Illustration by Tom Bachtell

To alter the human genome seems a dramatic act, one that places a solemn responsibility on those who would commit it. The genome of our species is, after all, an intimate part of who we are, the core of our biology, a repository of the many millions of years of evolution that have yielded the powerful, perverse creature that is a human being.

And yet we do something rather like this when we choose a mate. The raw material for evolution, the stuff with which natural selection can play, is genomic variation, and a main source of this variation is mixing the genomes of mothers and fathers. Whether you’re a teacher or banker, a carpenter or queen, who you reproduce with affects the genome of the next generation. This typically isn’t what people are thinking about when they pair up; choosing a mate fortunately doesn’t feel like an overwhelming responsibility to the species. Many forces affect our choice of partner: proximity, attraction, social compatibility, and just plain chance. What happens to the combining maternal and paternal genomes after fertilization is even less under our control. We mate and see what happens. Meanwhile, without our really thinking about it, the genome alters a little and carries on.

At least this is how it has been for most of human evolution. But things are changing. We are now able to make deliberate choices about the genome we pass on to our children and, by extension, to the species. Two main factors are at play here. First, we know a lot more than ever before about how genes shape us. And second, the technology for making precise changes to the genome is maturing quickly, most notably with the gene editing tool CRISPR (clustered regularly interspaced short palindromic repeats), for which Jennifer Doudna and Emmanuelle Charpentier won the 2020 Nobel Prize in Chemistry.

Walter Isaacson in The Code Breaker tells the story of the development of CRISPR gene editing, which is based on a system of clever molecular scissors that bacteria have evolved to protect themselves against viruses. Doudna and Charpentier didn’t discover the CRISPR system in bacteria; that can be attributed to many scientists working in relative obscurity for decades, as Isaacson also illustrates. But they described how it functions: which protein forms the molecular machine that snips DNA, and how it can be guided by RNA (ribonucleic acid) to a precise spot in the genome. Crucially, they showed that CRISPR could be fashioned as a tool for genome editing. Isaacson, a prolific biographer and the former editor of Time, orients his book around Doudna, the Berkeley biochemist who co-led the team that showed exactly how CRISPR works and who has been part of the effort to understand and guide its societal impact.

Though scientists have been co-opting natural molecular scissors to cut DNA since the 1970s—these tools are at the foundation of modern biology—the CRISPR scissors are vastly more flexible and make it much easier to target any specific DNA sequence. The technology is still being developed, and the term “CRISPR” now roughly refers to a whole phalanx of molecular snippers and nickers and cut-and-pasters that all do slightly different things. Together they add up to a relatively easy and precise way of deliberately making a change to any genome, including that of a human.

“Deliberately” is an important word here. Not only do we indirectly shape the human genome through our choice of mate, we also affect it through other common activities. Every major medical advance—antibiotics, vaccines, sterile surgical and obstetric techniques—influences who lives to reproduce and who doesn’t, and so, indirectly, alters the genome. Certainly wars, pandemics, and education do so too. What is radically new is that we begin to have the option to change the human genome deliberately. For better or worse, our species can turn its inexorable will to control nature back onto itself. Eventually, we could guide human evolution. Does this mean that we should?

In November 2018 news broke that He Jiankui, a scientist at the Southern University of Science and Technology in Shenzhen, China, had taken steps in this direction, using CRISPR for the first deliberate, heritable modification of the nuclear genome of a human being. (The much smaller human mitochondrial genome had already been modified, in 2016, as part of mitochondrial replacement therapy.) He Jiankui and his colleagues started with embryos resulting from in vitro fertilization (IVF) between couples in which the man was HIV-positive. They used CRISPR to edit the sequence of the CCR5 gene, which encodes a protein that the virus latches onto and that it needs to infect human cells. The scientists intended to change the CCR5 gene in the embryos to a variant that encodes a protein to which the virus can’t attach, which renders some people naturally HIV-resistant.


The team implanted CRISPR-modified embryos into their mothers and eventually reported the births of three CRISPR-edited babies. Because the editing was done when the embryos were very early in their development, before the precursors to sperm and eggs had formed, the change was heritable; these CRISPR-edited humans, if they eventually procreate, will pass it on to their children.

