The Best Possible World

A computer scientist, Noor Siddiqui, wants to help everyone “have a healthy baby.” Will we let her?

In 2023, the first baby was born with assistance from Orchid, a new reproductive technology startup which provides full-genome sequencing for human embryos. Orchid’s proprietary genetic testing service allows potential parents to screen their frozen embryos for over 1,200 monogenic diseases—including Huntington’s disease, cystic fibrosis, and Tay-Sachs—and compare polygenic (i.e. multi-gene) risk scores for hundreds more—including some of humanity’s worst ills, like schizophrenia and breast cancer. Parents can then implant the embryo of their choice based on the results. For the first time in human history, new children will be able to side-step thousands of diseases for which we currently have no cure.

Embryo selection is highly controversial. Debates about embryo selection tend to quickly prove Godwin’s law: that any debate allowed to go on for long enough will end in someone making a comparison to the Holocaust, or Hitler. In this case, the comparison is to the Nazi program of eugenics, via forced sterilization and mass murder. 

Most people, however, are not opposed—in principle—to using technology to improve the overall health of the human population. For one, the development of cures and vaccines for painful diseases is widely regarded as a good thing. Children used to regularly die, or become paralyzed, from polio; now, thanks to the polio vaccine, the disease has been eliminated. The world is a better place as a result. Suppose that a scientist announced today that she had invented a drug which could cure over 1,200 individual diseases, and which could eventually be used to cure 7,000 more. We would give her a Nobel Prize, not call her a “fascist.” 

Today, we live in a world in which over 400,000 children per year die from congenital diseases—genetic diseases which are present at birth. Using embryo selection, we can solve the problem of children dying from congenital diseases—by making it so that fewer children are born with those diseases in the first place. Instead, different children will be born in their place. We can use embryo selection, therefore, to bring a better possible world into existence: a world in which the prevalence of genetic disease is much lower, and where more people are able to live longer, happier lives as a result.

But while cures and embryo selection both work to make the world a better place, only cures can make particular people better-off. The key difference between cures and embryo selection is that, if curing a person’s disease can raise them from welfare level A to welfare level A+, then embryo selection can only ensure that more people are born at A+ in the future. Over time, however, at the population level, the results are the same: more people live for longer at A+. Does it matter if one of the greatest moral and technological advances of our age is technically good for “no one”?

The answer may lie in how we respond to one of the thorniest, and most emotionally charged, debates in contemporary philosophy.

The Non-Identity Problem

A fourteen-year old girl wants to have a baby. Because of her age, she will give her child a bad start in life. However, her child’s life will still be worth living. If she waits a decade in order to have a baby at twenty-five, then she will be able to give her child a much better start in life.

The girl has her baby, and predictably, the resulting person’s life is quite hard—although still worth living, as most people’s lives seem to be after they are born. Did the girl do something wrong? 

In 1984, a philosopher named Derek Parfit pointed out a problem with cases like these, where our actions will affect the identities of future people: although it seems like the girl did something wrong, she technically harmed “no one” by having a baby so young. If the girl had waited until she was older to have a baby, then the resulting child might have been better-off, but it would have been a different child: a different sperm would have fertilized a different one of the girl’s eggs, resulting in a different person being created. Her present child, then, could not have been made “worse off” by their mom’s decision to have a baby, since without it, they would not exist; nor were they conceivably “harmed” by coming-into-existence, since their life is still, on net, worth living. The question, then, is not whether the girl owed this child a better life, but whether, when procreating, she had an obligation to create the person who would have the best life possible.

Parfit’s “non-identity problem” has far-reaching implications. Prior to Parfit pointing out the non-identity problem, it seemed obvious that in order for an action to be “wrong,” it had to be bad for someone. But Parfit seemingly discovered an entirely new class of wrongful actions: actions which cause a worse-off person to come into existence in cases where a different, better-off person could have existed in their place. While these actions may make the world a worse place, they leave no particular person worse off. Are they still “wrong”?

In response to Parfit’s framing, some thinkers have doubled down on what is known as the “person-affecting” view in ethics: the position that an action can only be wrong if it is bad for someone—if it leaves at least one person worse-off (or conceivably “harmed” by coming into existence). Under the person-affecting view, there is nothing wrong with bringing a worse-off person into existence instead of a better-off one, so long as the worse-off person’s life will still be worth living. Taken to its logical conclusion, however, the person-affecting view implies that we have no obligation to create any future generations at all—since failing to procreate is technically bad for “no one.” Under the person-affecting view, if everyone on Earth were sterilized today and no one minded, the end of our species would not itself be a “bad thing.”

