The first epistemological challenge stems from prognostic uncertainty. When one makes a decision today, one can make educated guesses or even reliable predictions about the outcome, but the farther into the future one tries to look, the blurrier one’s vision becomes. Making predictions about how today’s decisions will influence the distant future longtermism has in mind, is close to impossible, because of the sheer number of unknown and unknowable causal relations between the decision itself, it effects, the effects of these effects, and so on.

Prognostic uncertainty alone is a problem that policymakers in the field of short- and mid-term governance need to handle. Even if they cannot know the effects of their decisions, they are held accountable for them in virtue of their mandate. However, when it comes to long-term governance, prognostic uncertainty transforms into a challenge that cannot simply be overcome by assigning or willingly accepting accountability. For, when the long-term outcomes of decisions or implemented policies come into effect, the original decisionmakers will no longer be there to be blamed or praised—or be ‘punished’ with non-election because of their short-sighted decisions.

But since accountability and governance are closely intertwined with each other (Kaler 2008), one needs to address this challenge. How can long-term governance work in the face of prognostic uncertainty when holding policymakers accountable for the results of their decisions is impossible?

One suggestion that shall be favored here is that accountability should not be connected to results of decisions, but to adherence to normative principles (Lechterman 2023). According to this suggestion, holding someone accountable does only require to know whether the decisions can be justified by an accepted normative principle. To illustrate this, let us suppose that the following were an accepted normative principle: ‘Burdens to future generations must be balanced with burdens to present generations according to the following balancing metric: […]’ Because of prognostic uncertainty, one cannot know how many people will exist in the future and under which living conditions. However, if one did not focus on outcomes but on normative principles, this knowledge would not be necessary. Policymakers making decisions possibly burdening future generations could nonetheless be held accountable for their decisions if one measured them by the degree of adherence to the previously formulated normative principle.

One could also extend the idea from adherence to normative principles to adherence to methods of making predictions. Accountability would then be based on how much policymakers use the most reliable prognostic methods in order to assess the long-term impact of their decisions. While these methods may not provide certain knowledge about the distant future, this proposal would reward evidence-based policymaking and, at the same time, discourage unfounded speculations or immoral policies based not on evidence but on short-term interests.

Needless to say, both approaches require much further discussion. For example, it would need to be determined how one should arrive at a consensus about the exact normative principles that would work as a foundation of accountability in long-term governance. There would also have to be a mechanism for assessing prognostic methods for their value in long-term policymaking. However, since basing accountability on the outcome of decisions is no appropriate solution for long-term governance, because of the problem of prognostic uncertainty, alternative concepts of accountability deserve further exploration.

The second epistemological challenge to be discussed is moral uncertainty. In contrast to prognostic uncertainty, which is uncertainty about facts, moral uncertainty is uncertainty about what one ought to do. It occurs when one has to decide between a set of alternative options with different results and a set of ethical theories, each of which recommend different options (MacAskill et al. 2020, p. 3). The crux is that in situations of moral uncertainty, all descriptive facts are well-known, but there is still uncertainty about what to do. In other words, moral uncertainty results from a lack of ethical guidance, which is particularly challenging when one has to make a decision no matter what.

To give a practical example from the context of long-term governance: When a regulatory body has to make a decision about an environmental policy, this body faces moral uncertainty if there is no decisive guidance on how to balance the interests of present and future people (and, to complicate matters further, non-human animals or even artificial life forms that may exist in the distant future). This body may have the advice of the best environmental experts with the most sophisticated models for the outcome of the available decisions—but there can still be a form of moral paralysis if there is a lack of ethical guidance, such as disagreement amongst ethical experts about justifiable discount rates when choosing mitigation pathways.

Given that our ability to make reliable predictions about the future decreases with a growing temporal distance from said future (see also the next section of this article), moral uncertainty adds further complexity to situations which require reasonable decisions. For, if we have neither sufficient factual knowledge nor sufficient certainty about which of our alternative decisions we ought to choose, one can ask how policymakers should make any justified decisions with long-term effects at all.

One solution in the field of public decision-making could be to adopt a principle that a policy should only be put into force if it can be publicly justified to reasonable people (Barrett and Schmidt 2024). This principle avoids making any decision about which policy is ethically justifiable because of its content. Instead, it focuses on public procedures, on how a policy is established in a public discourse. The proponents of this principle recognize the need for exceptions by stating that “the prohibition against enforcing publicly unjustified laws may need to be overridden in high-stakes cases where there is significant risk that failing to enforce a law is morally disastrous, even though some reasonable people disagree—for example, because the law protects a class of beings that some reasonable people deny have status.” (Barrett and Schmidt 2024, p. 17)

While this exception appears intuitively plausible—there do indeed seem to be cases that should not even be open to public debate because of what is ethically at stake

One example could be subjecting a vast number of people to slavery and leading miserable lives because of some seemingly ‘objective’ criteria that could even be used in logically consistent arguments.

For the time being, the problem of moral uncertainty appears to remain unresolved. Even the authors who drew attention to the issue of moral uncertainty acknowledge that many topics remain open for research, including the “implications of moral uncertainty for practical ethics” (MacAskill et al. 2020, p. 212). Since practical ethics can be regarded as a component or pre-stage of long-term governance, there also remains the open question of how practical ethics should inform policymakers in the face of moral uncertainty. However, as the public justification approach shows, there are ways to handle moral uncertainty, at least to some extent.

