for bigger molecules and larger structures. A single molecule, no matter how big, can be specified with only 200 bits (provided it adopts only one functionally significant conformation during normal biological operations). For example, specifying the position and orientation of a ribosome specifies the positions and types of all the atoms that compose it. Those familiar with data compression methods will realize that a variety of methods for reducing the size of the data encoding the information about the molecular structure of the tissue are available.
Molecular scans are generally divided into two types: destructive and non-destructive. Destructive scans, as their name implies, disassemble the cryopreserved tissue in the process of scanning it. Non-destructive scans preserve the tissue intact.
Reliable methods for conducting destructive molecular scans that entirely disassemble the tissue are relatively easy to envision (e.g., “Backups Using Molecular Scans”). Such methods might be based on high resolution Scanning Probe Microscopy (SPM) methods. SPMs rely on the physical interactions between a molecular-sized tip and the surface being scanned. The mechanism positioning the tip can be large (as in today’s SPMs) or could be very small, even molecular, in scale, in future SPMs built using molecular nanotechnology (MNT). A parallel array of SPM tips spaced approximately 100 nm (10-7 m) apart seems feasible, and would allow the surface of the brain (or other tissue) to be rapidly scanned. Assuming a moderately fast scan rate of 10 MHz (10 million pixels per second per tip) and an atomic resolution of 0.1 nm (10-10 m, one angstrom), means each tip would be able to scan its 100 nm x 100 nm square region in 0.1 second. Assuming a rate of penetration into the tissue of 1 nm per 0.1 second yields a molecular scan rate of approximately 106 nm/day, or 1 mm/day. A 100 mm thick brain could be completed in approximately 100 days. Thus we can readily envision at least one future molecular scan technology able to scan an entire cryopreserved human brain in a few months or less. Partial molecular scans would require less time.
There has not yet been published any detailed proposal for a non-destructive molecular scan technology able to scan a structure as large as a cryopreserved human brain. How this might be done is, at present, an open research question, although some intriguing research has been done in the area of high resolution MRI.
Processing Molecular Scan Data
Once we have the raw scan data from the molecular scan, that data must be processed. In the most favorable case, the cryoprotection went well and the data is beautiful, crisp, and complete. As the data becomes increasingly distorted and as increasing amounts of noise are introduced from various sources, the inherent redundancy in the original structure will be increasingly called upon to allow an accurate reconstruction. Accurate reconstruction in the face of noise is initially computationally inexpensive when the amount of noise is limited, but becomes computationally increasingly expensive until, at some point, it becomes prohibitively difficult shortly before the ability to provide an accurate reconstruction becomes infeasible and the data becomes inherently ambiguous.
Deep learning algorithms can be adapted to apply to the kind of data we’ll be able to generate from molecular scans: three dimensional high resolution atomically precise data. We’ll also have quite a bit of computer power available: at least 1012 GFLOPS/Watt. The cost of electrical power should then be at least 100-1000 times cheaper than today. That combination will give us quite a bit more computational power to apply to our image analysis. An object the size of the human brain has approximately 1027 voxels, assuming one angstrom voxels. We may be able to buy 1015 joules for as little as $10,000, giving us 1027 GFLOPS, or 109 FLOPS per voxel. That should be more than sufficient for most image analysis and deep learning purposes.
The deep learning and image analysis algorithms will have been developed for other purposes, and their application to whole brain emulation and reconstruction might have been pursued by others. However, it seems likely that at least some of this development will need to be pursued by members of the cryonics community, and possibly by Alcor.
It will be useful to plan how the image analysis will integrate with the data produced by the molecular scan. We’ll likely have to start with “model systems” and incrementally work our way up to bigger and bigger systems.
The “image analysis” or “deep learning” or “AI software” is assumed to produce, as output, an atomically precise description (possibly in some compressed format) of a biological system, such as a human brain, along with the surrounding support structures and interface systems.
This description could then be entered into a suitable atomically precise 3D manufacturing system (or “3D printer for atoms”) to fabricate the described structure. It seems reasonable to assume that manufacturing takes place at cryogenic temperatures and is followed by rapid warming.
The algorithms for processing molecular scan data will need to be developed, and it would be helpful to have as clear an idea of what these algorithms will look like as possible. One strategy for doing this would be to generate synthetic molecular scan data. If we assume that molecular scans will provide us with atomically precise information about the cryopreserved structure, then it should be possible to generate synthetic molecular scan data by creating atomically precise descriptions of cellular structures based on our current understanding of such structures, then applying damaging transformations based on our current understanding of the transformations involved in present-day cryopreservation methods. The resulting synthetic molecular scan data could then be used as input to aid in developing and debugging the algorithms used in processing molecular scan data.
Arguing against this approach is the likelihood that synthetic molecular scan data will deviate from actual molecular scan data in significant ways. While it would still be possible to test and debug the algorithms to be used in processing real molecular scan data on synthetic scan data, there would be a risk that the resulting algorithms, even if they performed well on the synthetic scan data, might still not perform well on real scan data. But developing and testing algorithms on synthetic scan data should speed development even if such testing was incomplete, and even if further testing and debugging on real scan data was still required.
Of course, it is also possible that molecular scans might prove to be significantly more detailed than is required, and that some lesser scanning method will prove to be sufficient (see “Lower Resolution Scans”), rendering the need to analyze molecular scan data moot.
Should it be possible to develop algorithms that are easily generalizable, then algorithm development could start today, with the understanding that any specific algorithm might not be used but that the general concepts developed could still form the framework within which the actual scan technology would be developed and the scan processing would take place.
A Molecular Scan is the Best We Can Do
A molecular scan provides us with all the information about the cryopreserved tissue that it is possible to obtain. No further information can be obtained. A molecular scan puts us in the best possible position to restore the scanned tissue to a healthy state. If we can’t restore a person with their memories and personality intact after a molecular scan, then there’s too little information in their cryopreserved brain to do this.
To put it another way, if someone has been cryopreserved and we pursue any other method for reviving them based on their cryopreserved tissues, we cannot, in principle, do any better than by starting with a molecular scan. In particular, if a cryopreservation went badly and we attempt to revive the person by warming them up and using some form of biological repair, such a biological repair process cannot, in principle, do a better job than a restoration process that started with a molecular scan. The reason for this is simple. After rewarming, the biologically oriented repair process must contend with the continuing deterioration of the damaged molecular and cellular structures. Ruptured membranes will continue to allow mixing of the contents of cellular compartments. Damaged molecular structures will continue to deteriorate and entropy will continue to increase. The biological repair processes, whatever they might be, will be fighting against extensive levels of damage and would have to move with implausibly great speed simply to limit the further spread of that damage, let alone to perform repairs.
By contrast, a molecular scan provides a snapshot of the system at the moment it was cryopreserved. There will be no further deterioration. Entropy is held in check. The computational processes that examine the digitized tissue can do so at leisure, mathematically restoring the digital representations of the structures to their appropriate state as though they were frozen in time.
Only after the full digital restoration has been completed and every detail has been attended to would the whole digitally restored structure then be actually converted back into a physical structure. This conversion process could take place either by carrying out a series of lowtemperature repairs on the existing physical structure, using the digital restoration held in computer memory as a guide; or by using what would amount to a 3D printer for atoms that allows the exact three dimensional structure to be printed in atomically precise detail.