Parfit himself rejected the person-affecting view outright, on account that it cannot explain our apparent obligation to create well-off future generations. His most plausible alternate account of what we owe to future people is known as the “total theory”: which holds that, all else being equal, we should strive to create the future generations whose lives will contain the greatest total sum of well-being minus suffering (a view with its own counterintuitive implication). 

Under the person-affecting view, embryo screening is neutral but unimportant: good for “no one,” save for the parents whose lives will go better if they have a healthier child, and the occasional thwarted baby who was really better-off not being born. But under the total theory, any time we create a new life, we have an obligation—all else being equal—to create the person who will have the best life possible. 

Publicity Wars

Consider another case. A thirty-five year old woman wants to have a baby. She has a choice: she can freeze twenty-or-so of her and her partner’s embryos, and then use Orchid to pick the “best” one to implant. Or, she can roll the dice and conceive a child the “old-fashioned” way. The woman decides that she does not want to use Orchid. Unfortunately, she has a child with a rare genetic disease. Her child’s life is very hard, but still worth living: it is even harder, though, than the life of the child of the fourteen-year-old girl. The woman did something that billions of people have done since the dawn of our species—she decided that she wanted to have a child, and gave birth to whichever child she happened to conceive. How could she have done something wrong?

But we live in the 21st century. By not using Orchid, the woman knowingly risked creating a child whose life would be compromised by genetic disease. As a result, she lowered the expected quality of her child’s life—similar to if she had a child in poor circumstances. Technically, the woman acted just as recklessly as the fourteen-year-old girl.

Orchid’s product produces strong reactions, to say the least. In April 2024, Noor Siddiqui, Orchid’s 29-year-old founder and CEO, was the subject of a scathing profile in Wired magazine titled “This Woman Will Decide Which Babies Get Born,” in which her interviewer accused her, among other things, of making his natural-child-having friends remember Nazis and worry about “the word that starts with ‘eu’ and ends with ‘genics’.”

But the 20th century eugenics regimes were horrifying because they mandated actions that would have been reprehensible no matter what the goal: namely, large-scale medical torture, mass public humiliations, and genocide. The Nazi Party was motivated by sadism and a desire to control and dehumanize their victims—elements which are not present when parents make consensual decisions about the genetics of their future children. 

To clarify, consider a different possible world. In this world, women are naturally endowed with the ability to choose the traits of their future children. Before conceiving a child, in this world, a woman can preview the expected features of her offspring—like height, temperament, and of course, health—and then choose which of her possible children she would like to create. One day, a dictator comes into power, and he decides that he does not like the sorts of children whom his subjects are creating. He passes a law which requires all female citizens to install a device that will cause them to have random children instead. The device also blocks women from seeing any specific information about their future children before birth; although they will still have access to abortion, if they want it. After the law, the world would be in an identical state of affairs to our own: but such a program would still seem like a horrifying, tyrannical violation of reproductive freedom.

Embryo selection actually allows potential parents—who are consenting, paying customers—to enact far more control over their reproductive futures than has ever been possible before. Is it widely accepted that just because a person wants to have a baby, does not mean that they will necessarily want to have a baby under any possible circumstances. And similarly, a person who is ready to have a child may not want—for any number of reasons—to have every possible child that they could conceive. 

Reproductive interest is not a binary, or a suicide pact: we do not force people to have children in all cases; similarly, we should not force them to have all possible children without affirmative consent, when possible.

In liberal societies, we generally allow people to choose the conditions under which they want to procreate, because we assume that individuals are best-suited to make sensitive decisions for themselves and their families. In our world, parents already have the ability to choose when and with whom they want to have children; we celebrate these rights as fundamental features of a free society. Now, parents will have the ability to choose which of their possible children they want to create in addition. We should celebrate that freedom, too. 

But the Wired interviewer’s most pointed critique of Orchid was that Orchid’s technology is, in a sense, self-defeating. Siddiqui founded Orchid in order to help people like her mother, who had a traumatic experience with retinitis pigmentosa—a rare genetic disease which causes gradual blindness—when Siddiqui was growing up. But if Orchid’s technology had been invented a century earlier, then Siddiqui’s mother would not exist.

