Playing it Safe?Precaution, Risk, and Responsibility in Human Genome Editing
Human germline genetic modification has long been a controversial topic. Until recently it remained largely a hypothetical debate: whether one accepted or opposed the idea in principle, it was not only too risky but impractical to execute in reality. With the advent of genome editing technologies, however, heritable modifications to the human genome became a much more concrete possibility; nonetheless, the consensus has to date remained that human heritable genome editing is not yet safe enough for clinical application. The announcement of the birth of two genome-edited babies in late 2018, therefore, was condemned almost universally as premature, irresponsible, and dangerous. But what does responsibility require, and from whom? How should risk and precaution be balanced in assessing heritable genome editing, and against what alternatives? This paper reexamines commonly held assumptions about risk and responsibility with respect to human genome editing and argues that the precautionary approach that has so far been favored is not well justified, that the risks of heritable versus somatic genome editing should be reassessed, and that a fuller account of responsibility—scientific, social, and global—is required for the ethical governance of genome editing.
On November 26, 2018, the world awoke to the news that genome editing had for the first time been used to create genetically modified human [End Page 111] beings. He Jiankui, a scientist then employed by Southern University of Science and Technology of China, Shenzhen, announced via social media and the popular press that he had performed genome editing on embryos with the aim of disrupting the CCR5 gene in order to induce immunity to HIV, implanted the embryos, and that twin girls had been born.
In the wake of this announcement, arguments, claims, and accusations flew. Most commentators declared He's actions "irresponsible," pointing to various statements by scientific organizations agreeing that human embryo genome editing was not yet at an appropriate stage for clinical use. Another common criticism was the lack of transparency surrounding the work, and that He failed to follow appropriate governance procedures with respect to consent and ethical review. Many also condemned the experiment on the grounds that the babies had been exposed to unjustified risk relative to the possible benefits.
This case, then, was almost universally regarded as an unfortunate occurrence. Nonetheless, in exposing some potential fault lines in our thinking about human genome editing with respect to risk, harm, and responsibility, it provides a timely opportunity to reexamine these concepts. He's actions may have been a failure of responsibility—but responsibility to whom and for what? What does it mean to "be responsible," "act responsibly," or "take responsibility" with respect to human genome editing? In addressing these questions, I will argue that in dealing with human genome editing, the following elements are required: an approach to risk and responsibility that goes beyond precaution; a more robust conceptualization of scientific responsibility and what it requires; a reconsideration of the nature and distribution of harms and benefits of genome editing; and attention to collective social responsibility and its global dimensions with respect to the development and governance of science.
Precaution and Responsibility in Genome Editing
The generally accepted approach to human heritable genome editing (HGE) so far has been cautious, if not outright precautionary. A recent review of over 60 statements on the subject found that most "were expressly against heritable genome editing at the current time" (Brokowski 2018), with both the risk of known possible harms and uncertainty and the possibility of unforeseen adverse effects being prominent concerns. Even those that were willing to entertain the possibility of HGE in principle, notable among them the US National Academies (2017) and the UK's Nuffield Council on Bioethics (2018), agreed that it should not be permitted in practice until issues of risk and safety had been adequately addressed, and that this was not yet the case. In light of these concerns, it seems justified to label He's actions as premature and irresponsible. But what level of caution in relation to genome editing is appropriate, and is a precautionary approach warranted? [End Page 112]
Early formulations of the precautionary principle emerged in relation to environmental policy. As originally articulated, it was intended to justify taking action to forestall harm: "Where there are threats of serious or irreversible damage, lack of scientific certainty shall not be used as a reason for postponing cost-effective measures to prevent environmental degradation" (UN Conference on Environment and Development 1992). Subsequent iterations shifted the focus of the principle to a presumption against doing, by placing the burden of proof on the would-be doer to demonstrate safety (SEHN 1998). This form of the principle, in which precaution is the opposite of pro-action, is the most common characterization in bioethical debate (Glannon 2002; Harris and Holm 2002). In this context, it also tends to incorporate assumptions about the relative merits of the status quo or natural, versus a human-engineered state of affairs, "a sense of nature's ultimacy that precedes or supersedes human ingenuity" (Fuller and Lipinska 2014).
