Epistemic status: 

This series, written for the "Essays on Longtermism" competition, was my first experiment in using Claude 4.5 Sonnet as a ghostwriter, and I learned some valuable lessons worth sharing. 

While I feel relatively good about this final product, I had to do extensive editing to correct errors on Claude’s initial version, and due to serious time constraints (the original series was written in 3 days) the version I submitted for the competition contained some obvious errors I missed and, more importantly, some very subtle errors which I am quite embarrassed of. 

While the large majority of the ideas in the series are my own, Claude did a good job and sometimes an excellent job of articulating them and a decent job of organizing them in a logical sequence. My main complaint is that the essays are occasionally repetitive and there are some elements which are duplicated across essays, however partly this was intentional in order to make it so that each essay could easily be read as a fully stand-alone piece. Claude can also be wildly mis-calibrated, often claiming something is “the best” or “the most important,” when there is no strong evidence for this or consideration of alternatives.

Additionally, while I have tried to remove any subtle errors, I have occasionally realized I missed one that was actually quite important, and so I apologize if any remain which I have missed. This is a big challenge, as Claude can sound very eloquent at times and yet be completely wrong, misconstruing ideas in logical sounding yet subtly misleading ways. 

My biggest recommendation to anyone using AI for ghostwriting is to very, very carefully read the writing and think about it and make sure the ideas are correct, as well as to perhaps explicitly mention AI has been used in the writing and there may be subtle errors of this type.

With those caveats in mind, here’s a link to the initial version of the essay submitted to the “Essays on Longtermism” competition.

I also want to note that while this series represents some of my best ideas over the last few years, it does not have the level of finesse and flavor of posts I typically like to write, perhaps with the exception of “Viatopia and Buy-In” which was an excerpt from my in-progress essay on “Deep Reflection.” 

Deep Reflection is the comprehensive examination of crucial considerations to determine a strategy to achieve best achievable future; this series was largely based on my research on Deep Reflection. (Deep Reflection Summary Available Here)

As mentioned, this essay, easily readable as a stand-alone piece, is the first essay and part of my submission to the "Essays on Longtermism" competition.

Brief and Comprehensive Series Explainers Here

TL;DR

This essay comments on two pieces from the Essays on Longtermism collection: Owen Cotton-Barratt and Rose Hadshar's "What Would a Longtermist Society Look Like?" and Hilary Greaves and Christian Tarsney's "Minimal and Expansive Longtermism." 

I argue that while the longtermist community has developed strong theoretical foundations, we face a critical infrastructure gap. We need concrete mechanisms that make longtermism practically achievable without requiring coercion or universal adoption of explicitly longtermist daily behaviors. 

The essay establishes several key frameworks that subsequent essays build on: the instrumental commoditization thesis (AI will soon make implementation trivial while direction becomes everything, making values work a high-leverage pre-AGI Better Futures intervention), the cooperative paradigm (community infrastructure that makes all longtermists more effective through novel systematic collaboration mechanisms and sharing of best practices), and the importance of systematic value reflection infrastructure. 

Cotton-Barratt and Hadshar describe what longtermist societies might look like but provide limited guidance on mechanisms to create them; 

Greaves and Tarsney note that expansive longtermism is less robust than minimal approaches. 

I show how concrete institutional designs and community infrastructure can make expansive longtermism more tractable, bridging the gap between their theoretical analyses and practical implementation. This essay provides the conceptual foundation that unifies the concrete mechanisms explored in subsequent essays.

While I believe extinction is rightfully a key focus within the longtermist community, this essay series focuses especially on MacAskill's "Better Futures," within longtermism, which concerns the quality of the long-term future, in addition to whether or not we have a future at all. 

"Viatopia" (explored extensively later in this series,) is an especially critical concept within Better Futures, and connects directly to Cotton-Barratt and Hadshar's concerns as to what a longtermist society might look like, as MacAskill states: "a state of the world where society can guide itself towards near-best outcomes, whatever they may be. We can describe viatopia even if we have little conception of what the desired end state is. Plausibly, viatopia is a state of society where existential risk is very low, where many different moral points of view can flourish, where many possible futures are still open to us, and where major decisions are made via thoughtful, reflective processes."

A Brief Longtermist Autobiography

When I was a kid, I discovered longtermism on my own, and have been one ever since. For most of that time, it has been a lonely journey.

