The Saturation View
A new theory of population ethics
This article was created by Forethought. Read the full article on our website.
In collaboration with Christian Tarsney, I’ve developed a new theory of population ethics, which I call the Saturation View. I think that, from a purely intellectual perspective, it’s probably the best idea I’ve ever had. It was certainly great fun to work on.
The motivation is that many views of population ethics, like the total view, suffer from some major problems. Some are already widely discussed:
The Repugnant Conclusion: For any utopian outcome, there’s always another outcome containing an enormous number of barely-positive lives that is better.
Fanaticism: For any guaranteed utopian outcome, there’s always some gamble with a vanishingly small probability of an even better outcome that has higher expected value.
Infinitarian Paralysis: Given that the universe contains an infinite number of both positive and negative lives, no finite or infinite change to the world makes any difference to overall value.
These are pretty bad!
But there’s another less-discussed problem, too.
The Monoculture Problem
What would the best possible future look like? Essentially all extant views in population ethics give the same, surprising answer: create a monoculture. Find whatever life or experience generates the most value per unit of resources, then produce endless identical copies of it.
This implication has received remarkably little attention from philosophers. But I think it’s maybe as bad as any of the other problems listed above.
Consider two possible futures:
Variety: A vast population of individuals leading very good lives, extraordinarily diverse in form, personality, interests, and accomplishments. No two individuals are identical. Inequality is limited — all lives are very good.
Homogeneity: The same vast number of individuals, but each is a qualitatively identical copy of the best-off person in Variety.
Intuitively, Variety is better. A future containing only one life-type, repeated as many times as physics allows, feels impoverished — like a song with only one note.
Yet virtually all existing population axiologies prefer Homogeneity. Total utilitarianism does, because Homogeneity has higher total wellbeing. Average utilitarianism does too. Critical-level views do. Even egalitarian views prefer Homogeneity — it’s perfectly equal!
This follows from two principles that nearly all views accept: Pareto (if everyone is at least as well off, and someone is better off, the outcome is better) and Anonymity (only welfare levels matter, not who has them). Together, these entail that Homogeneity beats Variety. So essentially all extant impartial accounts of population ethics suffer from the monoculture problem.
What’s more, future technology will allow us to copy minds perfectly and search for maximally welfare-efficient designs. If so, standard axiologies recommend essentially producing just one optimal life-type as many times as possible. Endless galaxies containing nothing but the same blissful experience, repeated and repeated, would be the ideal.
The Saturation View
In light of these problems, I propose a new axiology: Saturationism. It's able to deal with all four of the problems I listed using the same basic machinery.
The core idea is that experiences1 come in different types, defined by their qualitative characteristics — hedonic tone, complexity, representational content, and so on. These types form a kind of landscape, where similar types are closer together and dissimilar types are farther apart. When an experience comes into existence, it contributes intensity to its location in this landscape and to nearby locations.
The realisation value of a type is determined by both the wellbeing of the experience and by how many very similar experiences already exist. A region’s contribution to overall value is a concave function of the welfare-intensity at that region: the first instances contribute substantially, but additional near-duplicates contribute progressively less, approaching but never quite reaching an upper bound. A world’s total value is the integral of these contributions across the entire landscape.
Here’s an analogy. Imagine the space of possible experiences as a colour wheel, lit from above by an array of tiny lights. Each point on the wheel represents a possible type of experience — its hue corresponds to its qualitative character. When an experience comes into existence, it adds current to a light pointed at its location, illuminating that region.
Crucially, illumination is a concave function of current: the first instances make a region noticeably brighter, but additional near-duplicates contribute progressively less. There’s an upper bound on brightness that can never quite be reached.
A world’s value equals the total illumination across the wheel. On this view, Homogeneity concentrates all welfare in one region, lighting up only one small area. Variety illuminates the whole spectrum.