The Nuffield Council's report on human heritable genome editing, in an interesting reversal, entertains a more pro-actionary version of the principle itself. Considering possible harms that might be averted by genome editing in a situation where looming environmental catastrophe threatens human survival, they write that "at minimum [the precautionary principle] permits taking action in the present even in the absence of clear evidence of the likelihood of a harm that might occur in the future" (Nuffield Council 2018, 90). Nevertheless, the main approach to managing risks and uncertainties of heritable human genome editing can best be described as either that we ought not to proceed until we can be fairly sure that nothing bad will happen; or that, because we cannot be sure of this, we ought not to proceed at all, notwithstanding the Nuffield Council's exception for a threat to human survival—where, in a sense, all bets are off.
Adopting this sort of precautionary attitude to human genome editing combines (at least) two intertwined claims about responsibility. One is quasi-empirical, about where the balance of risk lies and what would therefore be a responsible course of action to take; the other is a moral claim about what our responsibility would be for the consequences of our choices in various circumstances.
To explain further, the empirical balance-of-risk claim concerns how likely it is that a change deliberately engineered by humans will produce unintended deleterious side effects. Existing genetic variation at least has history on its side, having survived thus far; what therefore are the chances that any change we make is on-balance more likely to disrupt this and produce negative side-effects? Advocates of a precautionary approach favor the idea that we ought to give nature the benefit of the doubt (President's Council 2003), a claim sometimes expressed in terms of the "wisdom of nature" (Bostrom and Sandberg 2009). Powell and Buchanan (2011), in critiquing this claim, call it the Master Engineer Analogy: the idea that "from an evolutionary perspective, the human organism is like the product of an engineering genius—a delicately balanced, completed, well-functioning masterwork" and that therefore attempts to interfere with this, [End Page 113] even if well-intentioned and scientifically-grounded, are likely to be "disastrously counterproductive" (Powell and Buchanan 2011, 7). This analogy, they argue, relies on misconceived interpretations of evolutionary theory: we should resist the presumption that intentional genetic modification is likely to be more dangerous than "unintentional genetic modification"—that is, the process of mutation, inheritance, and selection that occurs whether we like it or not. It may be that random genetic changes are more likely to be deleterious than beneficial, but this is not necessarily true of targeted modifications designed with some understanding of their likely effects.
Moreover, judgments about risk are not absolute but comparative: whether a likely consequence is better than or worse than the alternative. Precaution as a presumption against action assumes that the consequences of doing are probably going to be worse than not-doing. In the case of genome editing for serious disease, it is far from evident that this is so.
To Do or Not to Do?
The moral responsibility claim concerns blameworthiness for consequences: specifically, whether more blame should attach to those consequences we cause via our action, than those we allow via inaction. Here the precautionary approach is often expressed in terms of "letting nature take its course." By analogy with the "wisdom of nature," we might call this a "responsibility of nature" claim: that if something happens as a result of nature, we are not responsible for having allowed it.
Evaluating risk requires us to consider both how severe a harm would be and how likely it is to occur. A persistent issue with heritable genome editing is that we cannot necessarily foresee all possible harms in order to assess their severity nor their likelihood. This is not, however, by any means a problem unique to genome editing. We also cannot predict all the possible consequences of mobile phone use or non-use; of fracking or not-fracking; or of failing to stop climate change. There will be consequences either way, of a decision to use or not to use technology; why then should we focus in the case of genome editing on the possible consequences of doing, more than of not-doing?
The precautionary approach to human genome editing seems to presume that it would be worse for a harm to occur as the result of our deliberate action, than for a similar-magnitude harm to occur as the consequences of a deliberate omission. This, however, seems to conflate causal and moral responsibility. Are we failing to do what would be most responsible, in order to avoid being held responsible—that is, held to blame—in the event that something goes wrong?