The biology research community received this news with shock that bordered on horror. First, there was no medical need here, since a combination of sperm washing (separating sperm from semen) and antiretroviral drugs would have reduced the likelihood of an HIV-positive embryo to virtually zero. Second, we are still not certain just how precise CRISPR is. The human genome has many millions of base pairs, the building-blocks of DNA. It’s a lot to ask that a molecular scissors like CRISPR snip at only the right spot, and that the cell then deal with this snip in exactly the right way. While CRISPR could one day be precise enough to do this safely in embryos, the consensus view in 2018—as it remains now—was that it was not technically ready. Finally, there could be unexpected consequences of making even this desired genomic change. The CCR5 variant that the scientists wanted to introduce is rare and may have negative consequences, such as increasing susceptibility to other viruses.

As it happened, the scientists did not succeed in producing the CCR5 variant they were aiming for, but rather introduced new ones, the consequences of which are not yet known. On several counts, then, the work amounted to experimenting on humans without a commensurate need. He Jiankui and his colleagues also did their work largely (though not entirely) in secret, ignored the view within the research community that the technology was not ready, and didn’t follow the correct procedure for obtaining informed consent from the parents. Though heritable genome editing was at the time not apparently illegal in China, as it was and remains in many countries, the scientists did break other Chinese medical research laws. He Jiankui lost his prestigious research position, was sentenced to three years in prison, and was fined heftily by the Chinese government.

In Editing Humanity Kevin Davies, a former editor of the journal Nature Genetics and the current executive editor of The CRISPR Journal, focuses on the powerful personalities—the scientists, entrepreneurs, doctors, and commentators—who shape human genetics and genome-editing research. The book is a prolix, gossipy account of modern genetics, expert but sometimes overexuberant, swerving into numerous asides that a reader new to the subject might find difficult to navigate. Isaacson’s Code Breaker covers similar ground but is more deftly written, conveying the history of CRISPR and also probing larger themes: the nature of discovery, the development of biotech, and the fine balance between competition and collaboration that drives many scientists.

Though Isaacson makes clear that working out the mechanism of CRISPR, like most major discoveries in biology, cannot be attributed to one or even a few people, he focuses on the combination of serendipity, knowledge, and instinct that made Doudna such a vital figure. He tells of her knack for formulating rewarding research questions and of how her decades-long work on the molecular structure of RNA gave her the expertise to crack the RNA-based CRISPR. Isaacson writes of Doudna’s mild mid-career dissatisfaction with pure research, which may have made her particularly receptive to the idea of a technology that could have a direct impact on human health.

Both books also devote several central chapters to He Jiankui, discussing his motivations and leaving a complex impression of his ambition, naiveté, patriotism, and vainglory. Yet whatever his motivations, the problems with He Jiankui’s experiment define the minimum conditions for deliberately changing the human genome. There must be a genuine need and a lack of alternatives. The technology must be safe. And we must know enough about the desired change to be reasonably sure that making it won’t have unexpected consequences.

The most immediate reason for heritable genome editing is to make changes to a single gene in an embryo in an attempt to avoid disease. An alternative technology, also advancing quickly, is IVF coupled with preimplantation genetic testing (PGT). In this procedure, one or more cells are removed from an early in vitro–fertilized embryo without harming it. The genome in these cells is then analyzed for variants that cause disease, and only embryos without those variants are implanted into the mother (or a surrogate). IVFPGT can be used to screen embryos for several conditions, including Tay-Sachs disease, cystic fibrosis, Duchenne muscular dystrophy, and thalassemia—all of which cause great suffering. Carriers of such a disease will typically pass it on to some, but not all, of their children. With PGT, prospective parents can tell which of their embryos dodged it. The first baby selected with PGT was born in 1990, and the technology is now legal in most countries, with some exceptions. IVFPGT for gender selection is illegal in India, Canada, China, and the UK, for instance, though not in the US, where regulation is generally lighter than elsewhere (in some US clinics, it is even offered for the selection of eye color). In 2016 about a fifth of all IVF cycles in the US involved some form of PGT.