In 2021, I wrote a book draft on longtermism and broad societal strategies to prevent existential risk and achieve the best possible future. I was planning to call it either "Ways to Save the World" or "Paths to Utopia." 

The primary motivations behind writing this book were to:

• Launch a longtermist movement
• Design high level societal mechanisms which simultaneously:
  • Dramatically improve the near-term and long-term future trajectory
  • Create moral progress/progress on values
  • Prevent existential risk.

Then, in January 2022, while preparing for a Master's degree on Social Entrepreneurship, I discovered there already was a longtermist movement! 

I became obsessed with the Effective Altruism group at my university and dropped out of school to pursue direct longtermist community building and research while living at a longtermist group house in Berkeley, California.

I was quite thrilled when Will MacAskill wrote the book, "What We Owe the Future," proposing trajectory change as on par with extinction risk due to the threat of value lock-in (1) (2). Trajectory change was half of my book, and I had been sad to see most longtermists had gone sour on “Broad Longtermist” strategies (1) (2), which my book preferred, with most longtermists favoring narrow strategies such as technical AI safety.

In 2024, I requested a debate week on this topic, and in 2025, I got my wish. Unfortunately, I got carried away writing an essay for the competition and spent about half a year writing a 35,000-word (still unfinished) essay—although an intermediate summary version is available here.

When I asked for feedback, and Will MacAskill shared his own essay series with me, I realized I needed to update some of my own ideas on "Seed Reflection," which was my new name for my (unpublished) book "Paths to Utopia," which is quite similar to what Will MacAskill calls "Viatopia."

While I am in the middle of a much deeper research process that is nearing its final phases, this essay series represents some of the ideas that I think are most important from my evolving work, if presented in a somewhat rough-and-tumble way to get them out in time for the "Essays on Longtermism" competition.

My deepest gratitude to the EA Forum, Effective Ventures, Toby Tremlett, Will MacAskill, David Thorstad, Hilary Greaves, Jacob Barrett, and Eva Vivalt; as well as all of the wonderful authors who contributed essays, for their inspiring work on longtermism, and for making this competition possible.

I will begin by commenting on two of the "Essays on Longtermism." The first, by Owen Cotton-Barratt and Rose Hadshar, is on what a longtermist society might look like. The second essay, by Hilary Greaves and Christian Tarsney, explores and compares minimal and expansive versions of longtermism.

Section 1: Building the Bridge: From Longtermist Theory to Institutional Reality

In "What Would a Longtermist Society Look Like?", Owen Cotton-Barratt and Rose Hadshar (2025) explore what societies with various levels of commitment to longtermism might look like. The authors claim the definition of longtermism they use is roughly that of "Strong Longtermism" from Greaves and MacAskill (2021); the authors state: "A longtermist perspective is a perspective which assesses actions almost entirely on the basis of their expected impacts on the far future."

Their analysis reveals an important tension. They note that it seems implausible that both a state and all of its citizens would be strictly longtermist, adhering strictly and exclusively to longtermist ethics. More realistic scenarios include a strictly longtermist state with mixed citizen commitment, or a partially longtermist society where both state and citizens only partially prioritize longtermism. Yet crucially, Cotton-Barratt and Hadshar identify what longtermist societies would need without fully addressing how to create these conditions.

They provide an illuminating list of instrumental goods even a strictly longtermist society must pursue: 

• Having and raising children
• Education
• Mechanisms to match people with suitable work
• Technology and built environments for productive work
• Psychological health through communities, entertainment, and therapy
• Nutritious food
• Housing and domestic goods
• Healthcare
• Stable governance

They note two reasons for maintaining citizen wellbeing: political stability (maintaining power when not all citizens are longtermist) and productivity (citizens need psychological health and energy to be effective in longtermist work).

I strongly agree with their framework. For interventions to be realistic, they must serve public wellbeing; for them to be effective, they must create public goods. However, I believe we can go significantly further in bridging the gap between their theoretical analysis and practical implementation.

The Missing Link: From Description to Mechanism

Cotton-Barratt and Hadshar describe what longtermist societies might look like, but provide limited guidance on the mechanisms to create and sustain them. This is the central challenge: how do we move from our current myopic society to one that naturally embodies longtermist values, without coercion or requiring everyone to be explicitly longtermist in their daily decisions?