In moral philosophy, the distinction between acts and omissions has been most addressed in relation to the ethics of killing versus letting die. Rachels's (1975) thought experiment of the evil uncles (one of whom kills his nephew in the [End Page 114] bath, while the other "merely" but deliberately allows him to drown) attempts to demonstrate that the same harm brought about by action and by deliberate omission is morally equivalent. Genome editing, however, is not quite analogous: this is a case where in acting to attempt to prevent one foreseeable harm, we incur a risk of causing a different harm. How then should we weigh our responsibility for allowing versus causing harm in each instance?
Perhaps a more appropriate analogy is a Good Samaritan case, in which one is presented with an opportunity to render aid to the needy. In such cases, a person who is "not responsible" for the situation, such as a passer-by, is not usually legally liable for failing to help. A person who "takes responsibility" by choosing to attempt to help, however, might subsequently be "held responsible" if their efforts to help somehow go wrong, providing they are sufficiently careless in the attempt.
The law in general attaches more responsibility to actions than omissions, a distinction that has led to some convoluted legal reasoning particularly with respect to end-of-life decisions (Airedale NHS Trust v Bland (1993) 1 All ER 821). Imposing a precautionary, rather than a pro-actionary, approach to human genome editing, however, would seem to broaden this rather narrow legal approach to acts versus omissions much more widely in order to draw the same conclusion about moral responsibility—namely that we should be considered more blameworthy for harms we cause (even unintentionally) than for those we deliberately allow.
The acts/omissions distinction certainly has legal relevance, but should it have moral relevance? Moral philosophy is quicker than law to recognize obligations of beneficence, such as a "duty of rescue," where accomplishing the rescue would not cause undue or disproportionate burden to the rescuer. Moreover, even within a legalistic frame, consider that in the Good Samaritan example, a genuine attempt to assist by acting in good faith, with the reasonable belief that one's actions would help rather than harm, is unlikely to result in either legal or moral blame if the attempt miscarries. Our evaluation of what responsible parents and scientists should do with respect to genome editing, and what they ought to be held responsible (morally blameworthy) for, ought also to take into account similar criteria of reasonable belief and intention.
Furthermore, parents are not bystanders: they do have special moral responsibility for their children's welfare. Given this, a presumption against action seems unwarranted. As the Nuffield Council (2018) report puts it, "we have to take responsibility for both acts and conscious omissions: deciding not to use an available technology and whether to discover knowledge about the genome or to intervene in it may still count as choices that engage moral responsibility" (73). [End Page 115]
Who Is Responsible?
To summarize so far, responsibility with respect to human genome editing encompasses both what it would be responsible to do, and for what consequences we should be held responsible. These two senses are entwined, in that (1) if we fail to behave responsibly, it is reasonable to think that (2) we ought to be held responsible for that failure. But there is one further element needed to connect these points: namely, (3) our having responsibility (or taking it, or being responsible) for doing the right thing in the first place.
For example, imagine a teenager charged with looking after his younger siblings for the evening; instead of watching them, he goes upstairs to read a book, falls asleep, and they set the house on fire. Clearly, he has been irresponsible and should be held to account for any harms that result. (Note also that this is responsibility for an omission, failing to act.) Alternatively, now imagine that his mother comes home earlier than expected and tells him, "Go upstairs and read; I'll watch the kids." If in this situation he falls asleep reading and, his mother failing to pay attention, the children burn the house down, it seems that he should not be held responsible.
One might argue that, despite having been "let off the hook," it would nonetheless "be responsible" for the teenager to take additional responsibility for ensuring all is well with his family and to check on them. All this shows, however, is that transferring responsibility is not necessarily simply a matter of two parties agreeing between themselves that one will be responsible and the other not; extraneous factors may still lead us to conclude that the would-be transferor retains some responsibility. Unilateral assumptions and assignations of responsibility also require cautious evaluation: simply declaring myself to be responsible does not necessarily make me so, nor can I render someone else responsible just by saying so.
