5th Oxford Workshop on Global Priorities Research

Talk Abstracts

Monday 7 December - Session A

15:20 - Teru Thomas (paper joint with Nick Beckstead), “A paradox for tiny probabilities and enormous values“

In thinking about the consequences of our actions on the long-run future, a lot seems to depend on how we should take into account tiny probabilities of achieving enormous payoffs (or preventing enormous harms). This talk will explain some of the theoretical trade-offs in this area. We argue that every transitive theory of the value of uncertain prospects is either "reckless" or "timid": either it recommends risking arbitrarily great gains at arbitrarily long odds for the sake of enormous potential, or it permits passing up arbitrarily great gains to prevent a tiny increase in risk. We draw out some of the consequences of each horn of this dilemma.

16:20 - Charlotte Franziska Unruh, “The constraint against doing harm and long-term consequences”

Many people hold the constraint against doing harm, which says that the reason against doing harm is stronger than the reason against merely allowing harm. So far, not much has been said about what these people should think about behaviour that leads to harm in the long-term future. In this talk, I first provide three progressive clarifications about the constraint against doing harm as a forward-looking principle. First, the constraint applies only to harm that is sufficiently proximate to the agent’s behaviour. Second, agents have limited prima facie permissions to impose risks of harm through everyday behaviour. Third, the constraint against doing harm does not tell agents to refrain from behaviour that does not increase anyone’s ex ante risk of suffering harm. I then use these clarifications to respond to Will MacAskill’s and Andreas Mogensen’s Paralysis Argument. According to the Paralysis Argument, the constraint against doing harm, when applied to actions with long-term consequences, leads to the absurd conclusion that we should do as little as possible in our lives. I argue that everyday actions do not, in fact, increase the risk that an agent brings about remote harm, and therefore, the Paralysis Argument does not arise.

Monday 7 December - Session B

15:20 - Claire Field, “Is Longtermism reckless?“

I introduce a challenge to Longtermism, the Recklessness Challenge, and explore some possible responses available to Longtermists. Longtermism promotes actions that involve substantial risk of moral squander – losing significant moral value for no gain. This is largely due to the extreme epistemic difficulty of making predictions about the long-term future. I argue that on most ways of justifying risk, this risk of squander is reckless. I argue that the best defence of Longtermism in response to this challenge is one that makes use of the observation that the risk of squander has important similarities to excessive risk aversion. Acknowledging this makes room for a response that is somewhat friendly to Longtermism – good moral decision-making requires us to take some risks, and those Longtermism endorses may not always be the wrong ones.

16:20 - Johan Gustafsson and Petra Kosonen, “Prudential Longtermism”

According to Longtermism, our acts’ overall influence on the value of the world is mainly determined by their effects in the far future. Given additive axiologies, such as total utilitarianism, there is a straightforward argument for Longtermism due to the enormous number of people that may exist in the future. This argument, however, does not work given person-affecting views. In this paper, we shall argue that these views may, in fact, also lead to Longtermism. The reason they may do so is that Prudential Longtermism may be true. Prudential Longtermism is true for you if and only if our acts’s overall influence on your expected well-being is mainly determined by their effects in the far future. We argue that-due to a combination of anti-ageing, cryonics, uploading, and biological uploading-there can be an enormous amount of prudential value for you in the far future. This potential value may be so large that it dominates your overall expectation of lifetime well-being.

Monday 7 December - Session C

15:20 - Loren Fryxell, “A lexicographic expected utility characterization for infinite ethics“

Aggregative consequentialist theories suffer from infinite paralysis—if there is any positive probability that the world contains infinite moral value, and individual actions can only cause a finite change in value, then we should be morally indifferent between all actions (the expected aggregate moral value is infinite/undefined for all actions). That being said, classic expected utility theory (von-Neumann-Morgenstern) does not apply in such environments. In particular, the continuity axiom prohibits moral preferences from considering any outcome as "infinitely better" than another. I generalize the von-Neumann-Morgenstern theory by only requiring that continuity holds in some cases. The result is a lexicographic expected utility representation that allows outcomes to be "infinitely better/worse" than others. The theory does not suffer from infinite paralysis. Indeed, for any pair of lotteries with the same probability of infinitely good (bad) outcomes, the preference is represented by a standard (and hence finite) expected utility representation.

