Capabilities & Obligations

Humans, alone among the loci currently in the biosphere, hold certain protective capacities the biosphere itself cannot generate.

The cleanest case is asteroid defense. Other plausible candidates include broad-spectrum pandemic response and large-scale ecological restoration after sufficient damage; more speculative deep-time candidates include off-world refugia or interstellar relocation. Each is contingent on continued technological civilization, but each is real in the sense that the capability exists or could exist within reachable development paths and no other current locus in the biosphere has any approach to it.

The question this raises is interesting and not as simple as it looks: what does it mean, structurally, that some kinds of biospheric protection are reachable only through human agency? Does this give humans a special moral standing? Does it vindicate the species against the harm humans also produce? Is "we deflect asteroids" a moral credit that offsets "we drive thousands of species extinct"?

Modal Path Ethics has tools for this. The tools say something more interesting than either the vindication argument or its biocentric rejection. They say: capability creates obligation, not credit. The obligation is real and demanding. The idea of credit is a structural error.

Background: Existential Risk and the Longtermist Argument.

The contemporary literature in this neighborhood was shaped heavily by Oxford and adjacent institutions, especially through existential-risk studies, longtermism, and the broader effective altruism orbit.

The community's central figures include Nick Bostrom, who founded the (now-closed) Future of Humanity Institute in 2005 and made existential risk a recognized academic topic; Toby Ord, whose 2020 book The Precipice argues that humanity sits at a uniquely dangerous and uniquely opportunity-rich moment; and William MacAskill, whose What We Owe the Future (2022) extends the analysis to claims about long-term moral significance generally.

The argument structure these thinkers share, simplified for readers who haven't followed the field, runs roughly:

The future could contain enormous numbers of people, possibly many orders of magnitude more than have ever lived, if humanity reaches its long-term potential. The moral weight of this potential population is, by simple aggregation, enormous. Threats that could prevent this future from being reached, such as extinction, civilizational collapse, or lock-in to permanently bad states, are therefore morally enormous in a way that ordinary present-moment harms are not. The work of preventing existential threats is, on this view, possibly the most morally important work humans can do.

That argument has produced real policy outputs. Funding for AI safety research has expanded partly through this reasoning. So has biosecurity work. Asteroid detection and deflection also overlap with the same existential-risk concern, though planetary defense has its own older NASA, congressional, and scientific lineage rather than being mainly a longtermist policy output. This community is still academically very serious, has produced major books, and has affected policy in concrete ways. It is not just an internet subculture.

It is also a community that has run into substantial criticism, some of it well-founded, some of it less so.

The criticisms relevant to Modal Path Ethics' engagement are mostly methodological. The argument's core move is aggregative: sum the welfare of vast future populations, and the sum dominates anything in the present. This is utilitarian arithmetic at scale. It produces conclusions that look strange to many critics, like that small reductions in extinction risk should dominate even substantial present-day humanitarian interventions, that our generation's primary moral duty is to ensure other generations exist.

The framework Modal Path Ethics has been developing is, on commensurability grounds, structurally incompatible with this aggregative move.

But the framework also doesn't reject the underlying insight that existential threats matter or that humans have specific protective capacities the biosphere lacks. Those facts are very real. The question is what to do with them.

What Capability Actually Creates.

Cleanly: a locus that holds a unique structural capacity for preserving field-continuation thereby holds a structural obligation to maintain and deploy that capacity responsibly. The obligation grounds in the capacity, not in the locus's identity. It transfers if the capacity transfers. It evaporates if comparable alternatives appear.

Three points of clarification matter for what follows.

First, the obligation is not a moral elevation. The locus that holds the capacity is not thereby graded as good. The capacity creates a specific structural responsibility: to maintain the capability, to develop it where development is feasible, to deploy it when needed, to refrain from misusing it. None of these transfers to any verdict on the locus as a kind of being. The framework does not grade kinds. It grades structural relationships.