If a non-destructive molecular scan technology can be developed, then it could be applied to every cryopreserved patient. It would provide valuable information that could be used to assist the repair process, whatever that repair process might be, and would cause no damage that might impair subsequent efforts to revive the patient. Further, it would provide an invaluable failsafe in case the repair process went awry.
However, if only destructive molecular scan technologies are available, then their application to a specific patient would require weighing the benefits of the information they provide against the possibility that damage to the original structure might impair subsequent steps in the revival process. Some cryopreservation patients would prefer recovery of complete information about themselves through a molecular scan, regardless of whether or not it was destructive. Other cryopreservation patients might elect to have a destructive molecular scan only if it were necessary for their successful revival and only if there were no other options [Alexandre Erler, “Brain Preservation and Personal Survival: The Importance of Promoting Cryonics-Specific Research,” Cryonics magazine, November-December, 2017]. There may even be some cryopreservation patients who would forego a destructive molecular scan altogether, even if this meant failure of their revival. A destructive molecular scan is compatible with, and could be used as the starting point for, the biological restoration of a patient.
A destructive molecular scan, followed by the use of a digital restoration algorithm, followed by the use of an atomically precise 3D printer to instantiate the resulting atomically precise digital restoration, might be effective at producing a high fidelity and biologically accurate reproduction of the original person, in cases where methods that did not involve digital restoration would produce unsatisfactory results.
For these reasons, further research on a purely non-destructive molecular scan should be pursued. This technology could be used in all cases, by all people, regardless of their philosophical views.
Backups Using Molecular Scans
At a deeper level, tissue is information: the two are interchangeable. Anyone who seeks a very long lifespan, and who acknowledges that accidents can happen, must at some point come to terms with the need for backups: sufficiently accurate descriptions of themselves from which they can be restored, should they suffer from a misfortune so catastrophically damaging that no recovery from that misfortune is otherwise possible. This is both feasible and obviously desirable.
If a destructive molecular scan is taken of your cryopreserved self and the processed scan is used as the blueprint from which you are restored, this is philosophically similar to awakening from a backup after a catastrophic mishap. Can existing proposals for destructive molecular scans, based on SPM technology, be carried out reliably? In other words, if we disassemble tissue in the process of scanning it, as is called for by existing proposals for molecular scans, then the scan needs to be quite reliable, as the tissue will be gone when the scan is finished. If the scan is lost, and the tissue that was scanned is no longer available, then the person being scanned will be dead—clearly an undesirable outcome.
An SPM can scan the exposed surface of a block of tissue, characterizing it completely. After the surface has been completely characterized, but not modified, the information about the surface could be digitized and stored. All information from this surface scan can be continuously and redundantly transferred to stable storage media as the exploration of the tissue block proceeds. Only after information from the ongoing scan had been duplicated and stored redundantly, or even triplicated or quadruplicated, thus providing whatever level of reliability might be desired, need the scanning process proceed to the next step: removing the scanned surface layer to expose the layer beneath it. Very high reliability should be feasible.
This method of analyzing tissue is both conceptually simple, and can be made highly reliable:
1. Analyze the tissue surface using SPM technology.
2. Redundantly and reliably store the results of the surface analysis.
3. Only then, after confirming storage of the analyzed surface, remove the analyzed surface and expose the next layer.
While simple and reliable, and capable of providing molecular scans, this method does have the obvious disadvantage that is disassembles the tissue in the process of analyzing it.
Is there a method of carrying out a molecular scan that does not require disassembly? The answer to this question is more difficult. There could well be a way of gaining molecularly and atomically precise knowledge of tissue without disassembly, but it is not immediately obvious how this might be done. Magnetic Resonance Imaging (MRI) using nanoscale devices operating from adjacent capillaries and performing indirect scans of the intervening tissue offer intriguing possibilities, but a molecular scan of something as large as the human brain still presents significant technical challenges. The options made available by MNT and complete access to the circulatory system have not been fully explored. Further studies are needed to understand the possibilities and to provide a reliable answer to this question.
Lower Resolution Scans
A question of some interest is whether molecular scans are actually necessary, and if lower resolution scans might be sufficient. While we can be confident that a full molecular scan will be sufficient if anything is sufficient, lower resolution scans that provide less information about the tissue being scanned might also be sufficient, depending on the type of scan and the use to which the data is being put.
The question of what sort of information we need is one where neuroscience must inform our discussion. How much information is required to construct a satisfactory model of the human brain? While it’s rather obvious that we don’t need to know the location and orientation of every molecule in the brain (esp. the orientation of all the water molecules), how much information do we need to know? And what sort of scanning technologies might provide us with enough information at a sufficiently low cost? There are many existing research projects aimed at developing high resolution three dimensional images of biological tissues, including the human brain. At some point in the future, it should be possible to obtain funding to apply MNT to this problem. Again, further research is required.
A third alternative that some patients might explicitly request is to process the information from a molecular scan and use it to directly construct a whole brain emulation (WBE). This “scan-to-WBE” option might be simpler than the molecular scan-and-restore process, as it would eliminate the need for physically restoring a biological body. Scan-to-WBE would rely entirely on the information recovered from the cryopreserved tissue.
It is possible that the technology for molecular scans and Whole Brain Emulations might become available before the technology for in situ repair. Alcor members wishing to return to an active life as quickly as possible might want to take advantage of whatever technology arrives first. Of course, those members who wish to be revived as a WBE would have to communicate this wish to Alcor before they are cryopreserved as, once cryopreserved, further communication will not be possible. This process could be facilitated if Alcor provided forms enabling members to explicitly express their wishes in this regard.
As will be discussed later, scan-to- WBE will be an essential component of the process that we will use to ethically evaluate any proposed method of reviving a cryopreserved patient. As a consequence, methods for scanning-to-WBE are of interest to everyone in the cryonics community, not just those who are specifically interested in themselves becoming WBEs.
What criteria should be applied in deciding whether to use in situ repair or molecular scan-and-restore? Some might argue that we should always employ in situ repair, relying on the fact that in situ repair will include local assessments of tissue damage and utilize local molecular scans on an asneeded basis. These local molecular scans might be performed on a larger and larger percentage of the tissue as the quality of the cryopreservation became poorer and poorer.
Many patients in Alcor’s care have inevitably suffered extensive damage. Some have suffered such extensive damage that there are serious questions about the ability of any technology, no matter how advanced, to revive them with their memories and personality completely intact. In such cases, the use of a molecular scan followed by digital restoration prior to any attempt to carry out a biological restoration (guided by the digital restoration) would seem appropriate.
While Alcor seeks to comply with patient wishes, there might be two opposing wishes at work here. On the one hand, some patients may prefer to use in situ repair for philosophical reasons. On the other hand, some patients may want to get out of the dewar as quickly as possible. It is possible that fully developing the technology for in situ repair might take longer because it appears to be a more complex technology. There are plausible scenarios in which molecular scan-and-restore might turn out to be a simpler technology to develop and deploy. It is even possible that in some circumstances, scan-to-WBE might be available before molecular scan-andrestore, which in turn might be available before in situ repair. Molecular scan-andrestore might also be less prone to residual damage than in situ repair, and more likely to correct all the damage incurred by both the cryopreservation and any preexisting medical conditions. For example, if an existing region of tissue is evaluated as “good” during the in situ repair process and is warmed without being scanned, then there is no backup for that region. Any failure during the revival process, or any undetected damage in that region, could result in a less-than-optimal revival.