This structure makes diversity intrinsically valuable. Spreading welfare across many dissimilar types means each experience contributes at a steeper part of the concave curve, yielding more total value than concentrating the same welfare among near-duplicates would.
At small scales and with diverse experiences, the view behaves just like the total view. But at very large scales, the value of variety kicks in: it becomes increasingly less valuable to create an additional near-duplicate of some experience that has already been instantiated millions of times, and comparatively more valuable to create some wholly new form of positive experience.
Dissolving the Repugnant Conclusion
The classic path to the Repugnant Conclusion requires trading a utopian world for an enormous population of barely-positive lives. More precisely, the Mere Addition Paradox arises from three intuitive principles: that adding well-off people and improving existing lives is good (Dominance Addition), that more equal distributions with higher average welfare are better (Non-Anti-Egalitarianism), and that some sufficiently excellent world can’t be beaten by any world of barely-worth-living lives (Denial of the Repugnant Conclusion).
Once we accept the value of variety, we should reject the unrestricted versions of the first two principles — they fail when the “improved” world has much less variety. But we can accept variety-restricted versions.
Crucially, these restricted principles don’t generate the Repugnant Conclusion. To reach Z-world from A-world, you’d need a more equal, higher-average population that’s equally diverse while consisting wholly of barely-positive lives. But, on the Saturation view, barely-positive lives can only illuminate a tiny corner of the landscape. So no such world exists. The path to the Repugnant Conclusion is blocked.
Avoiding Fanaticism
Total achievable value is bounded above — there’s only so much experiential terrain to illuminate. That means no tiny-probability gamble can have arbitrarily high expected value.
Infinite Ethics
On Saturationism, the value of a world is finite and well-defined in any infinite universe — even if some locations have infinite wellbeing. Saturationism also discriminates between many infinite worlds that (for example) totalism treats as equivalent: a world that illuminates more of the landscape is better than one that illuminates less, even if both contain infinite welfare. What’s more, unlike other approaches to infinite ethics, it does not need to invoke the spatiotemporal structure of the universe or require a choice of ultrafilter, and therefore it avoids the problems that other do.
Separability
Like nearly all non-totalist views, Saturationism is non-separable — background populations can affect how we rank options. But this is a feature, not a bug. The value of variety just is an intuition that the correct axiology is non-separable.
Moreover, the violations are comparatively tame. If two populations have non-overlapping footprints in experience-space, their values simply add. At small scales, Saturationism approximates total utilitarianism. It’s only in unusual situations involving vast populations of near-duplicates that the totalist approximation fails.
Extant issues
There are still a lot of unresolved issues for Saturationism and, like any population axiology, it has unintuitive implications. Most importantly, the view’s implications in some highly-negative worlds are hard to stomach, though I think similar implications are unavoidable for any view that avoids fanatical implications.
Conclusion
If the Saturation View is right, then the best future isn’t the one where we’ve found the optimal experience and copy-pasted it across the cosmos. The best future is the one where we’ve gone exploring — where we’ve fully lit up the landscape of possible experiences. Not a single note, but a symphony.
This is a summary of a longer and more detailed write-up of Saturationism, which gives a “toy” version of the view to illustrate how it works before stating the full version formally. The full paper, with Christian Tarsney, is still work in progress.
I’ll focus on experiences, though the view could be defined in terms of lives or other “welfare events” (like instances of preference-satisfaction, achievement, and so on).
It's not true that views which put all wellbeing on equal footing imply that Monoculture is better than Variety. Monoculture would be a version of the experience machine, where everyone's relationships are 100% fake (since they're all qualitative duplicates of the best-off person in Variety, that means that none of the other people from Variety exist, so all their experiences of having a relationship with those other people must be fake). On either a preference or objective list theory of wellbeing, this is going to leave them much worse-off than the people in Variety. On the other hand, if you try to modify the scenario so that people in Monoculture aren't radically deceived about everything in their entire lives, then there's no longer any reason to think they'll be happier than the average person in Variety - most likely, they will be much less happy, since they live in a much more boring world and don't meet anyone who's different from them, so they get no interesting social interactions. I think this latter case is what leads to the intuition that a diverse world is much better than a homogenous one - it's because it's better for the individual people in the world for it to be diverse, not because the distribution of types of wellbeing matters for its own sake.