Second, the obligation is not a credit. Maintaining the asteroid defense capability is doing what the structural facts require here. It is not a generous act for which gratitude is then owed. The locus that fails to maintain the capability is failing an obligation. The locus that succeeds is meeting an obligation. Neither is virtuous in any species-elevating sense; both are structurally appropriate or inappropriate responses to the position the locus is in.

Third, the obligation does not aggregate against unrelated harms. A locus that holds Capacity X for protection against Threat Y, while simultaneously producing Harm Z that the capacity does not address, does not earn moral credit for X that offsets the moral debit of Z. The two are structurally distinct facts in the field. Each generates its own obligation. Each has its own analysis. The arithmetic that would let "asteroid defense funding" cancel "industrial extinction crisis" is exactly the aggregative arithmetic the framework's commensurability position rules out.

These three clarifications do most of the work. The capability creates the responsibility, not the standing; the maintenance is the meeting of an obligation, not a virtue; the obligations don't aggregate across unrelated structural facts. Hold these and the rest follows.

The Asteroid Defense Case.

Let's now try to run a weighting analysis on capability-maintenance versus capability-non-maintenance.

Without sufficient asteroid defense capability, the biosphere is exposed to large impacts, including the rare extinction-class impact events. Such events have definitely happened in the geological past. The Chicxulub impact about sixty-six million years ago is the canonical recent extinction-scale case. They will continue to occur on probabilistic timescales. The harm of an extinction-class impact is total biosphere contraction. Severity is maximum. Irreversibility is total. Breadth covers every locus. Centrality is total; every reachable future depends on the biosphere continuing. Asymmetry is catastrophic; a discrete external event closes everything. Distribution is universal; every locus shares the cost.

So, to be avoided.

The cost of maintaining the capability is moderate relative to what is preserved. Scientific infrastructure for detection (telescopes, computing, organizational continuity). Technological capacity for response (rocket delivery, deflection methods). Some opportunity cost for the resources involved. Some risk of dual-use misapplication.

The variables don't trade off here. Maintenance just dominates non-maintenance on every weighting variable that matters. The cost is small relative to what is preserved. This is a textbook dominance case from the commensurability essay: comparison is determinate, the framework yields a clear answer, no genuinely hard arithmetic is required.

Conclusion: the capability should be maintained, clearly. The obligation falls on the locus that uniquely holds the capacity, which is currently humans. Dinosaurs never saw this one coming.

This conclusion converges with what longtermists also conclude, but for structurally very different reasons. Longtermism gets there through aggregating the welfare of vast future populations and finding that the sum dominates present resource allocation. Modal Path Ethics gets there through structural dominance reasoning that doesn't require aggregation across persons or across futures. We just look at the field as it stands and the choice of transitions. The two frameworks reach the same practical recommendation through different metaphysics.

This matters highly for the cases where the practical recommendations now diverge.

Longtermism's aggregative reasoning produces the repugnant-conclusion-adjacent claim that small reductions in extinction risk should dominate substantial present harms. Modal Path Ethics doesn't produce this claim because it doesn't aggregate. Both frameworks equally endorse asteroid defense maintenance. They part ways on cases like "should we accept large present-day harms in exchange for tiny reductions in extinction risk." Modal Path Ethics' structural analysis there does not produce the longtermist conclusion at all. The dominance reasoning that endorses asteroid defense doesn't extend to endorsing arbitrary present sacrifice for marginal long-run probability shifts. The framework has tools for distinguishing these cases that pure aggregation does not have, and does need.

Capability Is Dual-Use by Structure.

Here is where the framework's analysis now sharpens.

The infrastructure that allows asteroid deflection is the infrastructure that allows precision kinetic-impact weaponry. The scientific cognition that produces vaccine development produces gain-of-function research. The geoengineering capacity that could mitigate climate damage produces climate-modification weapons. The AI capability that could accelerate medical research can accelerate destabilizing surveillance and autonomous weapons. Each of these is the same set of underlying capacities deployed out in different directions and embedded in different fields.