In the case of human genome editing, then, who is responsible, who takes responsibility, and who should be held accountable? It is now generally accepted that scientists bear some degree of responsibility for their work, collectively as well as individually: that is, being a scientist in itself imposes certain responsibility (Ehni 2008; Jonas 1984; Verhoog 1981). We should also, though, recognize the significance of responsibility-taking as a form of asserting authority. Calling He's work a "failure of self-regulation" (Cyranoski 2018) not only takes responsibility on behalf of scientists, it also stakes an implicit claim to governance. This claim, and the associated questions of how scientific responsibility is distributed and how it is to be discharged, demand more critical scrutiny.
What Does Scientific Responsibility Require?
Scientific responsibility in human genome editing is not just a matter of individual case-by-case risk management: calling something "irresponsible" is not [End Page 116] identical with saying it is "too dangerous." The statement issued after the first International Summit on Human Gene Editing in 2015 declared that:
It would be irresponsible to proceed with any clinical use of germline editing unless and until (i) the relevant safety and efficacy issues have been resolved, based on appropriate understanding and balancing of risks, potential benefits, and alternatives, and (ii) there is broad societal consensus about the appropriateness of the proposed application.(NAS 2017)
The implication is that conducting heritable human genome editing at the present time would be irresponsible because it is currently too dangerous, but that with proper investigation it might become less so. Some have suggested that responsibility may even require us to undertake such investigation: Lovell-Badge (2019), for example, comments that "it would actually be irresponsible for us not to explore this opportunity further." This is partly because of the medical need that genome editing might help to address, an argument that parallels bio-ethical claims about moral obligations to pursue gene therapy and enhancement. Additionally, however, scientific responsibility requires complying with norms of scientific knowledge production and communication, including "high-quality experimental design," "appropriate review," and "transparency" (NAS 2017, 24). This includes appropriate standards and procedures at a community-wide level, for which a thorough understanding of the relevant science is necessary.
Assigning and judging scientists' responsibility to avoid bad behavior is relatively simple. It is clear that in carrying out heritable human genome editing, against almost all scientific opinion and without any chance for public discussion, clandestinely, and without attention to appropriate standards of clinical research design or review, He breached scientific responsibility on multiple counts. Less easy, however, is determining what positive duties scientific responsibility might impose, and on whom. Who else might bear some responsibility in this case? Although He's name is most prominently associated with the work, he did not act alone: fertility specialists must have been involved in the procedures required. Further, from reports that have since emerged, it seems other academics in the US may have been directly involved, while an additional number knew or suspected what was afoot. This appears to be more than just an individual failure on He's part.
The more difficult problem, however, is what these other scientists could or should have done in response. If peer review and ethical oversight are the expected routes via which scientific work is subjected to examination, what should scientists do about work that avoids or evades these pathways? Are scientists expected to be aware of regulatory requirements across all national contexts and to act as whistleblowers to promote enforcement? How far does the positive obligation to discourage or prevent the malfeasance of other scientists extend: might there be a collective responsibility to develop community standards and [End Page 117] mechanisms that would have enabled those in the know more effectively to act? More work is required to turn the claims about irresponsibility that have attached to He's case into a robust framework for scientific responsibility in human genome editing. Does knowing that a scientist is abrogating agreed upon standards of research and behavior make those scientists in the know culpable?
Relative Risks and Sociocultural Harms
A focal criticism of He's experiment was that the procedure was largely unnecessary and of marginal benefit, given that HIV infection can easily be avoided in other ways and can be well managed with current treatments, and that therefore exposing the babies to the risks of heritable genome editing was unjustified.
He's argument was that with an HIV-positive parent, the children were at greater than normal risk of infection; that within the Chinese health system, access to top-line antiretrovirals is not necessarily assured; and that furthermore, in China, those who are HIV-positive face high levels of stigma and social discrimination. In a context where HIV is severely stigmatized and has drastic social, not only health, implications, might the culturally relative risk be higher and therefore change the risk-benefit ratio of the treatment?