As an additional confounding variable, some Alcor members might prefer being revived as a WBE living in a virtual world (if the technology is reliable). This arguably offers certain benefits, most notably the ability to make regular or even continuous backups and the opportunity to quite literally expand your mind. Patient preferences should be taken into account. The best course of action is probably to explicitly ask members what they prefer—before they are cryopreserved.
Did We Do It Right?
An obvious and rather awkward question is this: once we revive someone, how do we know we did it right? We could, of course, ask them: “How do you feel?” If they say “Terrible! I don’t feel like myself!” we might naturally be concerned. But how do we know that’s not the right answer? There are people who say that kind of thing quite a bit.
One solution is to conduct some sort of test before a person is cryopreserved, then test them again after we revive them, and compare the results. What sort of test might we conduct? How can we determine if we’ve done a high-fidelity cryopreservation and revival?
Evaluating an Animal Revival Protocol
Perhaps the most detailed functional information we could acquire about an experimental animal’s brain would be a record of every nerve impulse for some period of time. Is this feasible? Certainly with MNT, the answer appears to be “yes.”
We consider one possible approach: building “neurobots,” a class of medical nanorobots, and locating them on, in, or near nerve cells. Neurobots detect and record passing nerve impulses and have an accurate time base (either built in, or based on a centrally transmitted clock). When a nerve impulse passes by, the neurobots note the time and record the associated small fluctuations in voltage or electric field on a polymer “tape.” The tape is extruded into the extracellular space and finds its way out of the body, where it and many others like it are later recovered and analyzed. Other methods of communicating the data recorded by the neurobots are also possible.
We could record every nerve impulse in the brain by embedding a sufficient number of neurobots. A few back-of-the-envelope calculations show that the storage density of polymer tape is more than sufficient to hold all the data. Some specific proposals along these lines have already been advanced in the literature, though their effectiveness without MNT may be marginal.
The objective is to record all neuronal activity within the test subject’s brain (or other volume of interest). This has been a long-standing goal of neuroscientists. The major limitation facing neuroscientists today is the relatively large size of the devices needed to record the voltages and electric fields. MNT will enable the manufacture of devices of sufficiently small size and precision to enable this long-sought goal.
We could then record data from neurobots in the brain of an experimental animal before they were cryopreserved, cryopreserve them, revive them, and then record data from neurobots in the brain of the revived experimental animal, giving us two sets of neuronal data: “before” and “after”. Comparing the “before” and “after” data would let us tell if we had done a good job in cryopreserving and reviving the experimental animal. At a purely structural level, the connectome from “before” should be the same as the connectome “after,” except for those changes that took place because of learning, where we interpret “learning” broadly as “plastic changes in the brain caused by its normal functioning as a consequence of its interactions with a normal environment.”
To spell this out in more detail, if we wish to evaluate a protocol for cryopreserving a biological experimental animal and reviving them as a biological experimental animal, we would: (1) use neurobots to monitor all nerve impulses in a test subject, (2) construct a “before” WBE from the monitored nerve impulses, (3) cryopreserve the test subject while continuing to monitor their nerve impulses, (4) revive the test subject biologically, (5) use the neurobots to monitor all nerve impulses in the revived test subject, (6) construct an “after” WBE from the second set of data produced by the neurobots, and then (7) compare the “before” and “after” WBEs and see if there are any significant differences. If there are significant differences, then the cryopreservation and revival technologies are regarded as “not good enough”. If there are no significant differences, then the cryopreservation and revival technologies are regarded as “good enough.”
We construct “before” and “after” WBEs and compare them because it’s difficult to compare the raw data generated by the neurobots from “before” and “after”. Merely knowing that a nerve impulse passed neurobot A at time t1 “before” and that a nerve impulse passed neurobot B at time t2 “after” is not going to tell us much without a great deal of analysis.
Conceptually, the required analysis must convert the raw nerve impulse data into a picture of the neural connections of the test subject’s brain. This may be roughly likened to deriving the connectome, that is, the network of neural connections between the nerve cells in the brain, from the pattern of nerve impulses. The progression of a nerve impulse as it passes individual neurobots could be monitored, allowing the existence of a neuronal path winding along between those neurobots to be inferred. The generation of a new nerve impulse by the summation of several input nerve impulses could likewise be inferred from a sufficiently dense network of neurobots monitoring the nerve impulses in the brain. With a sufficient number of neurobots monitored for a long enough period of time, the entire connectome of the brain could be inferred. We can then use the connectome as a significant subset of the information required for a WBE.
Injection of Nerve Impulses
A question that needs to be addressed is whether or not passive data collection by neurobots will be sufficient to allow reconstruction of the connectome. That is, is it sufficient if neurobots simply monitor the existing neuronal traffic for some reasonable period of time? One can readily imagine that a particular synaptic connection between two neurons only occasionally plays a role in the pattern of nerve impulses actually generated. Monitoring nerve impulses between those occasions when that synapse plays a role would reveal nothing about that synapse.
A simple (but not necessarily realistic) example from computer science will serve to illustrate the point. A three-input MAJORITY gate has three inputs, input 1, input 2, and input 3. It will only fire if two of the three inputs take on the logical values of “1” at the same time. If we only knew the data values on the wires connecting the various logic gates, we might never realize that input 3 was connected if the actual pattern of data never had a logic “1” on input 3 at the same time there was a logic “1” on either input 1 or input 2. Thus, if there was a logic “1” on input 1 and input 2 at the same time, but never a pattern showing the gate firing when input 3 was at a logic “1” (because neither input 1 nor input 2 was at a logic “1” at that point in time), then we would conclude that the gate was a two-input AND gate, not a three-input MAJORITY gate.
While we don’t yet know whether passive collection of nerve impulse data is sufficient to allow correct inference of an individual’s WBE, we’ll need to determine the full set of synaptic connections even if passive collection is insufficient. To this end, we might need to inject signals into the nervous system, allowing us to interrogate the cellular circuits with a sufficient number of possible inputs to ensure that we have accurately determined all of the synaptic connections.
In our example of a MAJORITY gate that was incorrectly labeled as an AND gate, we would need to inject a “1” on input 3 at the same time that there was an input of “1” on input 1 or input 2. In this way, we could guarantee that we had enough data to deduce the nature of the MAJORITY gate, and correctly distinguish it from an AND gate.
Whether this will be necessary or not is unclear at the present time. If it is not necessary, then the neurobots will not need to inject signals into the nervous system, which could potentially simplify their design. If it is necessary, then the neurobots will need to be able to inject signals (nerve impulses, selective depolarization of the cell membrane) into the nervous system. A variety of methods for carrying out this task are possible.
Such an ability would, in any event, be desirable for other reasons, both in terms of treating a variety of medical conditions and in terms of diagnoses.
Comparison of WBEs
Once we have constructed a WBE from the raw data gathered by the neurobots, then it would become possible to compare two such WBEs to each other in a meaningful way, as we expect that information like the connectome of a primate before and after they have been cryopreserved should remain the same. Changes in the WBEs would either be the result of damage caused by the cryopreservation-and-revival process, or would be the result of learning that took place between the “before” and “after” WBEs. Assuming the neurobots remained in place during the cryopreservation, recording nerve impulses before and during the cryopreservation, and then later recording nerve impulses immediately following revival, there would be no loss of neuronal information. It should be possible to more directly compare the “before” and “after” WBEs with less concern about unaccounted-for changes that took place because of learning between the time the “before” WBE was taken and the “after” WBE was taken. The only unaccounted changes would then be those caused by damage due to the cryopreservation and revival process.