This is not a coincidence or a bug. It is structural reality. Capability is dual-use because capability is simply the ability to intervene in the world at scale. Once you can deflect an asteroid, you can deflect other things into other targets. Once you can engineer a pathogen, you can engineer a pathogen with different properties. Once you can model and intervene in climate systems, you can intervene in directions that don't preserve. The protective and the destructive uses share the same skill set, the same infrastructure, the same scientific community, often the same individuals. There's no escaping this one.

This complicates the vindication argument profoundly. The "humans uniquely protect" claim cannot be cleanly separated from the "humans uniquely threaten" claim at all, because the exact same capacity is doing both, just embedded elsewhere. A species that has nuclear weapons and asteroid deflection both is not getting the protection without the destruction. The capability is a package deal.

What this means for the structural analysis: the obligation is two-sided. Maintain the capability for protection. Discipline the capability against misuse for destruction. Both halves of the obligation are equal and ground in the same capacity. Failing either half is failing the obligation in the same way. Succeeding at one while failing the other is not partial credit here; it is partial fulfillment of a two-sided structural responsibility, and failure.

The framework can be more specific now. The Technological Singularity article warned about capability that produces protection at the cost of foreclosing playable extance, meaning alignment to "humanity as pattern but not agency, life as biomass but not ecology," compression as reality because we make the same mistake. The same warning applies to capability deployed for asteroid defense or pandemic response. A protection regime that produces its own structural contraction in fields the protection doesn't reach is not a clean win. It is two structural facts that need to be analyzed together. The framework asks: what is being preserved, what is being closed, what is being held in tension, what is the residue.

This is also the answer to a question the longtermist framework struggles with.

For aggregative existential-risk reasoning, what matters is reducing total probability of catastrophic outcomes. The means by which the reduction happens are secondary. If a more authoritarian global governance reduces extinction risk by twenty percent, a crude enough aggregation may endorse it.

Modal Path Ethics' structural analysis does not. The means are part of the field configuration; closing playable extance to reduce extinction probability is not a free trade. The framework can endorse asteroid defense funding without endorsing the political concentrations that some longtermist analyses imply.

The High-Emissions vs Low-Emissions Case.

Now, let's run the comparison this structural analysis requires from us.

The asteroid defense case has a particular structural shape: threat is exogenous (asteroids existed before humans), capability is protective, action is maintenance, cost is moderate ongoing investment, benefit is probabilistic protection over geological time. The framework endorses maintenance through clean dominance reasoning.

The high-emissions case has a different shape entirely: threat is endogenous (emissions come from human activity), capability is irrelevant in the same sense (the issue isn't whether humans have the capacity to address it, but whether humans will constrain the activity producing it), action is restraint, cost is forgone industrial activity and required reorganization, benefit is preserved field configurations across the entire reachable future of the biosphere.

These cases are clearly not the same kind of case.

In the asteroid case, the agent is a responder to an external threat. In the climate case, the agent is the cause of the threat. The protective relationship in the first case has no analog in the second. The structural obligation is differently grounded.

Next, try running weighting on emissions paths.

Severity: high-emissions paths produce catastrophic climate damage. Extreme heat events, sea level rise, ecosystem collapse, agricultural disruption, disease range expansion, mass displacement. Severity is very high.

Irreversibility: substantial. Some damage is fast (extreme weather), some is slow (sea level rise), but a significant fraction of the damage is functionally irreversible on civilizational timescales. Ice sheet collapse, mass extinction, ocean acidification, permafrost methane release have feedback dynamics that lock in damage even after emissions stop.

Breadth: every locus in the biosphere is affected, with disproportionate harm to vulnerable ecosystems, low-income human populations, and species without migration capacity.

Centrality: high. Climate stability is upstream of nearly every reachable future for most loci.