This challenge becomes particularly urgent when we consider what is sometimes called "skating to where the puck is going." Very soon, advanced AI will enable us to implement nearly any societal design we can imagine. At that point, the limiting factor won't be capability but direction (knowing what we want to create). As AI commoditizes everything except ideas and values, determining the best possible future becomes perhaps the highest-leverage work possible before transformative AI arrives. Whatever institutional designs and value frameworks are well-developed and readily available when this capability arrives may be what we actually implement, simply because we may use the tools that are lying around rather than conducting an exhaustive search.

This creates two critical imperatives. First, we need concrete institutional designs ready in advance, so that when the capability to implement them arrives, we have vetted, robust options available. Second, we need diversity in our institutional designs. If only one or two viatopia proposals exist, decision-makers might simply choose one and proceed. But if dozens of well-developed proposals exist, each compelling in different ways, this forces the serious reflection and debate necessary to avoid premature lock-in to suboptimal futures. The existence of multiple attractive paths makes it obvious that we should explore the space further before committing.

Viatopia as Practical Implementation

The concept of "viatopia" (an intermediate state that helps humanity converge on the best achievable future) provides the missing link between Cotton-Barratt and Hadshar's theoretical analysis and concrete action. While MacAskill introduced this term and MacAskill and Ord's "Long Reflection" represents one viatopian mechanism, there remains vast unexplored design space for institutions and mechanisms that could serve this function.

My work focuses on developing concrete viatopian mechanisms that address the specific challenges Cotton-Barratt and Hadshar identify. The Hybrid Market, for instance, directly implements their observation that "there is no reason that the state need provide these productivity-enhancing things directly, rather than leaving individuals to obtain them via markets. If the state wishes to tip the scales of consumption choices to account for the externalities they cause, they can do that via taxes and subsidies."

However, rather than relying on state implementation, the Hybrid Market represents a decentralized mechanism whose core premise is taxes and subsidies (automatically pricing in externalities, both positive and negative, across all timeframes, including proxies forecasting long-term effects on the far future). This allows society to efficiently move toward better futures by making longtermist considerations economically rational at the individual level, without requiring centralized coordination or everyone to explicitly adopt longtermist ethics.

The Children's Movement and Systematic Value Evolution

Cotton-Barratt and Hadshar emphasize "educating people so they are well equipped to tackle challenging research work" and "the education of children to become productive workers." This highlights an important and under-explored longtermist institutional intervention: systematically improving how we raise children to create a generation that naturally embodies the collaborative, long-term-oriented, epistemically rigorous mindset longtermism requires.

The Children's Movement I propose goes beyond education to comprehensive reimagining of how we support child development. By developing comprehensive systemic infrastructure to help parents and society at large show children deep care and attunement, empower them with autonomy, and teach them to take responsibility for their future, we create adults for whom longtermist interventions seem obvious and natural. This addresses both the political challenge (creating citizens who genuinely support longtermist priorities) and the productivity challenge (developing the psychological health and capabilities needed for effective work) that Cotton-Barratt and Hadshar identify.

More fundamentally, childhood represents perhaps our most crucial leverage point for systematic value improvement. We need institutions and interventions that help humans reflect on values, experiment with different value frameworks, engage in substantive debate, and systematically move toward better values over time. This could include not just childhood interventions, but psychological and social interventions throughout life, new institutions for value deliberation, and AI coaches that help individuals explore their values. (These will be explored in the future essay in this series "Shortlist of Viatopia Interventions.") The Children's Movement exemplifies this principle by targeting humans at the earliest stage possible, creating compounding effects across entire lifetimes and generations.

This systematic attention to value evolution is essential if we want to "keep the future human" (maintaining human agency and allowing human values to evolve carefully rather than immediately optimizing everything with AI). Some viatopia designs envision humans remaining the primary determiners of values into the far future, at least until we're confident we want AI to have significant influence over value formation. This requires institutional and technological infrastructure for deliberate, thoughtful value evolution rather than rushing to lock in our current values.

Existential Compromise: Resolving the Strict/Partial Tension

Cotton-Barratt and Hadshar note a concerning possibility: dystopian scenarios of a coercive state with a strict commitment to Longtermism, ruling over a people that does not share its views. This highlights a fundamental challenge: how can strict and partial longtermism coexist without coercion?