The third and final challenge is the concept of ‘value lock-in’, introduced by MacAskill (2023) in his already mentioned seminal book on longtermism. This concept is best expressed in his own words: value lock-in is “an event that causes a single value system, or set of value systems to persist for an extremely long time.[…] If value lock-in occurred globally, then how well or poorly the future goes would be determined in significant part by the nature of those locked-in values” (MacAskill 2023, p. 78). An illustration of the theoretical concept of value lock-in—albeit not on a global scale—is Christianity and the values deeply rooted in its central tenets, such as an ideal of poverty, subservience to a specific deity, or a universal right to life in all of its different stages.

Note, however, that a value lock-in does not require that each and every person lives in accordance with those values. There can be reasons such as weakness of the will or hypocrisy leading to non-adherence to values. However, in scenarios of a value lock-in, it is still much easier to enforce adherence to these values through the authorities.

There appears to be a form of global value plurality—but this might only be a temporary state of affairs. MacAskill suggests metaphorically, “that we are living through a period of plasticity, that the moral views that shape society are like molten glass that can be blown into many different shapes” (MacAskill 2023, p. 102). In other words, today’s societies could develop into multiple directions from an ethical point of view, depending on which values or value systems gain dominance over others. And given the rapid development of information and communication technologies, the spreading and cementation of value systems could occur much faster than in past centuries where information was spread through the spoken word, manually or—later—industrially produced books.

The relevance of the concept of value lock-in for long-term governance should now have become fairly salient. If one regards long-term governance not just as policymaking for the distant future but as the translation of values into rules and their implementation in legal norms, it makes a difference if these values are—to borrow from MacAskill’s metaphor—plastic and allow to be shaped or if they are solid and will remain solid over an extended period of time. Value-based decisions and policymaking in the context of long-term governance would have a much greater impact and would come with greater responsibility if one knew that decisions and policies were based on values that will remain solid because they will affect the lives of generations in the distant future, as opposed to people living in the present and maybe our close descendants.

Value lock-in turns into a challenge for long-term governance for a conceptual and an epistemological reason. From a conceptual perspective, values are typically vague to a certain degree, at least on some higher level of abstraction (Dougherty 2014). The already mentioned value of a universal right to life, for example, leaves ample space for interpretation. Ought it to be respected in each and every context, even in the context of lethal self-defense? Should it be interpreted as a prohibition to end a person’s life even if this person requests it? Are humans the only beings who are subjects of this right or should it be extended to other, maybe even artificial beings? These are conceptual questions directly related to the level at which one describes and interprets a value—but responses to them are of utter importance for policymakers in long-term governance.

The epistemological side of the challenge is that the quality of our predictions of how these values will manifest in future outcomes appears to depend on the level and specificity of interpretation. Revisiting the value of a universal right to life, one could imagine both a blissful and a bleak future. If one regards the scope of this right as maximal, encompassing all lifeforms, including humans, non-human animals and even artificial lifeforms, implementing policies based on it could indeed lead to a future worth living in for these lifeforms. However, narrowing the scope of this right to beings who are deemed to have ‘moral value’ on the basis of some arbitrary criterion could lead to a bleak dystopia in which those who have no moral value are ruthlessly exploited or killed, even if they have a desire to keep on living and if their well-being depends on some sense of their life being safe.

One possible solution to both the conceptual and the epistemological aspects of the challenge of value lock-in could be not to focus on values at the highest level (‘universal right to life’) but on principles or rules on a more actionable level (‘respect the right to life of every sentient being, regardless of their spatio-temporal location and their physical constitution, and do not intentionally threaten or end their life, if not for the purpose of self-defense’). A rule like this would at least exclude interpretations of the universal right to life that exploit the future in favor of the present or focus exclusively on humans as opposed to non-human animals or artificial beings. However, there would still be much work to be done when attempting to lock in a value interpreted in this way and implementing it in the form of long-term policies.

An alternative approach that shall be sketched here could be not to focus on the lock-in of values but of value-driven goals. A value-driven goal is a goal that is motivated by some underlying value, even if this value is not expressed explicitly. Since there are frameworks for formulating goals, such as the ‘SMART’ framework (Specific, Measureable, Assignable, Realistic, Time-related; Doran 1981), this approach could turn out to be more practical in long-term governance. It would avoid confusions about the exact interpretation of the values, but would still contribute to them being indirectly locked-in.

The value of a universal right to life could, for example, be expressed in value-driven goals for the reduction of carbon emissions or the non-proliferation of fully autonomous weapon systems. It could also be expressed in social policies improving the living conditions of non-human animals or preventing the ruthless exploitation of artificial, sentient beings if they come into existence in some distant future. The quality of such policies and goals would then not so much depend on theoretical interpretations of the underlying values but on their adherence to already established quality criteria for the formulation of goals. To give an example, it is clearly possible to formulate the goal to reduce carbon remissions by a specific, measurable degree over a certain period of time, and thereby refrain from threatening the lives of future people by creating hostile environmental conditions. Achieving this goal would then constitute the realization of the universal right to life of future people as a value driving this goal.

The challenge remains, though, to properly assess the long-term effects of realizing such value-driven goals. Hence, although the challenge can be shifted from value lock-in to value-driven goals, this proposal is far from being a definitive solution. Nonetheless, it shows that the longtermist idea of a value lock-in is not just an interesting thought experiment, but has real importance for long-term governance. For, as history shows, values can get locked in, compete with new values, and—possibly—get replaced by them. It also shows that while it can be difficult to overcome the challenge itself, there is the possibility to bypass it while still sticking to the core idea itself.