While this protocol works for experimental animals, we shall see later that it is ethically inappropriate to apply it to human test subjects. After some analysis of the ethical principles that must be followed, we derive a different protocol for evaluating a revival protocol that should be ethically acceptable for human use.
When is a Scan Technology Good Enough?
In the previous Section, we discussed how to evaluate a method for biologically reviving a cryopreserved experimental animal: gather data from the brain of the experimental animal before it is cryopreserved, and gather data from the brain of the experimental animal after it has been revived.
In this Section, we discuss how to evaluate a method for scanning a cryopreserved test subject and constructing a WBE.
That is, if our objective is not to biologically revive the test subject, but to construct a WBE directly from a scan, how might we evaluate the result? The scan might be a molecular scan, or it might be a lower resolution scan. We will also need to evaluate the algorithm used to construct the WBE from the scan data.
We will be using the same basic principle as before: constructing a “before” and “after” WBE and comparing them. However, while the “before” WBE will be constructed from neurobot data, the “after” WBE will be constructed directly from the scan data.
Again, we use neurobots to monitor all nerve impulses in a test subject, either a non-human test subject or, eventually, a human test subject. We construct a WBE from the recorded nerve impulses. We then cryopreserve the test subject. We then scan the test subject’s brain using the scan technology under investigation. We then do the scan-to-WBE using the algorithm under investigation. We then compare the two WBEs and see if there are any significant differences. If there are, then the scan technology in combination with the scan-to-WBE algorithm is judged “not good enough.” If there aren’t, the scan technology in combination with the scan-to-WBE algorithm is judged “good enough.”
It is worth emphasizing that in this case, the “after” WBE is constructed from scan data, not from neurobot data. That is, the existence of a neuron that carries nerve impulses is deduced from scan data, not from the pattern of nerve impulses. The manner in which a dendritic network processes incoming nerve signals and produces an outgoing nerve signal along the outgoing axon is deduced by examination of the scan data rather than from the incoming and outgoing nerve impulses recorded by neurobots. That is, we are deducing the existence of a neuron by examination of the scan data. The better the quality of the cryopreservation and the better and more accurate the quality of the scan, the easier it will be to determine the connectome of the test subject, and the easier it will be to build an accurate WBE of the test subject’s brain from the scan data. As the quality of the cryopreservation gets worse, and as the quality of the scan gets worse, the ability of the scan-to-WBE algorithm to recover the connectome information with high fidelity will become increasingly difficult, and will require increasingly sophisticated algorithms that are increasingly computationally intensive.
This comparison of before and after WBEs appears to be the best we can do in terms of evaluating the quality of the combined cryopreservation and revival technology, whether we are considering biological revival, or revival as a WBE. It certainly appears to be the kind of testing that Alcor and future physicians will have to carry out before reviving any patients.
Regardless of the specifics of how the before-and-after comparison is performed, the critical insight is that detailed information gathered from the entire brain, both before cryopreservation begins and after revival is complete, will be required to assess the quality of the overall process. Neurobots can gather this detailed information that is required for the “before” WBE, and neurobots can gather the same information for the “after” WBE when biological revival of animals is being evaluated, as they’ll be able to quite literally record every nerve impulse in the animal brain. If the objective is to construct a WBE without biological revival, then the “after” WBE can be constructed directly from the scan data of the experimental subject’s brain, whether that experimental subject is animal or human.
A Human Cryonics Revival Program
Applying the previously-mentioned Horizon Mission Methodology, we can now look back at the present from the perspective of a future in which the objective of human revival from the cryopreserved state has already been successfully achieved. These revivals are assumed to take place in the year 20xx, a future world in which MNT has been fully developed and in which medical nanorobots are used in medical diagnostic and therapeutic practice. This is a world in which today’s most common medical causes of death—e.g., cancer, heart disease, stroke, diabetes, and even aging—will be entirely curable conditions. “Terminal” patients are likely to be rare, unless cost or personal choice are issues.
By the year 20xx, Alcor will be reviving its patients. A substantial fraction of the patients will likely have been revived using in situ repair. Some patients who have suffered greater cryopreservation damage will have been revived by using molecular scan-and-restore. Patients who prefer being revived as WBEs might have been revived using scan-to-WBE. We now ask, looking backward from the year 20xx: How might this have occurred?
Ethical Principles for Revival
Before describing a possible revival program, we first must discuss the fundamental ethical principles that should govern the revival of cryopreserved patients. It would seem advisable for Alcor to convene an appropriate committee of well informed cryonicists to consider the issues raised by this Section so that the actual process of reviving patients can proceed smoothly and without causing unexpected concerns.
Four ethical principles seem applicable to the revival of cryonics patients:
Principle 1: Informed consent. Any party subject to an experimental procedure should be informed, as well as possible, about the procedure and its possible outcomes.
Principle 2: Before any procedure is applied to the revival of cryopreserved patients, it should be adequately tested on experimental animals, including primates, and also on human volunteers, if that is legally and ethically possible.
Principle 3: Adequate testing of a procedure for reviving cryopreserved patients should verify that as much personality-relevant information as possible is retained.
Principle 4: The risk of information-theoretic death to a patient should be minimized. As a corollary, before carrying out any procedure on a patient that might pose any risk, all available personalityrelevant information from that patient should be digitized, copied and securely stored—if this is legally and ethically possible and if, on balance, this reduces the risk of information-theoretic death.
Principle 1 is well known and generally agreed to. It has an extensive literature.
“Informed consent is a process for getting permission before conducting a healthcare intervention on a person. A health care provider may ask a patient to consent to receive therapy before providing it, or a clinical researcher may ask a research participant before enrolling that person into a clinical trial. Informed consent is collected according to guidelines from the fields of medical ethics and research ethics.”
Principle 2 is the application to the revival of cryopreserved patients of the more general principle that any medical treatment should be adequately tested before it is put into general use. How to “legally and ethically” test revival methods on human volunteers deserves further discussion.
Principle 3 is new. We have already discussed it under “Did We Do It Right?” When combined with our other principles it has consequences that require further consideration.
Principle 4 is also new. Minimizing the risk of information theoretic death is, in some sense, just a modern restatement of the age-old dictum: “First, do no harm.” The corollary given in Principle 4 will be familiar to anyone who has ever carried out a major edit operation on a file: back it up first, or you might regret it. Medical technology today cannot backup patients. However, we anticipate that the technology that will enable revival of cryopreserved patients will also enable backup of those patients. Because this is a new concept, generally accepted practices have not yet been worked out. Patient wishes must be properly taken into consideration before carrying out any procedure.
An Ethical Q&A
We review the ethical issues in the form of a Q&A, describing what we can and cannot ethically do. Following each question and answer, we discuss the ethical issues in greater detail, giving the justification for the conclusion in greater depth.
Q1: Is it ethical to implant neurobots into a human volunteer?
A1: Yes, provided the human volunteer provides informed consent and the process complies with appropriate medical safety guidelines.
In Q1, we must safely introduce neurobots into the human volunteer, gather data, and then remove them. Complying with principle 1, informed consent, should be feasible. Principles 2 and 3 are not applicable, as this step does not involve revival from cryopreservation. Principle 4 does not appear applicable, as introduction of neurobots does not appear to pose a risk of information-theoretic death.