A concept I came up with which I’ve been calling "existential compromise" provides a potential resolution. This idea was inspired by Nick Bostrom and Carl Shulman's analysis in "Propositions Concerning Digital Minds and Society," where they showed how vast resources enable win-win compromises. (In the example, “super-beneficiaries” are hypothetical future beings who may have unusually strong moral claims due to highly efficient use of resources, a reframing of "utility monsters," which is the pejorative form of this term.) They illustrated this with a specific example: "Consider three possible policies: (A) 100% of resources to humans (B) 100% of resources to super-beneficiaries (C) 99.99% of resources to super-beneficiaries; 0.01% to humans. From a total utilitarian perspective, (C) is approximately 99.99% as good as the most preferred option (B), and from an ordinary human perspective, (C) may also be 90+% as desirable as the most preferred option (A)." William MacAskill has developed similar ideas in his "Grand Bargain" framework in the Better Futures series, exploring how different value systems can reach mutually beneficial agreements given sufficient resources.

The vast scale of the far future (potentially 10^52 or more human-life-equivalents) makes possible agreements that would be impossible with smaller stakes. With sufficient abundance (which advanced AI could provide), we can satisfy both those who want to pursue immediate personal preferences and those who want to optimize for the far future.

For instance, we might guarantee that a substantial fraction of cosmic resources remain available for those who prefer to live in relatively unoptimized, human-scale societies experiencing the full range of human and near-human experiences. Simultaneously, we pursue careful reflection and eventual optimization with the remaining resources (potentially the vast majority). This makes it possible for both strictly longtermist and partially longtermist citizens to coexist within the same civilization, with neither feeling oppressed by the other's values.

Critically, viatopia can be designed to appeal broadly while still moving systematically toward better futures. By ensuring current generations experience genuine wellbeing improvement (not sacrifice for the future), we make longtermist institutions politically viable. This addresses Cotton-Barratt and Hadshar's observation that maintaining citizen happiness is essential both for political stability and psychological productivity.

Building on Existing Foundations

The effective altruism and longtermist community has developed substantial theoretical frameworks and important infrastructure for effective implementation. Existing meta work and community building (through organizations like CEA, 80,000 Hours, and numerous local and student groups) provides crucial foundations. Fellowship programs, career advising, and community platforms like the EA Forum demonstrate the power of well-designed infrastructure to coordinate and amplify individual efforts.

However, there remain significant opportunities to expand this infrastructure, particularly in ways that are more decentralized, crowdsourced, and scalable. These interventions are explored extensively in a post later in this series “Shortlist of Viatopia Interventions”, and will be summarized later on in this piece. 

This represents an important direction for the community: continuing to develop institutional and technological infrastructure alongside theory development. We need entrepreneurs, institution-builders, and engineers working alongside philosophers and researchers, creating systems that make it easier for more people to contribute effectively to longtermist goals.

Conclusion

Cotton-Barratt and Hadshar provide valuable analysis of what longtermist societies might look like. Their emphasis on instrumental goods, citizen wellbeing, and the distinction between strict and partial longtermism offers a rigorous framework for thinking about institutional design.

However, moving from description to reality requires concrete mechanisms (ways of structuring markets, raising children, organizing communities, and building institutions that naturally embody longtermist values while maintaining broad appeal and respecting human agency). Viatopia provides the conceptual framework; specific implementations like the Hybrid Market and Children's Movement provide the practical mechanisms; and existential compromise provides the political solution that makes it all feasible.

Section 2: Making Expansive Longtermism Tractable Through Infrastructure

In "Minimal and Expansive Longtermism," Hilary Greaves and Christian Tarsney (2025) explore a crucial distinction in longtermist thought. While standard arguments establish "minimal longtermism" (focused on targeted interventions against specific technological existential risks), many longtermists find themselves drawn to "expansive longtermism," which holds that nearly all personal and societal decisions should be made considering their impact on the far future.

Greaves and Tarsney present this tension clearly: minimal longtermism enjoys strong evidential support through clear causal mechanisms (certain technologies increase extinction risk; mitigating that risk increases expected future value), while expansive longtermism's arguments are "significantly less robust and significantly more speculative." For minimal longtermist interventions targeting technological x-risks, they note philanthropists could provide funding, talented individuals could devote careers, and policymakers could allocate resources and implement regulations. They state this could plausibly require less than 2% of GDP.

Expansive longtermism, by contrast, could justify spending over 50% of GDP on broad interventions like indirect existential risk mitigation, patient philanthropy, space settlement, accelerating growth, and improving values and institutions. On this view, even breakfast choices become longtermist decisions through opportunity cost or productivity effects. Yet Greaves and Tarsney express legitimate skepticism about whether we can reliably identify which expansive interventions genuinely improve the far future.