Risks can be differently materially relative—that is, the same event can have different consequences in light of different material circumstances. For example, suffering a simple cut to one's finger at home is fairly insignificant and unlikely to cause much of a problem; the same wound suffered in the jungle, without access to antibiotics and several days' journey from medical assistance, might result in infection, loss of a finger or limb, or even death. One would be justified in taking proportionately greater precautions to avoid the latter than the former. Insofar as cultural context makes up part of the material circumstances in which our decisions are taken, it may well be the case that a particular characteristic represents more of a harm in one social context than in another. In such circumstances it seems sensible that the level of risk that can justifiably be incurred in trying to avoid it would likewise vary.
One response might be that the appropriate solution to the social disadvantage of being HIV-positive is to seek to reduce discrimination, rather than to genetically engineer immunity. Disability advocates and scholars have long argued for this "social model" with respect to people with disabilities or genetic conditions such as Down syndrome (Shakespeare 2013): that the "harm" of such conditions is at least in part socially constructed, and that it is society that ought to change rather than imposing a technological solution. Nonetheless, it is commonly seen as justifiable to expose embryos in vitro or fetuses in utero to procedures such as pre-implantation genetic testing (PGT) and prenatal screening in order to avoid such conditions. [End Page 118]
Comparing responses to He's work with the practices and attitudes surrounding the use of reproductive technologies we already commonly accept, if we are to apply the "social model" to judge one, we should apply it in equal measure to the other. In both cases, we ought to give more consideration to the ways in which welfare is, as the Nuffield Council report recognizes, "to an extent, socially and historically determined." Importantly, both for this case and as genome editing develops globally, we should also acknowledge the socioculturally relative ways in which our attitudes towards different uses of technology are conditioned, and how this might differentially affect assessments of justifiable risk in each case.
Harm and Benefit to Whom?
A wrinkle in assessing the harms and benefits of HGE is that, at least as we presently envisage its application, it would need to be performed in advance of any genetic testing of the embryo, at the single-cell stage or even at the point of fertilization. If using it to increase the chances of obtaining a suitable embryo for transfer where some embryos might already have the desired genotype, as has been one proposed application, this would mean some embryos would have been exposed unnecessarily to the procedure—but we would not know which.
This is in some ways comparable to the existing distribution of risk in PGT: the embryos that are eventually transferred are subjected to blastomere biopsy only to prove that they are healthy. Insofar as this procedure might entail some risk of harm, had they been implanted without testing they might have been better off. For potential children produced from both PGT and genome editing, however, the application of the technique should be regarded as a precondition of their existence—assuming that parents would otherwise not reproduce or (more likely) would choose to reproduce in another way, meaning the creation of different embryos resulting in different future persons.
In the case of the genome-edited twins, their parents might still have chosen assisted reproduction in order to enable the sperm-washing procedure to prevent HIV-transmission, but differences in the timing and circumstances would surely have resulted in a different combination of egg and sperm, and thus a different child or children being born. And, although He's decision (on his account, in accordance with the parents' wishes) to transfer one of the embryos despite knowing that the genome editing had not worked as intended has been criticized for incurring the risk of HGE with no justificatory benefit, the alternative for that child would have been nonexistence.
The task of assigning harm or benefit to children who might result from potential genome editing procedures thus encounters the knotty philosophical problem of identity, or rather nonidentity (Parfit 1984): can we say a child has been harmed by being subjected to genome editing when the alternative, had their parents not chosen to use this process, is that a different child altogether would have been born? [End Page 119]
Identity issues raise various difficulties for reproductive law. The regulation of PGT and its acceptable uses has been criticized for inconsistency in this regard (Sheldon and Wilkinson 2004); indeed, the entire category of "wrongful life" cases is problematic in comparing the harms of existence to nonexistence. Such abstract philosophical considerations are also rather far removed from parents making decisions about their future families. For example, in the debate over mitochondrial replacement therapy (MRT), philosophers were quick to pounce on and dissect the question of whether MRT affects numerical identity and how this might influence assessments of harm (see, for example, Cavaliere and Palacios-Gonzalez 2018; Liao 2017; Palacios-Gonzalez 2017; Rulli 2017; Wrigley, Wilkinson, and Appleby 2015). Prospective parents and would-be users' views, however, centered much more on concern for the health of future children and how the parent-child relationship might be negotiated via genetic and social factors (Nuffield Council 2012).