We are, therefore, left with informed consent. As we are assuming that nanomedicine in general, and medical nanorobotics in particular, have already been developed and are in use, the introduction, operation and removal of neurobots into a test subject should be within the normal ambit of an experimental protocol of that time. This will be especially true when we consider that introduction of neurobots into human test subjects will be done only after they have been safely introduced into animals, including primates, and have been proven to be safe.
It should be possible to conduct an experimental protocol to validate the safety of neurobots in human volunteers while complying with the ethical principles given here, and also with the ethical principles that must normally be complied with in human clinical trials.
Q2: Is it ethical to use the data obtained in Q1 to construct and run a WBE?
A2: Yes, provided the volunteer provides informed consent. The rights of the WBE so constructed must be respected.
Q2 involves the construction of a WBE from the data gathered by neurobots, whose safe introduction and removal has already been validated using an ethical protocol, in accordance with the answer to Q1. It is unlikely that legal issues will be a problem in this step, as manipulation of data, even data that describes a human mind, is not yet significantly constrained by law, and will likely not be so constrained for some time. The primary issues will be ethical. It is worth emphasizing that experimental tests of WBEs of primates will have been successfully concluded, and that the long term stability of WBEs of primates in virtual environments will have been demonstrated before work on human WBEs begins. The primary remaining issue will be informed consent, including negotiations with the test subject over the appropriate protocols to be followed both during the debugging process and afterwards. For example, should a failure happen “quickly” (a few seconds? A few minutes?) then, with the prior consent of the subject and (hopefully) isolation of the problem, the data from the failed attempt could be erased.
Should a failure happen “slowly” (several weeks? A few months?) then the WBE could be preserved on stable storage until it was possible, sometime in the future, to be restored to full mental health. To put it another way, the unsuccessful WBE would be like an Alcor patient, awaiting future technology to be restored to full mental health.
Q3: Is it ethical to implant neurobots in a terminally ill volunteer, construct a WBE, and cryopreserve them with an experimental cryopreservation protocol following their legal death?
A3: Yes, provided the volunteer provides informed consent and the process complies with appropriate medical safety guidelines, and the rights of the WBE so constructed are respected.
Q3 presents a greater challenge. Following the implantation of neurobots and the construction of a WBE, we cryopreserve the test subject. Worse, we cryopreserve the test subject using protocols similar to those actually used on Alcor patients today, which in some cases are less than optimal by today’s standards, let alone the future standards in force at the time this test program will be conducted.
Is this ethical?
Legally, it seems likely that cryopreservation using today’s methods will still be viewed as causing legal death, at least at the time at which the protocols for reviving a patient cryopreserved using those methods are still being tested. It might therefore be difficult to legally ask for a healthy volunteer to submit to this experimental protocol, regardless of their motivations or expectations.
However, we could ask for terminally ill volunteers. After volunteering, they would be implanted with neurobots, from which a WBE could be constructed. We know this is ethically permissible from Q1 and Q2. They could then be (legally) cryopreserved immediately following legal death, as is done today in cryonics. This solves the legal problem. It would also be necessary to obtain informed consent.
What causes our experimental subject to volunteer? Arguably, when medical nanorobots are available and neurobots are available at least experimentally, most terminal diseases will be treatable. Our volunteer must (a) be suffering from a terminal illness which is, for some reason, not treatable by the available nanomedical technology, or (b) not be able to afford the nanomedical treatment, or (c) have declined the nanomedical treatment despite its effectiveness. Further, our volunteer must be willing to accept a sub-standard cryopreservation despite the fact that better cryopreservation technology will surely be available.
Those suffering from a terminal illness which was not treatable by the available nanomedical technology might view a WBE as a definite plus when considering whether or not to volunteer for this experiment, although the number of volunteers falling into this category might be small.
Those not able to afford the nanomedical treatment might likewise view a WBE as a major advantage when considering whether or not to volunteer for this experiment. On the other hand, it might be viewed as coercive to offer a WBE to a person who can’t afford nanomedical treatment. It certainly seems perverse to refuse admission to an experimental program to those who can’t afford nanomedical treatment on the grounds they might prefer a nanomedical treatment to the proffered WBE. It is also unclear how to reliably determine the motive for declining nanomedical treatment, as determining human motivations is frequently difficult. Whether or not a person declined nanomedical treatment because they couldn’t afford it, or because they wanted a WBE, might not be clear, even to the person making the decision.
Finally, those who declined nanomedical treatment despite being able to afford it and the fact they knew it would be effective might include some who were specifically interested in becoming a WBE, and declined treatment precisely because neurobots and WBE technology had recently become available. The opportunity to have someone else pay for the procedure might be attractive.
The latter two groups would be particularly interested in whether the neurobot and WBE technology were both fully tested and reliable. Offering members of these groups an untested technology might be viewed as unethical or coercive, suggesting that both the neurobots and the neurobot-to-WBE algorithm should be fully tested and validated prior to seeking volunteers who might fall into these categories.
The most straightforward course of action, ethically, would be to fully test and validate both the neurobots and the neurobot-to-WBE technology using healthy volunteers, relying on the answers to Q1 and Q2 to enable us to ethically carry out these validations, and then move on to the next task: seeking terminally ill volunteers.
If neurobots and neurobot-to-WBE have been experimentally validated, then the person who volunteers for the experiment to evaluate Q3 will receive an important benefit that many may regard as compelling. The volunteer will first be implanted with neurobots using a validated procedure that has been tested and has been shown to work, from which a WBE will be constructed using a procedure which has also been tested and has been shown to work. That is, the person volunteering for this can expect to get a WBE using known and tested procedures before the experimental cryopreservation procedure even begins. A volunteer who desires a WBE for whatever reasons will thus obtain one. Their WBE will survive, regardless of what happens to their biological body. Many people will find this a worthwhile proposition.
The prospect of helping cryopreserved patients may be an additional psychological bonus for these volunteers, generating a positive mental state analogous to the feeling experienced by today’s altruistic kidney donor who knows they’re saving the life of a fellow human being by their actions.
In summary, the “before” WBE is constructed from the brain of the human volunteer, using the neural traffic information provided by the neurobots. The volunteer is then cryopreserved using historical cryopreservation methods similar to those that were originally applied to the Alcor patients who are awaiting revival.
It should be possible to find willing volunteers for this part of the protocol who fully understand it and its consequences for themselves, while remaining in full compliance with accepted ethical principles.
Q4: Is it ethical to apply an experimental protocol to revive the volunteer of Q3 for the purpose of evaluating the effectiveness of the experimental protocol?
Some in the cryonics community might be disappointed at this conclusion, as reviving cryopreserved experimental human subjects would seem the obvious test of an experimental revival protocol. Unfortunately, the ethical problem can be stated quite succinctly. The problem is not getting the informed consent of the person who was cryopreserved. As discussed in Q3, there are conditions under which this consent can reasonably be obtained.
The problem is twofold. First, we must get the informed consent of the person who is revived, who might be a different person from the person who was cryopreserved. Second, we must get the approval from the broader society for what amounts to a new process for creating a human life.