Asymmetry: severe. Burden distribution is wildly uneven; populations that contributed least to emissions bear most of the damage, and the damage is concentrated in regions and populations with least adaptive capacity.

Distribution: deeply asymmetric. The benefits of emissions-producing activity have been distributed (unevenly) over the past two centuries; the costs are concentrated in the present and especially in the future, with vulnerable populations bearing disproportionate share.

So, not great. Let's compare to low-emissions paths.

The cost of low-emissions paths is forgone industrial activity, which has its own complications, since some present-day populations depend on emissions-producing economies for material survival. The cost is real, redistributively complex, and ethically demanding. Low-emissions paths also involve transitions that themselves have moral content, including workers displaced, regions hollowed out, infrastructure left stranded. These are not nothing at all.

But they are not catastrophic in the way the high-emissions costs are. They are bounded. They are distributable. Many are reversible or repairable; most of the costs of transition can be addressed cleanly through redistribution, retraining, regional investment. The structural shape is fundamentally different from the high-emissions costs, which compound, irreversibly, across all reachable futures, forever.

The framework therefore clearly endorses low-emissions paths through the same dominance reasoning that endorses asteroid defense maintenance, though here the dominance is less complete and the framework has to do real commensurability work on transition costs and burden distribution rather than just yielding a clean answer. But still, the result is clear.

The crucial point for the present argument: this case structurally is visibly not a place where humans get to "trade" capability-credit for restraint-debt. The asteroid case endorses capability maintenance. The climate case requires capability restraint. The two are structurally separate facts and in fact opposed. The framework gives them separate analyses. Each analysis grounds in the structural relations of that specific case. The aggregation that would let one offset the other is not available to you here.

This is the core anti-vindication move. Running both analyses honestly produces two distinct obligations: maintain capability against external threats; restrain capability where it produces internal threats. Both obligations apply to the same agent. Neither cancels the other. Failing one is not licensed by succeeding at the other.

Why Species-Vindication Fails.

With the comparison in hand, the structural failure of the vindication argument becomes visible.

The argument's move is to take the asteroid defense case (or some adjacent capability) and use it to credit the species against unrelated harms. The move requires aggregation: you have to be able to add up the protective capacity and subtract the destructive activity and get a verdict on humans-as-such. The framework refuses the aggregation operation.

But the failure is deeper than just refusing aggregation. The vindication is also confused about what kind of thing humans are within the framework. Modal Path Ethics doesn't grade species, only structural relationships. "Humans are good" is asking the framework to do something the framework does not do. There is no structural fact corresponding to "the moral standing of humans as a kind." There are many structural facts corresponding to specific human activities, capacities, relationships, and obligations to write about. None of them sums to a species verdict.

This is not some peculiar limitation of Modal Path Ethics. It is the framework recognizing what species-membership actually means. A species is a category. Categories don't have moral standing in the way structural patterns do. Humans qua humans is not a thing that can be Better or Worse. Specific human practices, institutions, capabilities, and decisions can be analyzed structurally and judged structurally, certainly, but those structural judgments don't aggregate to category-judgments any more than the structural analysis of one chess game aggregates to a verdict on chess as a kind. That's just not how this works.

Once you see this, the vindication argument starts to look suspiciously a lot like a cousin of the utility-monster argument.

Both move from local structural facts to category-level conclusions through illegitimate aggregation. The utility monster says "concentrated welfare in this being is large, therefore feeding the monster is good." The species-vindication says "concentrated protective capacity in this species is large, therefore the species is good." Both arguments share a common structural mistake: treating quantities of structural facts as license to skip the structural analysis and produce category-level verdicts. The framework rejects both for the same very good reason.

There is also, as always, the symmetric failure mode worth naming: the species-condemnation argument. "Humans are uniquely destructive, therefore humans are bad" is structurally identical to species-vindication, just running in the opposite direction. Also wrong. Both treat the species as the relevant unit of moral evaluation. Both ignore the structural specificity that the framework requires. Modal Path Ethics always refuses both moves, not just the vindication. A serious analysis of human structural responsibility refuses to grade humans as a kind in either direction. It analyzes specific structural facts and assigns specific structural obligations.