What this means for genome editing is that solving the nonidentity problem is not our biggest concern. We should not be so caught up in metaphysical conundrums regarding future persons that we fail to cash out harms and benefits to current persons. In fact, HGE is also for the benefit of prospective parents, in enabling them to fulfil significant procreative interests (Nuffield Council 2018). Likewise, although clearly it is to be hoped that the genome-edited twins are healthy and unharmed, we should consider who else might be harmed by He's actions, or by others who might be inclined to attempt a similar exercise. Notably, if premature, reckless or "irresponsible" (in any sense of the word) application of genome editing leads to a delay in development of valuable therapies, those patients and their families who miss out on beneficial treatment will be harmed. We should therefore consider the possible dangers of an over-hasty approach, not only for eager patients who may be willing to try experimental genome editing procedures no matter how risky, but also for those who await benefit from genome editing developed in a responsible and acceptable way.
Intergenerational Risk and the Somatic/Germline Distinction
It is usually taken for granted that somatic genome editing is "less risky," and that this justifies pursuing it in preference to germline editing. This assumption, however, warrants further questioning. The risks and harms involved in germline editing may be of a different nature and differently distributed, but it is not necessarily straightforwardly the case that assessment of risks and possible harms should lead us to favor somatic genome editing.
For example, somatic gene therapy for inherited anemias involves bone marrow harvesting (a nontrivial procedure), in vitro manipulation, and re-transplantation (Dever and Porteus 2017). Each stage has its own attendant risks; there is [End Page 120] uncertainty regarding whether it will succeed overall or have some unintended adverse consequence; and during the whole procedure the patient suffers the effects of the disease. The latter can be significant: a recent case of somatic therapy, for instance, involved a young boy with junctional epidermolysis bullosa, a severe skin adhesion disorder (Hirsch et al. 2017). The treatment was a success for the patient and as an example of cell and gene therapy, but had it been possible to cure the disease at an embryonic stage instead, this would have saved the patient from enduring several years of pain and suffering, including losing 80% of the skin from his body before being treated.
Additionally, if the recipients of somatic genome editing wish to have genetically related children and avoid passing the disease on, they must resort to IVF and PGT, procedures which carry their own risks and burdens and are not always successful. These procedures are also currently not available for carrier status, at least in the UK; thus, a patient with a recessive genetic condition who has been cured by somatic genome editing would, unless their reproductive partner also has the condition or is a carrier, not necessarily have access to PGT to remove the mutation from their line of descent.
In other words, to continually prefer somatic to germline intervention is to say that each manifestation of disease in subsequent generations should be treated with somatic therapy, which in itself entails a not-inconsequential degree of risk and suffering. So even if a single instance of germline editing carries greater risk in comparison to a single instance of somatic editing, a decision over which to pursue must take into account the cumulative risks and harms of the multiple instances of somatic editing that are being traded off.
A further consideration is that the risk of germline editing falls primarily on one individual, while the risk of repeated somatic procedures is distributed amongst many, including persons who do not yet exist. We might of course say that in the case of germline editing, not only the "editee" but their potential future descendants are exposed to risk. This depends, however, on the reproductive choices made by the editee, choices that are likely to be strongly influenced by whether the genome editing procedure has worked as intended, or whether some unintended adverse consequence has occurred.