If the revival process goes awry, it is possible that it might create a new person, a person who has not been consulted and who has not been given the opportunity to provide informed consent. Worse, the cryopreservation protocol utilized was not in keeping with the standard-of-care for that future time, thus creating an enhanced risk that the experimental revival protocol might produce an inappropriate outcome.
From the societal perspective, we are no longer dealing with the traditional, and actually rather safe, issue that cryonics normally deals with: that of saving an existing human life: we are now dealing with the issue of possibly creating a new human life. This issue is well known to be socially, politically, and ethically divisive. The purpose for which we are asking society to deal with this issue is the rather abstract one of testing a cryopreservation revival protocol.
If there are any who might argue that it would be ethically acceptable to move forward with an experimental program on human subjects that poses questions about creating new human life, consider just the pragmatic political vulnerability this creates. While the legal system usually works relatively slowly, laws against human cloning appeared well before any actual practice of human cloning posed any societal risks. It would be reasonable to be concerned that adverse publicity that paid little attention to the facts but was framed in a manner intended to play on people’s emotions could adversely impact any organization that pursued experimental work in this area.
Possibly creating human life for the purpose of evaluating protocols for reviving cryopreserved patients is not something we should pursue.
Q5: Is it ethical to scan the legally dead cryopreserved brain of the volunteer of Q3?
Carrying out scientific research on legally dead human remains that have been donated for the purpose is an established activity carried out in the context of a well-established regulatory environment, and provides valuable information that saves lives. Indeed, we would be scanning a cryopreserved human brain to develop methods for saving the lives of cryopreserved patients, and will, inter alia, be providing a great deal of medically useful information.
Q6: Is it ethical to use the brain scan data from Q5 to construct a WBE using an experimental algorithm?
It is worth noting here that we are running an experimental scan-to-WBE algorithm, but are carefully not running the WBE. The ethical issues involved in running a WBE only arise when the WBE is “switched on”. The data describing a WBE is just that: data. The data describing a WBE that is sitting on a DVD, for example, cannot feel pain. Only when the DVD is loaded into a computer and the computer starts running are we concerned that the WBE, now a running process, might feel pain (depending on exactly what input is provided and exactly what the WBE is doing).
As a consequence, we can confidently state that running an experimental scan-to-WBE algorithm, but not actually running the WBE, avoids the ethical issues that are typically associated with WBEs.
Q7: Is it ethical to run the WBE constructed in Q6?
Recall that the WBE was constructed using an experimental scan-to-WBE algorithm applied to data derived from a scan of a volunteer cryopreserved using a method similar to those used on existing Alcor patients, and therefore not best practice (and likely falling significantly short of best practice) for the time that it was done.
This combination of an experimental scan-to-WBE algorithm applied to data derived from a less-than-best-practice cryopreservation makes it ethically dubious to “switch on” the resulting WBE.
Q8: Is it ethical to compare the WBE from Q6 with the WBE constructed in Q3?
This answer is simple. A WBE that is not running is simply data. We are comparing one set of data with another set of data. This is trivially ethically permissible. We are not running the WBE from Q6, and are not running the WBE from Q3 for the purpose of evaluating the revival protocol. Simply comparing two sets of data does not pose any significant ethical issues.
An Ethically Acceptable Protocol for Evaluating a Revival Protocol
Now that we have walked through the ethical issues, we can describe an ethically acceptable protocol for evaluating a revival protocol, an experimental scan technology and an experimental scan-to-WBE algorithm on human volunteers.
1. Implant neurobots into a terminally ill human volunteer, after asking the volunteer for informed consent, using a successfully tested and validated protocol that complies with appropriate medical and ethical principles.
2. Cryopreserve the volunteer following their legal death using a cryopreservation protocol modeled after those actually used to cryopreserve Alcor patients.
3. Use the data obtained from the neurobots implanted in 1 to construct a WBE using a successfully tested and validated neurobot-to-WBE algorithm. Run the WBE in a suitable environment if that was part of the agreement with the terminally ill volunteer.
4. Carry out those parts of the revival protocol that take place at cryogenic temperatures.
5. Scan the cryopreserved brain produced by step 4 using the experimental scan technology.
6. If the purpose is to evaluate a biological revival protocol, then the scan in step 5 should either be a non-destructive scan or a destructive molecular scan. If it was a destructive molecular scan, rebuild an atomically precise duplicate of the cryopreserved brain that was destructively scanned using the scan data. Otherwise, simply continue.
7. Re-warm isolated one-cubicmillimeter samples of tissue from the cryopreserved volunteer to verify the re-warming phase of the protocol. This must be permitted by the informed consent obtained in step 1.
8. Construct a WBE from the scan data obtained in step 5, using the experimental algorithm. Do not run this WBE.
9. Compare the WBE from step 2 with the WBE from step 8.
10. Compute the percentage difference between the two WBEs. Subtract this percentage difference from 100 and call it the “revival fidelity”. If there is no difference, we will have achieved 100% revival fidelity. A 100% revival fidelity is the best that can possibly be achieved.
This is an ethically acceptable protocol which can be used to evaluate both a scan technology and a scan-to-WBE algorithm. It will yield a result that ranges from 0% to 100%. If the focus is on the scan-to-WBE algorithm, we can use a molecular scan technology. The use of a molecular scan eliminates the concern that the scan technology itself might have resulted in information loss. By using a molecular scan and a high-quality computationally intensive scan-to-WBE algorithm, it should be possible to achieve the highest possible revival fidelity given the level of damage caused by the possibly poor quality of the cryopreservation technology. Ideally, the revival fidelity would approach 100%. Should this prove to be the case, the revival protocol could be deployed with confidence that it would faithfully revive Alcor patients with high fidelity. If the revival fidelity is significantly below 100%, and if, after sufficient work and effort, the revival fidelity does not improve but remains significantly below 100%, and if a molecular scan has been used in step 5, then the poor revival fidelity must be attributed to the poor quality of the cryopreservation carried out in step 2.
At some point, if molecular scans are used and after sufficient work on improving the scan-to-WBE algorithm has been carried out so that further improvement in the revival fidelity is not expected, it will be necessary to move forward with the revival of Alcor patients even if the revival fidelity is significantly less than 100%. The result will be a healthy patient with some degree of amnesia. Ideally, amnesia about commonly known facts (the ability to speak and understand a common language, for example) could be filled in by adjustments to the damaged memory system.
It is worth reviewing and emphasizing a few points.
First, neither of the two WBEs is actually “switched on” during the evaluation process for the purposes of the evaluation. Both WBEs are constructed, one from neurobot data, the other from scan data, for the purpose of evaluating the changes resulting from the cryopreservation process and the cryogenic stages of the revival process.
The human volunteer might have requested that the WBE constructed in step 3 be switched on. This action, if taken, will be taken solely in the context of steps 1 and 3 of the protocol, will be entirely independent of the other steps in the protocol, and will not be done for the purposes of the evaluation protocol. As such, the ethical issues involved in switching on the WBE constructed in step 3 should be evaluated without consideration of anything that might happen in any of the other steps. This action will have been taken in consequence of the wishes of the human volunteer, and must be evaluated in that context.
Second, at no point does the evaluation of the experimental revival process result in either living or even biologically functioning whole tissues. At one stage of the protocol, isolated one-cubic-millimeter tissue samples are taken, similar to isolated tissue samples that are taken today in laboratories around the world for medical testing purposes. At all other times, the experimental process works with tissues that are at cryogenic temperatures. The purpose is to evaluate the revival process, not to revive a patient.