Siddiqui deferred (“I’m not trying to delete my mom”). In the interview, she explained that she does not think that Orchid would have deleted her mom, only made it so that her mom would not have had to suffer so much. But Siddiqui’s interviewer was right: if Siddiqui’s mom had appeared in a batch of embryos, and if her grandparents had screened those embryos for genetic diseases with the intention of having the healthiest baby possible, then Siddiqui’s mother would never have been born.

Of course, there are functionally infinite ways for a given person to not come into existence—none of which are in any way “personal,” in the sense of being targeted to “delete” a particular person. For one, if Orchid’s screening technology—and the requisite IVF capability—had actually been air-dropped into the world a few decades before Siddiqui’s mother was conceived, and if her grandmother had decided to use it, then, aside from her retinitis pigmentosa, Siddiqui’s mother would almost certainly not have been one of the embryos on option: because the retrieval process would have taken only fifteen-or-so of her grandfather’s (223)2, or 70 trillion, possible genetic contributions and combined them with the same number of her grandmother’s 300,000 available eggs to create an entirely novel array of possible people—with the odds that one of those embryos would have “been” Siddiqui’s mom coming out to something like one in a quintillion. 

In general, the odds of “you,” or any particular person, ever existing are jaw-droppingly, astronomically low—around one in 102,685,000, compared to the 1080 atoms in our universe. In order for you to exist, your parents had to exist, and meet, and conceive a child at that exact time and in the exact, er, manner that would cause the sperm that made “you” to fertilize the egg that made “you”: nearly impossible, since, to start, your father only had a 0.75% chance of ever producing the sperm that made “you,” and even then, that sperm was only one of ~500 billion that he produced in his lifetime. Your mom was born with 1-2 million eggs, only one of which had the right genetic contribution to make “you,” and most of her eggs were already gone by the time she reached puberty. Given that your parents existed and had a child together, then, the odds of “you” existing are roughly one in 1020 (two powers of ten larger than a quintillion). And in order for your parents to exist, the same one in 1020 odds had to hit for each of them, and for each of their parents, and back and back, so that if pretty much anything had been different about the trajectory of our universe from the Big Bang until your mom’s walk home on the day you were conceived, then you would not exist.

A Better Place

It can be difficult to separate our judgments about the potential benefits that could come from embryo selection from our (justified) fear of forced genetic “cleansing.” Most discussions of consensual genetic screening tend to imply that it is functionally equivalent to, or a short slippery-slope away from turning into, the latter. So suppose, for a moment, that the global rollout of embryo selection goes well.

Today, most of Orchid’s clients are progress-oriented, early-adopter types on American coasts. But the technology could scale much further. Imagine that, in 2040, Orchid’s screening technology is available in fertility clinics all across the United States, and is covered by insurance (Orchid screening currently costs $2500 per embryo). By 2060, Orchid is taught in sex-education classes across the country as the “responsible” way to have children; women, even those who could easily conceive a child the “natural” way, increasingly opt for IVF in order to pre-screen their embryos and have the healthiest baby possible. By 2100, 80% of all babies born in the U.S. and a significant fraction of those born in Europe come from pre-screened embryos: soon, children roll their eyes when their grandparents recall the days of “random” conception. In the 2120s, a micro-trend of having “roulette babies” arises among the ultra-rich, who flaunt their copious wealth by paying for expensive treatments for their children, but the trend ends abruptly when the child of a prominent influencer dies an extended painful death from Huntington’s disease. By 2150, Orchid is able to screen for all 7,000 known monogenic genetic diseases, and the predictive power of their polygenic risk scores increases such that insurance rates fall all across the developed world. Finally, in 2350, the Gates Foundation is horrified to learn that people in the Third World are still dying from preventable diseases, and launches an initiative to set up free IVF and embryo-screening clinics for the world’s poorest one billion people; by the end of the century, monogenic disease has been all but wiped from Earth—save for a few rare mutations found in the children of accidental pregnancies.

Consider the possible world in 3000. It is, in expectation, a much better place than the counterfactual one where human reproduction was allowed to proceed as planned. It is a world where more adults are able to live autonomously, and to pursue their dreams and passions without constraint; a world where more parents live long enough to meet their grandchildren, and where more grandchildren live long enough to see their 10th birthdays. We can think of the difference between a post-Orchid world and its alternative as similar to the difference between our world and an alternate version of 2024 in which the polio vaccine had never been invented: except that polio only killed 3,000 people a year at its peak, which is less than 1% of the number of children who currently die per year from genetic diseases.