Nevertheless, it seems that there may be a relevant difference between a single risk-taking event followed by decisions that can be made knowing what the consequences in fact were, versus multiple sequential risk-taking events each involving possible negative consequences and influencing subsequent decisions. What this demonstrates is that we need a more thorough ethical, as well as scientific, analysis of risk in both somatic and germline genome editing, in order properly to be able to make decisions about when to pursue each possibility. [End Page 121]
Responsibilities to Global Society
Thus far we have focused mainly on the responsibilities that scientists, scientific communities, and parents have with respect to genome editing at the individual level. What might collective social responsibility with respect to genome editing require?
In a wider social context, we might also have obligations regarding how we use (or refrain from using), in the broadest sense of the word, human genome editing. The Nuffield Council suggests that heritable genome editing should be "permitted only in circumstances in which it cannot reasonably be expected to produce or exacerbate social division or the unmitigated marginalization or disadvantage of groups within society." We might take this a step further: if we can envisage a way in which this technology might be used that would give us a reasonable expectation of reducing social division, marginalization, or disadvantage, these circumstances might provide a reason not only to permit HGE, but a positive reason to use it. Additionally, though, we ought not to assume that it is only the direct application of genome editing itself that will lead to marginalization, or that only genome-edited humans will either be its victims or agents.
What is usually envisaged when equity concerns over genetic modification are raised is the division of society into GeneRich and GenePoor: that those who are already sufficiently advantaged to afford genetic technologies will become even better off as a result, widening existing inequities. Extreme versions of this concern envision this resulting in the fragmentation of humanity into different species, who will then inevitably turn on each other (Annas, Andrews, and Isasi 2002). This argument ignores, however, that as history unfortunately shows, humans are quite capable of creating social divisions liable to end in genocide even without the help of genetic modification. Avoiding human genome editing only solves part of this problem, and not the most immediate part: the "fix" here must be social, not anti-technological. Indeed, if the spectre of genome-edited speciation prompts us to develop better moral and political frameworks in order to avoid future posthuman genocide, this might contribute to reducing conflict and promoting more equitable treatment of all human beings today—which would certainly be a good thing!
In any case, however, there are at least two other, probably more immediate ways in which genome editing might lead indirectly to social division and marginalization. The first is via the geneticization of discourse: even before genome editing comes into widespread use, the way in which we discuss its potential and our moral obligations with respect to it might promote "geneticized" thinking, which in turn can lead to discrimination. Overbroad versions of the "moral imperative" argument, which posit genome editing as the remedy for all manner of complex biosocial problems such as health-care resource allocation and inequalities in socioeconomic status and educational outcomes (Gyngell, Bowman-Smart, and Savulescu 2019), may pose a particular danger here: it is not a very great leap [End Page 122] from urging that genes are the solution we have a moral obligation to pursue, to concluding that genes are the cause of the problem. Given how far we are from the promissory future in which genome editing does solve such problems (if indeed it ever can), this is just as likely to result in pseudo-genetic discrimination.
Second, genes aside and on a global scale, we are already ScienceRich and SciencePoor. Vast disparities exist between countries in terms of scientific capital: those countries with more advanced scientific capacity also tend to dominate ethical and regulatory discourse and wield greater influence in determining the global norms of science. Looking at the recent history of human genome editing as well as other emerging technologies, there is a very real possibility that the trajectory through which human genome editing is realized, including the discourses surrounding and shaping its realization, will increase inequity between countries and worsen divisions amongst global scientific communities (Chan, Palacios-Gonzalez, and De Jesus Medina Arellano 2017). Even before He's announcement, China's human genome editing endeavors were already disadvantaged by stereotypical perceptions of a "Wild East" deficient in ethics, regulation, and scientific norms (Ho 2016; Sipp and Pei 2016). Regardless of whether "objectively" justified or not, such perceptions are also a product of existing scientific capital; reiterating them, whether as justification or critique, further reinforces inequalities. We must attend to the effects of this in considering how we move forward with, and shape discourse around, human genome editing and its governance at a global level. Discussions of genome editing must therefore move us to new modes of global ethical discourse and scientific governance, not only because of the technology's potential to cross borders, but because of its role as a paradigmatic example of emerging, ethically contested science.