Third, the possible use of a destructive molecular scan as part of the evaluation process is part of the evaluation of the revival process, not part of carrying out a revival. The philosophical interests of the volunteer have already been addressed by informed consent. They are very different from the possible philosophical interests of an Alcor patient who might be revived by a protocol being evaluated by this evaluation protocol.
Because the evaluation interrupts the revival protocol after it has completed all of the cryogenic phases of the process and before it begins warming the experimental subject, there is no direct evaluation of the effectiveness of the post-cryogenic phase of the revival protocol. For this reason, the post-cryogenic phase of the revival process must be evaluated separately. Fortunately, this evaluation is less complex. As discussed previously, any cracks, fractures, or other damage to membranes that would compromise cellular compartments must be repaired at cryogenic temperatures, as failure to do so would result in further damage upon rewarming. Further, selection of “good” regions and repair of “bad” regions must already have been completed, and all preparations for rewarming must have been completed.
While we cannot complete the revival protocol and evaluate the effects of rewarming on the cryogenically repaired brain, it is ethically allowable to carry out a direct experimental test of the results of warming samples of cryopreserved tissue following cryogenic repairs.
Take small samples of cryopreserved tissue following the cryogenic repairs to the tissue and rewarm the samples at the same rate as if had they stayed part of the whole, and examine the results. For example, if a randomly chosen single cubic millimeter of cryopreserved tissue was selected and rewarmed, it could be used to experimentally demonstrate what happened upon rewarming without creating any ethical risk. A single cubic millimeter of tissue is neither conscious, nor can it feel pain, nor could it exhibit any organized neurological response. Taking this to its logical conclusion, the entire legally dead cryopreserved volunteer could, following cryogenic repairs and with appropriate informed consent, be divided into separate one-cubic-millimeter-cubes, and all the tiny cubes could be rewarmed in isolation and the results analyzed. Provided these tiny cubes remained completely isolated from each other, there would continue to be no ethical risk, as each tiny cube would simply be a small piece of isolated tissue.
By these means, it is possible to carry out a complete experimental evaluation of a cryopreservation revival protocol on a human patient who has been cryopreserved using a sub-standard protocol without ever actually reviving a human patient, and while remaining in compliance with sound ethical principles.
The risk that an experimental program to develop a cryopreservation revival protocol might result in an experimental revival that “goes bad” and results in a human subject in pain and anguish can, and should be, completely avoided.
Whole Brain Emulation and Cryonics
Because neurobots and scan technologies will be essential tools for providing the information that allows us to evaluate whether or not our cryopreservation and revival methods have been successful, and because the construction of WBEs from neurobots or from scans will be equally essential tools in allowing us to evaluate the success of our revival procedures, the cryonics community as a whole has an interest in making sure that all of these technologies are developed.
Will the mainstream community develop neurobots, scan technology and WBEs in the absence of cryonics community support?
While there does seem to be research interest in the neuroscience community in developing better tools for monitoring nerve impulses, neurobots as described here have been described as “second- generation” approaches even for the nanorobotic treatment of Alzheimer’s disease, so there does not appear to be the kind of urgent driving force behind the development of neurobots and WBEs that will be needed for cryonics applications. Scan technology also does not appear to be high on the priority list of mainstream research. This suggests that the cryonics community might have to step up to the plate and facilitate the development of neurobots, scan technology and WBEs. The interests of the cryonics community are quite specific. A focused effort by the cryonics community might be decisive in consolidating the rather diffuse interests of other groups.
In other words, the hope that others will develop the specific tools that we want and hand them to us on a platter seems overly optimistic. A more realistic strategy is to assume that if we want neurobots, scan technology and WBEs, we’ll have to, at the very least, be at the forefront of their development, if not actually shoulder the major burden of their development.
It’s helpful to reiterate that the primary motivation for developing neurobots, scan technology and WBEs is to evaluate and demonstrate the validity of the revival technology that we need to develop, and not merely to satisfy the predilections of those in the cryonics community who are personally interested in uploading.
The third item in Alcor’s Mission Statement is: “Eventually restore to health and reintegrate into society all patients in Alcor’s care.”
The technology that will allow us to carry out this component of our mission is becoming clearer. We have now reached the stage where we can begin the process of planning for the revival of Alcor’s patients.
While molecular nanotechnology and nanomedicine will eventually be developed regardless of what the cryonics community does, cryobots (medical nanorobots capable of operating in and repairing cryopreserved tissue) might not be developed by mainstream science for some time. Members of the cryonics community should systematically review all of the technologies needed to revive cryopreserved patients, identify those technologies that the mainstream might not develop, and plan for their development.
The areas of greatest interest to the cryonics community include at least the following:
1. Directly funding the development of cryobots.
2. Actively promoting mainstream reasons for funding the development of cryobots, thereby securing mainstream funding for their development.
3. Actively promoting the development of neurobots, scan technology, and WBEs, and likely pursuing direct development of all of them.
4. Encouraging and promoting the investigation of non-destructive molecular scan technologies.
Just as computer simulations have proven useful in the development of other new technologies, it appears that extensive use of computer technology and computer simulations can be used to reduce the cost and speed the revival of cryopreserved patients.
It is the author’s pleasant duty to acknowledge the assistance and comments of the people who have pointed out the various flaws and defects in earlier versions, which the author has endeavored to correct. The author, of course, remains responsible for any faults that remain. The author gratefully acknowledges the comments and assistance of Robert A. Freitas Jr., Mike Anzis, Greg Fahy, Tad Hogg, Aschwin de Wolf, and Brian Wowk.