What Modal Path Ethics Actually Says About Humans and the Biosphere, Briefly.

Strip the analysis down to its load-bearing claims:

Humans hold capabilities the rest of the biosphere does not. This is empirically true and structurally significant. The capabilities create specific obligations, both positive (maintain protective capacities) and negative (refrain from destructive ones, restrain endogenous threats).

The obligations don't aggregate to a species-level verdict. There is no operation in the framework that lets the obligations cancel each other out or sum to a kind-judgment.

The capabilities are structurally dual-use. The same infrastructure and cognition that allows protection allows destruction. The protective and destructive capacities cannot be cleanly separated, because they are the same underlying capacity deployed in different directions.

The framework converges with longtermist thinking on specific recommendations like maintaining asteroid defense capability, while diverging from longtermist thinking on the metaphysical grounding of the recommendation. Aggregative welfare reasoning is not the framework's tool. Structural dominance and structural obligation are.

The framework converges with environmental ethics traditions on specific recommendations like reducing emissions, while diverging on grounding. The framework does not require treating the biosphere as a moral patient with intrinsic value or as a holistic entity with standing. It requires treating the structural future-space of the biosphere as morally real because structural future-space is what the framework treats as morally real, full stop.

The framework rejects species-vindication and species-condemnation alike. Neither is a legitimate move in Modal Path Ethics' grammar. What is legitimate is the structural analysis of specific patterns of human activity, capability, and obligation. Some of those patterns are field-preserving. Some are field-contracting. Naming them honestly, on their own structural terms, is what the framework does. Producing a verdict on humans-as-such is what the framework refuses to do.

What humans actually owe the biosphere, to the degree such a thing can truly be said within this framework: the maintenance of protective capacities humans uniquely hold; the restraint of destructive capacities humans uniquely hold; the truthful contact with what the structural facts actually are; the willingness to bear the costs of the obligations rather than smuggling in compensating goods that don't actually balance the structural ledger.

This is enough to get going with. It is also a substantial obligation that humans, as a locus, are largely failing to meet at the present moment, which is not great. The framework's analysis is not a vindication of human present conduct. It is also not a condemnation of the human kind. It is a structural account of what the position humans hold actually requires, and an honest naming of where the actual conduct falls short.

Closing.

The temptation of vindication is real. It is more comfortable to read about asteroid defense and conclude "humans are now essential to biosphere continuation" than to read about the same case and conclude "humans hold a structural obligation they are largely meeting in this domain and failing in many adjacent ones." The first conclusion is flattering. The second is honest and kinda wordy.

The framework's posture is that flattering conclusions about one's own kind are exactly the structural distortion the framework was built to resist.

The same expectation-driven cognition that produced N-rays produces species-vindication: people see what they want to see, in conditions that don't independently verify, and call the result evidence. The vindication argument is the cosmic-scale version of the move; humans, looking at humans, finding that humans are uniquely valuable.

What a coincidence. The framework treats this with the same skepticism it would apply to any locus's self-flattering analysis of its own moral standing.

What the framework offers instead is more useful. Specific capabilities create specific obligations. The obligations are demanding. They are also the actual thing humans owe the biosphere by virtue of holding particular structural positions that load particular structural responsibilities, not because they are human. Meeting the obligations is what's required. Naming them as credits would be exactly the kind of self-aggrandizing distortion that prevents them from even being met.

The framework is not anthropocentric. It is also not biocentric in the way some environmental ethics traditions are. It is structural. It analyzes specific patterns. It assigns specific obligations. It refuses to grade kinds.

This is what the position humans hold actually means. It is also what the framework looks like when it is being honest about a tempting question rather than giving you the comforting answer.