There are thousands of diseases caused by changes in just one gene. Whether CRISPR editing is needed for such cases is debatable, since IVFPGT could be used instead. But often, prospective parents must go through several rounds of IVFPGT, at great financial and emotional cost and at some health risk to the woman, and even then it’s not guaranteed that IVF will yield embryos that are both healthy and have the desired gene variant. Using CRISPR to fix a disease variant could eventually be much simpler. There are also very rare (some would say vanishingly rare) situations in which, even for a single-gene disease, a couple cannot have a healthy, biologically related child without editing out the disease mutation.1

To be minimally safe, CRISPR should make only the desired change but not undesired ones, which could increase the risk of other diseases, like cancer. Safety concerns about CRISPR are being actively addressed by researchers; it’s quite likely that they will eventually be solved. Although the ethics of embryo editing are widely discussed, much further advanced is the development of CRISPR to repair faulty genes in people well after they are born. In this somatic gene editing (“somatic” means body cells, as opposed to eggs and sperm, which are germ cells), a genetic change is made to the person who receives the treatment and isn’t passed on to descendants.

The first clinical trials for somatic gene therapy with CRISPR are already in progress. They target diseases caused by errors in a single gene, like Leber congenital amaurosis, a hereditary form of blindness, and blood disorders like sickle cell anemia and beta-thalassemia. In December 2020 researchers announced that ten patients treated for these blood disorders with a CRISPR-based therapy were living free of transfusions and disease-induced crises; in this case the therapy appears to be not only safe but a cure.

CRISPR is also being put to work in immunotherapy, a form of cancer treatment in which the body’s own immune cells are engineered to target tumors. In the first reports of safety trials, also from 2020, three people with advanced blood cancer were treated with their own CRISPR-edited immune cells and experienced no ill effects over long periods. These types of efforts, and the technological development surrounding them, are the leading edge of therapeutic gene editing with CRISPR and are where at least some of the safety problems will be worked out. This certainly has its challenges; gene therapy based on pre-CRISPR technology took several decades to be realized. But once CRISPR-based therapy is dependably safe, its efficiency and flexibility will create complex ethical scenarios.

How the genome shapes human traits and diseases is one of the central questions of modern biology and directly determines what we can (and should) do with CRISPR, as well as with IVFPGT. Human genetics is hugely complex, and our ability to predict whether a particular genetic change will have unexpected consequences is very incomplete. This is one of the main reasons we should proceed with heritable genome editing with profound caution, if we proceed at all. Unlike the single-gene conditions I’ve been discussing, the genetic risk for many diseases is linked to the combined contributions of dozens or hundreds or possibly even thousands of genes. This is true of diabetes, cancer, and neurodegeneration, for example, and also of traits like height or intelligence (as measured by IQ).

Jennifer Doudna

Jennifer Doudna; illustration by Tom Bachtell

Most of these complex conditions are not only genetically determined—environmental factors also play a considerable part in how they manifest. To make things more complicated, the effects of some gene variants depend on other gene variants or the environment or both. So there could be hundreds of genes underlying some human diseases and traits, and many of them could affect one another. Without enormous advances in our understanding, the notion that we can predictably edit the human genome to alter complex multigene traits is, at best, wishful thinking.

Jamie Metzl, a self-described technology futurist and former member of the National Security Council and State Department, argues in Hacking Darwin that hubris is an inextricable part of human nature and that if we are capable of engineering human beings, we will. Though I disagree with some aspects of this provocative book’s sensibility—the genome as hackable computer code, competition as the dominant human instinct, the good life as the individualistic achievement of maximum potential—here Metzl makes a good point.

The technology for IVF was controversial when it was first developed, but finding solutions for infertility is proving a powerful emollient. The European Society of Human Reproduction and Embryology estimates that by 2018, about eight million people had been conceived in vitro since the birth of the first, Louise Brown, in 1978. In 2018 alone, about 2 percent of all births in the US—more than 70,000—were after conception in vitro. Our willingness to accept so new and unusual a process for the fundamental act of reproduction suggests that heritable genome editing could become widely accepted too, if it solves genuine problems and doesn’t create others. A Pew survey that assessed Americans’ views on heritable genetic editing reported that 60 to 72 percent of respondents were supportive if it was a means of avoiding serious disease.2

Certainly, the science is not standing still. Already it is possible to predict height or risk for complex conditions like breast cancer or cardiovascular disease, using scores that combine all those tiny effects of the many relevant gene variants in a person’s genome. There are at least two US companies that offer polygenic risk testing, in which an embryo is selected for implantation based not only on single-gene variants but on these polygenic scores. (One could select an embryo not only to be free of cystic fibrosis but also to have a low risk of breast cancer, for example.) Though it is still debated just how useful polygenic scores will be for this purpose, Metzl points out that our knowledge will continue to increase, with hundreds of thousands of human genomes becoming available for analysis as genome sequencing becomes cheap and even routine in some situations. Perhaps, as he predicts, embryo selection for complex polygenic diseases will one day be common. Perhaps we will eventually even use CRISPR or something like it to make changes to hundreds of genes in human embryos. I doubt this will (or should) happen in the next few decades, but in ten years or so, we will probably meet the minimum conditions of safety and predictability for editing out single-gene diseases.