16:20 - Leopold Aschenbrenner, “Existential Risk and Growth”

Human activity can create or mitigate risks of catastrophes, such as nuclear war, climate 

change, pandemics, or artificial intelligence run amok. These could even imperil the survival of human civilization. What is the relationship between economic growth and such existen- tial risks? In a model of directed technical change, with moderate parameters, existential risk follows a Kuznets-style inverted U-shape. This suggests we could be living in a unique "time of perils," having developed technologies advanced enough to threaten our permanent destruction, but not having grown wealthy enough yet to be willing to spend sufficiently on safety. Accelerating growth during this “time of perils” initially increases risk, but improves the chances of humanity’s survival in the long run. Conversely, even short-term stagnation could substantially curtail the future of humanity.

Monday 7 December - Session D

16:20 - Isabel Juniewicz, “The Persistence of Institutions: Evidence from the Partitions and Population Transfers of Poland”

A challenge in evaluating the effectiveness of institutions is that it is difficult to separate the effect of institutions from the people that created them. Recent institutional research attempts to work around this by evaluating institutions imposed by an external force, usually due to conquest or colonization. Poland presents an excellent opportunity for just this sort of analysis, as between 1772 and 1795, in a series of three partitions, Poland was conquered by and divided between Prussia, Russia and Austria-Hungary. In addition to the effects of the original partitions, this setting provides an opportunity to distinguish between different mechanisms of transmission of the effects of institutions due to the forced relocation of the Poles living to the East of the Curzon line.

16:40 - Stefan Schubert, “People are more cosmopolitan about the distant future”

People tend to prioritise their own country and its citizens over other countries and their citizens. In a series of online experiments, we study whether greater temporal distance reduces this tendency and makes people more cosmopolitan. In Study 1, we ask American participants a series of questions measuring their degree of cosmopolitanism, and find that it increases when the questions concern the distant future, as opposed to the near future. Study 2 shows that the effect remains when participants are asked to imagine that the US will still exist in the distant future. In Study 3 we find, using a donation paradigm, that participants also become less likely to prioritise other ingroups—their local community, and their family and its descendants—when their help would be directed to people in the distant future, as opposed to people in the near future. In Study 4, we turn to whether greater spatial and social distance make people more temporally impartial; i.e. whether the converse also holds. We do not find evidence for that: participants prioritise present ingroups (e.g. Americans today) over future ingroups (future Americans) and present outgroups (e.g. foreigners today) over future outgroups (future foreigners) to the same extent. So we find that temporal distance reduces partiality related to spatial and social distance, but spatial and social distance do not reduce temporal partiality.

Monday 7 December - Keynote Speech

17:00 - Krister Bykvist, “What, if anything, is good about sustainable development?”

In my talk I focus on the economist Geir Asheim’s definitional schema of sustainable development: 

A generation’s management of the resource base at t is sustainable iff it constitutes the first part of a feasible development, starting at t, that sustains x onwards. 

I ask whether this can be turned into a morally significant factor (constraint on permissibility, constraint on justice, or a good-maker), for a set of popular candidates for x (needs-satisfaction, wellbeing growth, or a certain wellbeing level). My conclusion is pessimistic. But, on the positive side, I argue that sustainable development can instead be seen as a politically useful proxy for other morally significant factors.

Tuesday 8 December - Session A

15:00 - Tyler John, “Intergenerational Legitimacy”

A number of recent writers have defended deontic strong longtermism, the thesis that what we ought to do is primarily determined by the effects of our actions on the very long-term future. Meanwhile, political leaders and political philosophers are wringing their hands about taking even modest action to prevent catastrophic climate change, believing that a legitimate government must respect the outputs of democratic deliberation, however odious they may be. This talk seeks to reconcile these ideas. I argue that political philosophers have long overlooked the crucial fact that obligations of political legitimacy are owed to future people as well as present people. A legitimate government is one that wields its power in a way justified to all generations who fall under it, not simply present generations. 

I reconstruct and analyze several popular theories of the fundamental grounds of political legitimacy. The theories of political legitimacy that I consider ground obligations of legitimacy in the value of autonomy or in the value of equality. As I show, the most promising theories of the grounds of political legitimacy imply that States have duties of legitimacy to future generations as well as to present citizens. The actions that States take today can infringe on the freedom of future people, and they can also constitute a kind of subordination which runs afoul of egalitarian political relationships. Given that current and future generations each have the properties on which duties of legitimacy supervene, a legitimate State must act to ensure the freedom and equality of future generations in a way that the talk will aim to make clear. While theories of intergenerational legitimacy differ in their normative verdicts, several plausible accounts imply strong duties of legitimacy that stretch into the distant future.