1. “On Creating New Horizons: Integrating Non-Linear Considerations To Better Manage the Present From the Future” by John L. Anderson, National Aeronautics and Space Administration (NASA); Michael Radnor, Professor of Management, J.L. Kellogg Graduate School of Management, Northwestern University; and John W. Peterson, Manager, Technology Strategy, Switching and Access Solutions R&D, Lucent Technologies Inc. (1998). [return]
2. 78% of respondents in a survey of the cryonics community thought the first cryopreserved patients would be revived by the year 2100. Swan, Melanie. (in review). “Global Cryonics Attitudes about the Body, Cryopreservation, and Revival.” In: special journal issue “Posthuman and Transhuman Bodies in Religion and Spirituality,” Ed. F. Ferrando. Springer. Sophia International Journal of Philosophy and Traditions. [return]
5. Tad Hogg, Matthew S. Moses, Damian G. Allis, “Evaluating the Friction of Rotary Joints in Molecular Machines,” Molecular Systems Design & Engineering, 2:235-252 (2017) DOI: 10.1039/C7ME00021A, https://arxiv.org/abs/1701.08202. [return]
6. Ralph C. Merkle, Robert A. Freitas Jr., Tad Hogg, Thomas E. Moore, Matthew S. Moses, James Ryley, “Molecular Mechanical Computing Systems,” IMM Report No. 46, April 2016; http://www.imm.org/Reports/rep046.pdf. [return]
8. For example, IARPA funding might be available for an objective that is simpler than reviving cryopreserved patients: recovery of any information at all from a cryopreserved human brain. “The Intelligence Advanced Research Projects Activity (IARPA) invests in high-risk, high-payoff research programs to tackle some of the most difficult challenges of the agencies and disciplines in the Intelligence Community (IC).” https://www.iarpa.gov/index.php/about-iarpa. [return]
9. This is a good approximation for most molecules, including most proteins, which fold into one of only a few characteristic three dimensional structures in a healthy person, though not for some very large molecules, such as DNA, which would require additional information to describe their three dimensional shape. Molecules like DNA would require a description of their linear information content (the information contained in their base sequence) coupled with some additional information describing their geometry—although here, too, DNA structure is often quite stereotypical (e.g., being wrapped around histones). [return]
10. 20 bits would allow selection of molecular type from a library of 220 ~ 1 million types. Informal estimates (e.g., Ellert van Koperen, 2 Oct 2014; https://chemistry.stackexchange.com/a/16952) put the number of known biologically naturally-occurring molecules at approximately 500,000 types. [return]
11. An estimated 98.7% of all molecules present in a typical human cell are water molecules. Robert A. Freitas Jr., Nanomedicine Vol. I, Landes Bioscience, 1999, “Table 3.2 Estimated Gross Molecular Contents of a Typical 20-um Human Cell,” http://www.nanomedicine.com/NMI/Tables/3.2.jpg. [return]
12. Other factors such as isotopic composition, electronic state and electronic charge of constituent atoms may require special consideration in certain instances but should not materially alter this conclusion. [return]
13. Molecules that adopt multiple functionally significant conformations could either be restored to one “standard” conformation, or additional bits could be added to specify which functionally significant conformation they had adopted. Similar considerations apply to biomolecules containing minor random atomic-level structural errors that do not affect functionality. [return]
14. Alternatively, in situ molecular scans could take advantage of the approximately 10 m2 surface area of the capillaries in the brain, which would increase the surface area over which scanning could take place by 1,000-fold, decreasing scan time to a few hours assuming local nanodevice power consumption does not become excessive. It should be feasible to design molecular machines able to carry out molecular scans able to fit into the space available in the circulatory system, although more detailed size estimates will be needed before such a conclusion can be drawn with confidence. The 1,000-fold increase in surface area should more than offset any decrease in scanning efficiency resulting from the tighter size constraints on the scanning equipment. [return]
18. Ralph C. Merkle, Robert A. Freitas Jr., Tad Hogg, Thomas E. Moore, Matthew S. Moses, James Ryley, Molecular Mechanical Computing Systems, IMM Report No. 46, April 2016; http://www.imm.org/Reports/rep046.pdf. [return]
21. MNT should enable precise, uniform warming of brain-sized objects in microseconds by the use of embedded heating elements in the cryogenically manufactured tissue. See, for example, “Cold Starting” by Ralph C. Merkle, Cryonics, November 1990, http://www.alcor.org/docs/cryonics-magazine-1990-11.txt. [return]
22. All physical objects can be described in perfect detail by a sufficient number of bits, as a consequence of the Bekenstein bound. “In physics, the Bekenstein bound is an upper limit on the entropy S, or information I, that can be contained within a given finite region of space which has a finite amount of energy—or conversely, the maximum amount of information required to perfectly describe a given physical system down to the quantum level. It implies that the information of a physical system, or the information necessary to perfectly describe that system, must be finite if the region of space and the energy is finite.” https://en.wikipedia.org/wiki/ Bekenstein_bound. [return]
24. After Alcor has discharged its current mission, that of reviving its patients, it might find that it is well positioned to carry out a new mission: that of providing backup services to its members. Indeed, after reviving current members Alcor will already have the necessary backup data for many of its newly awakened members under the scenarios envisioned here. Offering backup services as a component of the revival and reintegration package for awakened patients seems both obvious and useful to the patient. It represents a new opportunity for Alcor that could be offered to future members. Of course, backup services can only be provided if, at a minimum, a scan of the entire patient’s brain has been conducted at a sufficient resolution to support restoration. In situ repair might not include such a scan. [return]
25. Purists might note that, in the case of the backup, you will lose all the experiences and memories gained between the time the backup was made and the time the catastrophic mishap occurred. In the case of reviving a cryopreserved patient using a destructive molecular scan, there would be little or no such loss of experiences. [return]
27. As discussed in “Backups Using Molecular Scans”, a conceptually simple method for carrying out a destructive molecular scan can be described. This might avoid the technical complexities of designing cryobots, entering the circulatory system at cryogenic temperatures, etc. [return]
28. Proposals along these lines have been made before, including a draft Advanced Draft Reanimation Directive and accompanying Reanimation Preferences Addendum that were prepared as part of the efforts by the LifePact organization. [return]
32. A connectome is a comprehensive map of neural connections in the brain, and may be thought of as its “wiring diagram”; https://en.wikipedia.org/wiki/Connectome. Whether or not the “connectome” provides all the data necessary for a WBE might be viewed as a definitional matter, but it seems clear that there is information necessary for a WBE, that can be derived from the raw data provided by the neurobots, which would not fit into the usual definition of the “connectome”. This being the case, the information required for a WBE would be a superset of the connectome, and the connectome would be a proper subset of the information required for a WBE. [return]
33. The connectome, viewed as a static entity, is not sufficient to construct a WBE, as a WBE must also provide sufficient information to enable learning, including changes to the connectome. This requires additional information in the WBE, information which is not included in the connectome but should be extractable from a sufficiently long time series of neural data provided by longer-residence neurobots. [return]
36. We assume that, whatever the law might say at the time, a successful WBE of a human is, in fact, a human and deserving of all the rights that a biological human might have. [return]
37. These time frames would, of course, have to be reviewed and agreed to by the experimental subject prior to the experiment, and would constitute part of the experimental protocol that would have to be disclosed pursuant to Principle 1, informed consent. [return]
38. This would be particularly true if the failure mode was such as to cause pain or suffering to the WBE. Note that the data originally collected from the neurobots would be retained, in compliance with Principle 4. The data erased would be the “current simulation state” of the WBE sometime after the initial state. [return]
39. It is possible that it will eventually become a legal right for a person to transfer their mind from a biological substrate to a non-biological substrate. If this happens, it might become legal for a person to record their brain activities before they become unconscious, then to become unconscious, for the data describing their brain activities to be used to construct a WBE, which would then be “switched on,” and for their biological bodies then to be “switched off” without ever regaining consciousness. Should this become legal, those wishing to do this could then donate their no-longerneeded biological bodies for experimental use in evaluation of cryopreservation protocols. [return]
40. “Death with dignity” laws, while potentially beneficial for cryonics, do not change the fact that cryopreservation would still take place following legal death. They do, however, allow for the scheduling of the time of legal death, and reduce the risk of pre-mortem dementia and brain damage that might be irreversible, even by advanced future technology. [return]
42. The “before” WBE should normally use neurobot data taken as close in time to the subsequent cryopreservation as possible. Otherwise, there will be a loss of conscious experience following the time when neurobot data stops being recorded and the time that legal death occurs. In some cases, loss of this data might be seen as a benefit. Conditions for deleting undesired data could be arranged in advance with the volunteer. [return]
44. The author recognizes that the term “sufficient” has not been fully defined. The closest analogy that comes to mind is the effort involved in validating the security of an encryption function, which is likewise not fully defined. [return]
45. Robert A. Freitas Jr., “The Alzheimer Protocols: A Nanorobotic Cure for Alzheimer’s Disease and Related Neurodegenerative Conditions,” IMM Report No. 48, June 2016; http://www.imm.org/Reports/rep048.pdf. [return]