Even so, there may be reasons not to go ahead with heritable genome editing. This is a central concern of Françoise Baylis’s Altered Inheritance, which focuses, more so than the other books discussed here, on its societal consequences. Baylis is a philosopher and bioethicist at Dalhousie University in Canada. With several eminent scientists, she has called in the pages of Nature for a temporary global halt to heritable genome editing while we debate, as a society, whether or not to do it. She offers an authoritative, comprehensive guide to the ethical issues around CRISPR, and her central message is clear: heritable human genome editing shouldn’t be treated as inevitable, and the decision to undertake it should be a collective one. She takes to task scientists who believe they need not answer for the societal consequences of their research and argues that we should adopt heritable genome editing only if it results in a more just and equitable world.

As Baylis illustrates, this is not the most likely outcome. Genome editing is almost guaranteed to increase social inequality. Like most advanced medical procedures, including IVFPGT, it will be expensive. The first (non-CRISPR) gene therapy to be approved in the United States, Luxturna, designed to correct a hereditary form of blindness, was put on the market at $850,000 for both eyes. Even if editing embryos proves cheaper than this type of somatic gene therapy, and no matter the health insurance models, it will be affordable to only a few. This is nothing new—medical care is distributed grotesquely unequally, both globally and within many countries, including the United States. But crucially, and unlike antivirals or drugs against cardiovascular disease, in the case of heritable genome editing, economic inequality would over time be inscribed in the genes. This alone should give us sustained pause.

Another potential consequence of deliberately changing the human genome is that it could increase intolerance of weakness or perceived flaws. Baylis uses the often-cited example of deafness. Since many forms of it are caused by single-gene mutations, CRISPR could make it possible for some deaf parents to have children who can hear. This isn’t theoretical—not only has CRISPR editing been shown to restore hearing in deaf mice, but Denis Rebrikov, a scientist at a major IVF clinic in Russia, is seeking to use it in embryos to reverse a common gene variant causing congenital deafness in humans.

For some deaf people and their advocates, this is objectionable. Contrary to being a handicap, they argue, deafness is a way of being in the world that opens up possibilities and creates human community and culture, including language, that its members wish to preserve. Congenital deafness is relatively unusual in that it can have both a single-gene origin and has a valued culture; such concerns may not directly apply to many other conditions (though achondroplasia, which causes one form of dwarfism, is another, and similar arguments are made by some people who have it). But the important point is that deaf or other “disabled” people may perceive more clearly than the rest of us that the definition of normal is in part socially constructed and that there are losses when we narrow it. If we’re not careful, we could wield a tool like CRISPR to make people more alike, the species more genetically uniform and biologically vulnerable, and our societies harsher and more intolerant than they already are.

It would be hard-hearted (and incongruent with modern medicine) to argue that we should conserve, in the name of societal diversity, gene variants that cause great suffering or drastically shorten life. The ethical concern is rather that, once we’re technically proficient at altering ourselves genetically and have become psychologically accustomed to doing it for serious diseases, we’re likely to want to extend genetic cures to milder conditions (like deafness), and from there to making changes that aren’t related to health at all. Many human traits one might consider modifying—height, intelligence, empathy, musical or athletic or literary talent—are almost certainly affected by hundreds or thousands of genes that we don’t understand and that interact in complex ways with one another and with nongenetic factors, so that even the technical possibility of using genome editing to alter such traits is well in the future, if we can get there at all. (There are exceptions: for instance, a single-gene mutation improves the oxygen-carrying capacity of the blood, a genetic advantage that some world-class athletes have by chance.)

These concerns converge in the title of the philosopher Jonathan Glover’s 1984 book, What Sort of People Should There Be? Glover recognizes that most of us recoil from the idea of changing our genes for the purpose of changing our nature. (This, he points out, can be true even if the change were in principle for the better—toward more empathetic people, say.) But he also makes the challenging observation that it takes an almost willful blindness, a deliberate ignoring of human affairs, to hold the view that humans could not be better than they are. Indeed, many of us find the notion of human improvement untroubling when it comes to nongenetic methods: education for instance, or hours at the gym.