15:40 Jeff Russell - “Infinite values and large finite values”

I'll present a theorem about the structure of value and explain why I find it puzzling, especially for utilitarians. The theorem says, roughly, that a dominance-style principle for preferences between risky prospects requires that regimes of infinite value are well-ordered, and that finite values within each regime are bounded.

16:00 - Stefan Riedener, “Existential Risks from a (Thomist) Christian Perspective”

The most straightforward argument for x-risk reduction is utilitarian. But there are few utilitarians. By contrast, there are about 2.4 billion Christians. So how should Christians assess x-risk reduction? Based on the moral philosophy of Thomas Aquinas, I’ll outline three Christian rationales for why anthropogenic extinction would be morally catastrophic: it would (i) constitute a total failure of our attaining our natural end, (ii) amount to an enormous form of hybris, and (iii) destroy the most valuable aspect of creation. So Christians should agree with utilitarians that x-risk reduction is extremely important. Indeed, given the gravity of the effect and the negative form of the duty, they have reason to support it even at considerable cost to classical forms of caritas. I’ll end by suggesting that similar rationales emerge on other religious and non-religious worldviews.

Tuesday 8 December - Session B

15:00 - Jacob Barrett, “Institutional selection”

Normative political philosophy typically takes an “institutional design” perspective, attempting to come up with institutions that would, if implemented, achieve certain moral or political objectives. This contrasts with an “institutional selection” perspective, which instead asks how we might redirect or strengthen the selection pressures driving the evolution of institutions in order to achieve better institutions in the long run. The goals of this talk are, first, to put institutional selection on the agenda for political philosophers and longtermists, given its great importance to the production of long-term value or justice; second, to begin to develop a general framework for theorizing about institutional selection; and, third, to draw out some preliminary normative conclusions.

15:20 - Orri Stefansson, “What is the point of offsetting?”

I shall defend two related claims. The first claim is that, contrary to what Broome has argued, offsetting is not a legitimate way of meeting one's duty not to cause greenhouse gas emission. Moreover, in light of a recent argument by Barry and Cullity, I argue that this first claim holds also for offsetting by sequestering, not only for offsetting by forestalling. The second claim is that, because of (the truth of) the first claim, we have no reason to offset rather than using the money to do more good in other ways. And since it turns out that offsetting is not an effective way to do good, we should not spend money on offsetting. I nevertheless end by suggesting that it may be rational to offset one's emissions, even if one is quite convinced by my argument, as long as one is not certain that I am correct. Offsetting is very cheap, and hence it would be rational, for those who are risk averse and morally uncertain when it comes to the grounds for their duties, to offset their emissions.

16:00 - Vuko Andric, “Does democracy require reducing the legislative power of present persons on behalf of future persons?”

In my talk, I critique two arguments for the claim that democracy requires reducing the legislative powers of present persons on behalf of future persons. Such restrictions might either result from the attempt to partly confer legislative powers upon future persons – e.g., by reserving some seats in parliament for representatives of future persons – or from limiting the power of current legislatures through constitutional courts that restrict current legislatures on behalf of future persons. The first argument is based on the claim that future and present persons belong to the same demos. According to the second argument, democracy requires that current legislatures be restricted to ensure that the rights of future persons are not violated by current law-making.

Tuesday 8 December - Session C

15:00 - Peter Bayer, “Farsighted manipulation and exploitation in networks”

Economic agents with an increased sophistication sometimes use their advantage to exploit their more naive counterparts. In public goods games played on networks, such an agent will attempt to manipulate as many of his neighbors as possible to produce the public good. We study the exploitation of a myopic population by a single farsighted player in such games. We show the existence and payoff-uniqueness of optimal farsighted strategies in every network structure. In the long run, the farsighted player's effects are only felt locally. A simple dependence-withdrawal strategy reaches the optimal outcome for every network if the starting state is unfavorable, and reaches it for every starting state if the farsighted player is linked to all opponents. We characterize the lower and upper bounds of long-run payoffs the farsighted player can attain in a given network and make comparative statics with respect to adding a new link. The farsighted player always benefits from linking to more opponents (sociability) and is always harmed by his neighbors linking to each other (jealousy).

15:40 - Timos Athanasiou, “Strategy-proof procurement of intergenerational public goods”

The talk takes up the problem of procurement of public goods whose benefits extend over several generations. We focus on issues pertaining to incentives in the context of mechanism design. 