I find myself in substantial agreement with both their analytical framework and their cautious optimism. Technological x-risk prevention should indeed be prioritized. However, I believe the case for certain forms of expansive longtermism is stronger than their analysis suggests—not due to further philosophical considerations, but because we can build infrastructure that makes it tractable.

The Tractability Problem and Its Solution

Greaves and Tarsney's primary concern with expansive longtermism is not that it's wrong in principle, but that it's difficult to identify which broad interventions reliably improve the far future. The causal chains are long, complex, and uncertain. How do we know if improving education or institutional decision-making actually helps millennia hence?

This is fundamentally an information problem: we lack the capacity to systematically analyze hundreds of potential broad interventions, strategy considerations, and crucial considerations to determine which are robustly good. But this is precisely the kind of problem advanced AI can help solve, and indeed, the kind of problem where early preparation is essential.

The fact that there are so many different crucial considerations, and it is very difficult to know how they all affect each other and how they affect the long-term future, significantly raises the value of automating high-level strategic work early. William MacAskill has expressed enthusiasm for "automated macrostrategy," recognizing that developing methods for throwing massive amounts of compute at these problems could help us map complex relationships and forecast likely interactions between different strategic considerations.

My 2024 work on research automation using AI workflows demonstrates this potential. Rather than attempting to fully automate research (which current AI cannot reliably do), the focus is on creating an AI research automation tool library that longtermist researchers can use to dramatically accelerate their work. These tools enable systematic analysis of intervention proposals, generation and evaluation of strategic considerations, and exploration of a breadth of crucial considerations and their interactions which humans might not have the cognitive bandwidth to consider.

For example, we could run numerous automated Monte Carlo simulations to predict a large number of possible plausible interactions between different interventions and circumstances. AI can generate vast numbers of ideas, vastly exceeding what humans produce in a given timespan, potentially uncovering crucial considerations we simply wouldn't think of because there are many topics, events, and ideas we're not aware of that AI could use to inform its analysis. As AI capabilities approach and exceed human level in strategic domains (and in some narrow domains, they already far exceed human capabilities), we want these research automation tools already developed and refined, enabling researchers to navigate the intelligence explosion wisely.

Intervention development and incubation using AI tools seem nearly equally important as overall strategic automation. It's not just important to have an accurate overall picture of what good futures look like and how to achieve them; we also need to develop specific tactics for enacting that vision. Strategy without concrete implementation pathways remains abstract. Systematic intervention development transforms strategic insights into actionable projects that can actually move us toward better futures. A highly systematized intervention generation and evaluation process such as Charity Entrepreneurship’s charity development process could already be amplified, and perhaps partially automated, and this is probably increasingly true over time.

Moreover, interventions that leverage AI to enhance human researchers are possible now and will become increasingly effective as AI improves. By figuring out how to get the greatest amount of AI leverage per human researcher now, that leverage grows in lockstep as AI becomes more powerful. Developing these tools early creates a compounding advantage: as base models improve, the same infrastructure becomes dramatically more capable without additional human effort. These tools may allow significant expansion of confidence in expansive longtermist interventions by enabling unusually comprehensive macrostrategy research that can deeply evaluate a very large number of potential interventions for effectiveness and robustness against failure and downside.

The Instrumental Commoditization Thesis and Strategic Focus

One important consideration is what I call the instrumental commoditization thesis: as AI capabilities grow, implementation becomes increasingly trivial while direction becomes overwhelmingly important. Soon we'll be able to implement nearly any institutional design, technological development, or societal intervention we can specify clearly. The binding constraint shifts from "Can we do this?" to "Should we do this? What exactly should we do?"

This suggests that some of the highest-leverage work in the pre-AGI period may not be direct implementation but rather determining the best possible goals, developing multiple viable paths to good futures, and creating the deliberative infrastructure to choose wisely among them. In other words: expansive longtermism focused specifically on values, institutions, and strategic clarity.

This means recognizing that once AI makes everything else easy, knowing what we want becomes the last remaining challenge. Focusing attention on this challenge now (before path-dependencies are accelerated by transformative AI) represents an unusually high leverage way of "skating to where the puck is going" so we can know how to utilize AI when it reaches transformative capability levels.