What is standing between us and creating that world? Mostly, that world where you choose your children would be a massive disruption of the reproductive status quo. For the entirety of human history until now, people “got what they got” on children, and while child-rearing is always hard, they typically experienced great love for their children, and bore unanticipated burdens and heartbreaks as facts of life beyond human control. The value proposition of embryo selection is that some genomes will produce people who will, in expectation, have “better” lives overall (the difference lies in the expected quality of a given new life; in individual cases, there are certainly people with genetic diseases who will lead “better” lives, on any number of metrics, than other people who do not have genetic diseases).

Within an individual life, we are quite comfortable comparing “better” or “worse” outcomes: we accept that someone developing a disease is a tragedy, while someone developing the cluster of traits associated with “good genes” is a lucky break. But if we admit now, after hundreds of thousands of years of random reproduction, that we have good reason to bring “better” lives into existence instead of “worse” ones, then we would destroy the illusion that all of our previous children were, in some sense, the best ones to have. 

For the Future

The first time I read Siddiqui’s Wired interview, I was reminded of Parfit’s greatest fear when he first identified the Non-Identity Problem: that someday, people would notice that their efforts to improve the world for future generations were actually changing the composition of those generations, and conclude that they had no more reason to “help” the future at all. After all, if we are only morally obligated to help people, and not to bring about the best possible world with our actions, then there would be nothing wrong with, say, burning through all of Earth’s resources in the next century or burying slowly-decaying nuclear waste in random places around the globe—since curtailing our present-day indulgences will certainly make existing people worse-off, and “no one” in the future will be harmed if we do otherwise.

From our vantage point in the present day, embryo selection feels different from other ways that we might help the future, because it directly alters who will exist in that future at the same time that it works to make the future a better place. But in truth, there are no “default” people who live in the future, no “neutral” next generations who our present-day meddling will or will not allow to come into existence.

In fact, every action you take in the present day, from going to the grocery store to grabbing a quick cup of coffee, produces a chain of novel events—of lines made longer or shorter, of traffic made faster or slower—that would not have occurred otherwise. Eventually, that chain will alter a planned conception event, and cause a different person to come into existence than the one who would have existed otherwise. And that person’s actions will alter further conception events, and those people will go on to alter further conception events, and so on and so on, until, at some point, it is possible that every single person who exists will be different from the people who would have existed if you had done otherwise.

In particular, when we make vast, sweeping policy changes in the name of helping future generations—from “Just Stopping Oil” to funding large-scale, blue-sky R&D projects to improve public health—we almost certainly “swap out” the next generation for an entirely new group of people: since our actions affect so many people’s behavior in the present day at the same time.

As a result, when we act to “help” future generations, we likely do not benefit any particular people who will live in the future, because at the same time that our actions work to make the future a better place, they will simultaneously change who will be around to benefit.

Instead, it would be more accurate to say that we are guiding a different possible world into existence, with different people: one which we hope will be a better place overall. 

The Risk of a Lifetime

In the past decade, there has been a sharp uptick in anti-natalist sentiment—as we humans grapple with our present era’s uneasy combination of increasing existential risk and abundance. Some anti-natalists argue that we should stop having children instrumentally: namely, in order to slow the effects of climate change (although I am not sure who we are trying to stop climate change for, if not for the benefit of future generations). But others make a more daunting, and sweeping, claim: that most human life is simply not worth living, and that, despite the occasional pleasures of being alive, we should stop having children right now in order to spare our potential offspring from the overwhelming harm of coming into existence.

The most obvious rebuke to the previous point is that most people find their lives to be well worth living, after they are born. But even within a life that is overall worth living, the intervening “bad stuff” is still bad: better to never have been experienced, if possible.

Every time we create a child, then, we must take a leap of faith. We must believe that the “good stuff” that we can create by adding lives to the world will ultimately outweigh the “bad stuff”—pain and suffering—that we inevitably create by exposing our children to the risks of existence.

Today, for the first time in human history, we have an opportunity to systematically push nature’s hand towards creating lives that will contain, in expectation, more of the “good stuff” that makes life worth living and less of the “bad stuff” that makes procreation fraught. The great unfairness of the genetic lottery is that the same chemical codes that make people distinct from each other—that make me “me” and make you “you”—also set some people up to have much better experiences of living than others. But if there is nothing inherently better or worse about bringing one person into existence instead of another, then, all else being equal, it seems that we should strive to create the people who have the most to gain from coming into existence, and who will miss out on the most by never seeing this world: the people who will have the best lives possible.

If we have any obligation to make this world the best place it can be, then we ought to be more intentional about who we bring into it.