On this topic there are interesting, unresolved issues. This is so for two principal reasons. First, the horizon is extended or infinite. Second, at the time of decision the preferences of future affected generations are not realized. Hence, the mechanism needs to operate under multiple veils of ignorance. The incentive problem takes a form particular to IPGs. 

We propose a novel approach that tackles the incentive problem and discuss the class of mechanisms that it may generate.

16:00 - Sareh Vosooghi, “Coalition formation under uncertainty and the power of Information Design”

I examine coalition formation in a public-bad game under uncertainty and characterise the unique stable coalition structure at each belief about the social cost of the public bad. I show that higher expectations about the social cost of public bad lead to improved efficiency. This result resolves the small-coalition paradox in the literature on stochastic International Environmental Agreements. Furthermore, invoking the communication of information by central authorities, I propose a new approach for implementing stable coalitions.

Tuesday 8 December - Session D

15:00 - Eva Vivalt, “Forecasting models”

One of the main threats to longtermism is that it is not clear to what extent one can positively affect the farther future. While standard empirical approaches provide little guidance regarding events on this timescale, forecasts may be a helpful tool. However, it is still an open question as to whether forecasts that resolve in the short run can be informative about the accuracy of forecasts on a much longer time horizon. I hypothesize that some types of items are harder to forecast accurately than others and, in particular, that we may have more confidence in long-term forecasts if we can show that individuals can accurately forecast complex models. I use an example of researchers forecasting models of COVID-19 deaths to highlight the promise and challenges of such an approach.

15:30 - Benjamin Tereick "Intuitive forecasting in the short and long run: An experimental study"

We aim to investigate whether human performance at forecasting the short-run is associated with performance at forecasting the long-run. To do so, we plan to set up an experiment in which participants are asked to predict values from several time-series both into the near  and into the far future. We will analyze whether performance in short-term tasks predicts performance in the long-term tasks and study whether strategies employed by participants favor performance in one of the tasks over the other. 

(joint project with Maxime Cugnon de Sévricourt)

16:00 - David Reinstein, “Barriers to effective giving and concern for global priorities: evidence synthesis and a field experiment on donors' response to effectiveness information”

While hundreds of billions of dollars are donated to charity each year, the effectiveness of these charities differs by orders of magnitude, even within similar categories. Furthermore, many individuals do not donate substantially even though they believe that the cost of saving a life is small. Our collaborative and dynamic synthesis (https://bit.ly/eg_barriers) considers this ‘puzzle’. We outline and categorize potential barriers to effective giving and assess the evidence for each barrier. This directly informs “how to motivate effective giving”. It also offers insights into the drivers of ‘concern for’ and ‘willingness to act’ to address the most neglected, tractable, and consequential global and humanitarian priorities, including long-term well-being and existential threats. 

 Our current focus: How do potential donors react to quantitative measures of charitable impact? The EA movement and organizations like GiveWell and Impact Matters/Charity Navigator are increasingly presenting these measures to larger, more mainstream audiences. We survey the (limited and often tangential) evidence on the effect of such measures. We also present evidence from our recent field experiment (https://bit.ly/dv_field_exp) with a US-based international relief charity. In each of two annual Thanksgiving email solicitations (2018: N= 182,594; 2019: N = 79,754) a randomly-selected half of all recipients were presented the ‘treatment’: measures of impact per-dollar. Our results suggest the ‘unrealistic’ impact presentations (presented in 2018) increased the propensity-to-give in direct response to the email (95% Bayesian credible intervals: approximately +50% to +150% of the mean incidence rate of 5.4 per 1000). More realistic impact information (presented in 2019) seems to have had a negative or near-zero effect on donation incidence (credible intervals: -80% to +40%). 

Note: the credible intervals presented above are preliminary. 

Co-Authors on ‘Barriers’ synthesis: Dr. David Reinstein, Nick Fitz, Ari Kagan, Janek Kretschmer, Luke Arundel and others. Co-Authors on Field experiment: Scott Dickerson; experiment organized jointly with Kiki Koutmeridou and members of the Donors’ Voice advisory team

Tuesday 8 December - Keynote Speech

17:00 - Chad Jones, “The End of Economic Growth? Unintended Consequences of a Declining Population”

In many models, economic growth is driven by people discovering new ideas. These models typically assume either a constant or growing population. However, in high income countries today, fertility is already below its replacement rate: women are having fewer than two children on average. It is a distinct possibility --- highlighted in the recent book, Empty Planet --- that global population will decline rather than stabilize in the long run. In standard models, this turns out to have profound implications: rather than continued exponential growth, living standards stagnate for a population that vanishes.