I also think the saturation view has its own version of the Repugnant Conclusion, which is much, much worse than the original version. You have to accept that an arbitrarily small increase of wellbeing for a single person could outweigh an arbitrarily large increase of wellbeing for arbitrarily many people as long as the latter was a type of wellbeing that had been experienced many times already and the former wasn't. On this view, it would be better to give one person a single M&M, rather than to give infinitely many people lives of perfect bliss (who currently live lives not worth living) as long as no one has ever eaten an M&M or something similar to it before, while many people have already had lives of perfect bliss of the same sort you're going to give to the infinity people. So we still end up with a situation where it could be better for lots of people to live lives barely worth living (or even not worth living at all), rather than everyone living lives of perfect bliss, except that now the tradeoff is not a massively larger population, but a single person experiencing the tiniest of pleasures. If negative wellbeing is handled the same way as positive wellbeing, you can get even worse implications - it could be good to take infinitely many people living lives of perfect eternal bliss, and instead send them to lives of eternal torment, all so that a single person can eat an M&M, so long as the space in the wellbeing landscape for the particular kinds of bliss and torment in question are already saturated by other people.
It also doesn't solve infinite ethics, but makes it much worse. On the Saturation view, if infinitely many people have already had some kind of experience, then its marginal value goes to zero. So in an infinite world, nothing would matter at all - every type of wellbeing is already saturated, so any marginal change in wellbeing makes no difference whatsoever. You can go around murdering people, and it will make no difference at all to how good the world is.
I also think that, structurally, this theory just doesn't match up with what really matters. There's a common objection to utilitarianism that says that it ignores the separateness of persons, and while that argument is complete nonsense when it comes to traditional versions of utilitarianism, which treat every individual's wellbeing as a distinct token of goodness that matters for the sake of that individual, and whose value isn't affected by others' wellbeing, the objection applies in full force against the Saturation view. Different individuals' wellbeing can't be treated separately here, because it turns out your wellbeing doesn't matter as much if it's a type others have already experienced. Wellbeing is no longer good for the sake of the individual experiencing it, but for the sake of making to total collection of pleasures in the world as diverse as possible.
Hi Will,
I really enjoy your thinking around philosophy and this piece specifically. However, I fear the Saturation View makes a large incorrect assumption which causes many of its benefits to disappear.
You write that "There’s an upper bound on brightness that can never quite be reached" clarifying that "Total achievable value is bounded above". I agree that a genuine upper bound resolves issues of fanaticism and the repugnant conclusion. But what motivates a hard ceiling compared to a logarithm which is also concave with diminishing returns; but unbounded? With an unbounded function this means the problem of the repugnant conclusion returns: Instead of trading n barely happy lives for a utopia, we could now trade 10^n (or whatever the discount rate implies).
I think intuitions in base utilitarianism push against such a ceiling. Regarding your analogy, it would say the brightness limit is in the eye and not the territory. The Point of View of the Universe would delight in a new happy person existing, no matter how many similar minds there are already.
This isn't to endorse pure maximizing EV utilitarianism for population ethics however. I think moral uncertainty and risk aversion can resolve some of these problems more naturally. Having diversity of minds hedges for uncertainty given both what is good and the uncertainty around sentience.
These also resolve repugnant conclusion worries in particular. Different moral theories have different definitions of "a life barely worth living". It ranges from hedonistic utilitarianism's sum Planck second of epsilon utility to Mill's hours at the opera, to something even more demanding. If you have credence across many theories and are risk averse, then satisfying "a life barely worth living" leads to a dimensional and rich life; removing the repugnancy from the repugnant conclusion's Z world.