Community Infrastructure Interventions and Cooperative Dynamics

An underappreciated aspect of the minimal versus expansive debate is that we're not operating as isolated individuals but as a community. This opens up a crucial category of expansive interventions: community infrastructure interventions that make the entire longtermist community more effective.

Greaves and Tarsney note that expansive longtermism raises coordination challenges. But we can flip this: infrastructure that reduces coordination costs represents high-leverage expansive interventions. The EA community has built valuable foundations through organizations like CEA, 80,000 Hours, and numerous fellowship programs and through platform technologies for coordination, such as the EA Forum, LessWrong, and the AI Alignment Forum. These demonstrate how good infrastructure compounds effectiveness across the entire community.

Building on these foundations, we can develop additional community infrastructure interventions, again, explored extensively in a post later in this series, “Shortlist of Viatopia Interventions.” These include:

Building on these foundations, we can develop additional community infrastructure interventions, explored extensively in a post later in this series "Shortlist of Viatopia Interventions." These include:

• Fellowship programs and incubators (like a Charity Entrepreneurship-style Better Futures fellowship) that systematically generate, evaluate, and launch 200-300 concrete interventions while upskilling participants
• Research automation tools and workflows (including automated macrostrategy infrastructure) that help researchers be exponentially more productive by systematically breaking down strategic thinking into reusable components
• AI ghostwriting platforms for researchers (with custom instructions optimized for longtermist work) that remove writing as a bottleneck, letting researchers focus on strategic thinking while AI handles communication
• Platforms like Crux for systematically mapping and collectively evaluating crucial considerations, creating a "firehose of the most important ideas" that enables rapid knowledge transfer across the community
• Coordination platforms and collaboration mechanisms for sharing AI best practices, workflows, and human-AI interaction architecture insights, including GitWise-style systems where community members rate and refine each other's tools
• Human-AI Symbiosis infrastructure that provides AI with comprehensive personal and collective context, enabling longtermist researchers and field-builders to leverage AI as "mega-coaches" that dramatically multiply individual effectiveness
• Community knowledge management systems and best practice repositories that make collective wisdom easily discoverable and reusable

These community infrastructure interventions have the unique advantage that they generate additional capacity to create a greater quantity of higher quality interventions. When one person develops a research workflow that increases productivity by 10%, and that workflow is shared community-wide through a tool library, the community's collective research output could potentially increase by the same margin. Each such infrastructure project multiplies the impact of all subsequent work.

This cooperative paradigm shift is fundamentally important. The longtermist community has many motivated individuals struggling to find paid longtermist roles. But if we create systemic infrastructure that is extremely good at upskilling people and helping them contribute to performing research, and developing and implementing interventions, we can achieve dramatic improvements in community-wide effectiveness by making much greater use of this latent untapped talent.

The key insight is that individual effectiveness is not solely determined by individual talent. It's heavily influenced by available tools, clear strategies, supportive networks, synergistic platforms, and systematic practices. By focusing on these communal resources, we can dramatically increase the productivity of typical community members, help them develop the specific skills they need to be high impact, and help highly effective individuals become even more impactful.

Systematic Value Reflection as Core

If the instrumental commoditization thesis is correct (if AI increasingly handles implementation while direction becomes crucial), then interventions focused on improving humanity's ability to determine good directions become centrally important. This means infrastructure (including institutional and technological infrastructure) for systematic value reflection.

We need mechanisms that help humans reflect on their values, experiment with different frameworks, engage in substantive debate, and systematically evolve toward better values. (again, from “Shortlist of Viatopia Interventions.”) These include:

• Psychological and mental health infrastructure (including AI coaches, therapists, and personal advocates) that help individuals explore and refine their values, process experiences reflectively, and develop wisdom necessary for long-term thinking
• Educational approaches that build strong epistemics, collaborative mindsets, and moral reflection capacity from childhood, exemplified by the Children's Movement which systematically improves how we raise children to create generations naturally inclined toward careful reasoning about values
• Crowdsourcing initiatives like the Good World Project that enable broad participation in envisioning good futures through AI-assisted conversational elicitation, systematically mapping the possibility space of desirable futures through Utopedia
• Forecasting and futarchy infrastructure that leverages AI's ability to model consequences and predict satisfaction with different goals, helping separate values (human decision) from beliefs about outcomes (AI-assisted prediction)
• Deliberative institutions and mechanisms for collective intelligence that aggregate diverse preferences while guarding against premature value lock-in, enabling society to converge on better values through structured dialogue
• Market-based mechanisms (like the Hybrid Market) that price values, incentivize scale-sensitive value-creating behavior, allow for explicit moral trade, and encourage moral progress by rewarding early investment in emerging types of value which society is currently under-valuing
• Perpetual reflection organizations that continuously map where humanity wants to go, facilitate synthesis between different visions, and nudge society toward collectively endorsed trajectories as values evolve
• Political and legal frameworks (like extended children's rights conventions) that enshrine systematic support for human development and value evolution while maintaining agency and preventing coercion