Perhaps we see genetic improvements as inherently more troubling than nongenetic ones because they are relatively new and unfamiliar. Perhaps this is a legacy of eugenics, or of the belief in a divine creator. Perhaps, as the Harvard philosopher Michael Sandel argued subtly in The Case Against Perfection (2007), it’s because we have an instinctive appreciation of giftedness, because we value the notion that some part of every person’s abilities must be attributed outside of ourselves, to chance or nature or God. Also important are the principal corollaries to the question in Glover’s title: Who decides what sorts of people there should be? And is it acceptable that some may decide this for others?

In the early twenty-first century, it’s hard to imagine leaving such decisions in the hands of the state. An alternative, more compatible with our capitalist societies, is what has been called a “genetic supermarket,”3 in which prospective parents are offered a menu of genes and traits for their children by reproductive clinics. Some see this as eugenics by the back door. Others argue that parents have a moral duty to give their children the genetic traits that will contribute to their greatest well-being, and also to that of others. In between is the position that choosing from a genetic portfolio resembles decisions parents already make for their children, such as not exposing their fetuses to alcohol and influencing their children’s nutrition, education, and emotional development in the formative years.

While I’m in favor of avoiding the genetic determinism that treats a genetic change as inherently graver than a nongenetic one, it’s powerfully unsettling to imagine that human evolution could be driven in part by a kind of consumerism, a force so vulnerable to fashion and to selfish or herd behavior. Loving parents are hardly likely to choose their children’s genes in the same way they choose blue jeans, but even thoughtful choices could trade off benefits between the individual and the collective: imagine, as the ethicist Jonathan Anomaly and his colleagues have suggested, a society consisting only of extroverts.4 Also, if such a supermarket were largely profit-driven, there could be no guarantee that widespread and enduring human well-being would be the only or even the dominant goal.

We now face the questions of whether and how to proceed with heritable genome editing, and of who will contribute to the answer. Many scholarly groups have addressed these issues—most recently an expert panel convened by the US National Academies of Sciences and Medicine and the UK Royal Society, which issued a report in September 2020—with three essential messages. First, CRISPR has not been shown to be technically safe enough for heritable genome editing. Second, a path to such use for single-gene diseases could be envisioned once technical problems are solved, but should at first be for only the rare people who could not otherwise have a healthy, genetically related child. Third, decisions of whether or not to use heritable genome editing should be made by individual nations, based on the views of their citizens, and there should be national and global governance of these activities. The report is careful not to deliver judgment on whether heritable genome editing should be permitted, but strikes me as simultaneously conservative and radical on this point. If one assumes that heritable human genome editing is inevitable, the recommendations are extremely cautious. Yet in laying out paths to the clinic, an assumption of inevitability is made.5

The books discussed here are, sensibly, not prescriptive on this central question. Each acknowledges the need for broader discussion. Isaacson, like his principal subject, Doudna, eventually comes to the position that heritable genome editing is, on balance, a force for good in its promise to eliminate certain diseases. Davies and Metzl also see human genetic engineering as essentially positive, and unavoidable in some form. Metzl points out that a country like China, with a relatively lower emphasis on individual privacy and autonomy and huge investment in genome sequencing and big data analytics, has a significant advantage when it comes to assembling the genetic data needed to understand how genes shape complex traits and diseases. To avoid a genetic arms race, he counsels citizen engagement and global governance of human genetic engineering. (He is on the WHO panel working on guidelines for such governance.) Baylis takes a slightly different view. She emphasizes that only with a broad societal consensus, inclusive of groups who would not normally be involved in such decisions, should we go ahead with heritable editing of the human genome. The challenge of the moment, as she sees it, is building global and local ways to enable such a consensus.

Heritable genome editing won’t be widespread soon, and its consequences won’t manifest overnight. We still have the opportunity to think about what these consequences might be, whether we can shape them and are willing to face them, and if not, how to change course. We must determine not only whether we think it right or wrong to alter the genome—and direct human evolution—but also how, in our societies where power is concentrated and trust is weak between the few who decide and the many who bear the consequences, we might find the wisdom and the will to do it well.