The Children's Movement exemplifies one approach: by systematically improving how we raise children, we create generations that are better equipped to navigate these challenges. But we need a whole ecosystem of value-improving interventions spanning individual psychology, social institutions, and AI-augmented deliberation.

Crucially, this work becomes more urgent as AI timelines shorten. We don't know whether we have decades or years before transformative AI arrives. Pre-AGI institutional design takes time to develop, test, and refine. The institutions and ideas we have ready when advanced AI arrives may be what we end up using, as there may be great pressure to make decisions relatively quickly about what path humanity should take once these capabilities exist. This creates strong reasons to develop these institutions now, even if their deployment is years away.

Keeping the Future Human

One important consideration for choosing between minimal and expansive approaches is uncertainty about what kind of future we want. Some envision AI rapidly optimizing everything according to well-specified values. But we might instead prefer to "keep the future human"—maintaining human agency and allowing human values to evolve carefully over time within societies that aren't radically different from our current experience. Note that this is a related but separate consideration from Anthony Aguirre’s primary rationale in his "Keep The Future Human" framework, which primarily emphasizes ensuring advanced AI is controllable.

If we take this more gradual path seriously, then expansive interventions focused on human psychological health, moral development, and institutional design become critically important. We need viatopian mechanisms that help human values evolve in positive directions without coercion, that maintain meaningful agency while guarding against catastrophic choices, and that enable us to eventually converge on excellent futures without rushing there immediately.

This connects to a key design criterion for viatopia: balancing agency and guidance. We want to maximize human freedom to explore different possibilities while systematically encouraging movement toward better values. This is a difficult design problem, but it's one worth solving if we want to preserve meaningful human choice in shaping the far future.

Conclusion and Integration

Greaves and Tarsney are right that minimal longtermism enjoys stronger evidential support than expansive longtermism in general. But I argue that certain expansive interventions (particularly those focused on infrastructure, values, and community effectiveness) are more tractable than their analysis suggests.

The key is recognizing that we're in a unique historical moment. AI is about to make implementation trivial but direction critical. We have a brief window to build the infrastructure (including institutional and technological infrastructure) that helps humanity reliably choose good directions. And we have a community of motivated individuals who could be far more effective with better coordination, tools, and systematic practices.

By prioritizing interventions that create broad safeguards against existential risks, systematically improve human values and institutional quality, maximize community effectiveness through infrastructure, and leverage AI in ways that compound over time, we can pursue a form of expansive longtermism that is both philosophically defensible and practically tractable.

Moreover, I agree strongly with Greaves and Tarsney's implicit emphasis: we must not downplay the importance of current generation’s wellbeing. As Cotton-Barratt and Hadshar emphasized in their analysis of longtermist societies, focusing on wellbeing is essential both for political viability and psychological health. Fortunately, there is no fundamental conflict between creating good worlds today and ensuring excellent futures. By prioritizing high-leverage, wellbeing-promoting interventions and ensuring everyone's needs are met, while simultaneously creating high leverage interventions for values research, reflection, experimentation, and debate, we can create a good present that naturally evolves into an even better future—and perhaps eventually the best future achievable.

In the next essay, "Why Viatopia is Important," I provide the theoretical foundation from my Deep Reflection work, explaining Will MacAskill's viatopia concept and why it matters for achieving the best achievable future. The essay introduces the multiplicative crucial considerations framework, showing why dozens to hundreds of interacting factors make comprehensive reflection potentially orders of magnitude more valuable than narrow approaches. It explores the instrumental commoditization thesis in greater depth, explains how diversity of viatopia paths prevents premature lock-in through MacAskill’s bootstrapping mechanism, and discusses parallels and differences between MacAskill's Better Futures framework and my own work. This theoretical foundation establishes why the practical mechanisms explored throughout this series are essential for navigating the challenges ahead.