The Argument

There is a certain grave irony in the position of Ethan Muse, Joe Schmid, and rationally informed Catholics more generally who defend the doctrine of Eternal Conscious Torment (ECT): they are plausibly exactly the sort of people who would end up in hell if God exists.

The argument is very simple.

Assume that God exists.

If someone sincerely and reflectively affirms that eternal conscious torment is compatible with moral perfection and perfect love, then that person is morally corrupt to such an extent that they deserve to go to hell.

This premise should be taken very seriously. Eternal conscious torment is not merely a severe punishment, an exceptionally long punishment, or even a punishment of unimaginable intensity. It is suffering without end. There is no final restoration, no eventual release, no point at which the suffering has accomplished its purpose and comes to an end. The person suffers, and suffers, and suffers—and there is literally no future moment at which this ceases to be true.

To understand what this means and nevertheless affirm that such a fate is compatible with perfect love and perfect moral goodness is not an ordinary moral error. It represents a truly immense corruption of moral judgment.

There are mistakes about morality that reasonable and decent people can make. There are difficult questions on which thoughtful people can disagree. But there comes a point at which calling something a mere “disagreement” risks obscuring what is actually being affirmed. If someone were to contemplate the endless conscious torment of another person, fully understand that the torment will never stop, and then declare that inflicting or sustaining this state is perfectly compatible with being maximally loving and maximally good, something has gone very seriously wrong with that person’s moral judgment.

And the qualification matters: the argument concerns those who defend ECT sincerely, seriously, and reflectively. Ignorance, confusion, inherited formulas, or failure to have thought carefully about the doctrine are one thing. But someone who has considered the objection, grasped the nature of eternal torment, understood what “eternal” means, and nevertheless deliberately maintains that it can be an expression of perfect justice, perfect goodness, and perfect love is in a different position.

The more clearly one understands what ECT entails, the more morally disturbing its endorsement becomes. At some point, the willingness to call endless torment “perfectly good/moral/just” ceases to be a simple theological mistake and begins to look like evidence of a grave deformation of one’s moral sense.

And if God exists and is morally perfect, then such moral corruption presumably matters. Indeed, if there really are states of character sufficiently corrupt to merit damnation, it is difficult to think of a more plausible candidate than becoming so morally disoriented that one can look upon everlasting conscious torment and sincerely call it compatible with perfect love.

So the reasoning is straightforward:

1. God exists and is morally perfect and perfectly loving. (let’s assume for the argument)

2. Sincerely and reflectively affirming that eternal conscious torment is compatible with moral perfection and perfect love manifests extreme moral corruption.

3. Extreme moral corruption of this kind deserves damnation.

4. Therefore, people who sincerely and reflectively defend ECT deserve damnation.

Ethan Muse, Joe Schmid, and all Catholic infernalists who seriously and rationally defend ECT fall into that category.

Conclusion

This is the grave irony. The people insisting that eternal conscious torment is compatible with the love and goodness of God may, by the very act of sincerely and reflectively defending that proposition, be displaying precisely the kind of grave moral corruption for which damnation would actually begin to seem plausible.

That conclusion is partly ironic, but the underlying point is not a joke. If one genuinely believes that perfect goodness can look upon the endless conscious agony of a person and say, in effect, “this is just, this is good, and this is compatible with perfect love,” then one should perhaps be considerably more worried about the condition of one’s own soul than about the alleged damnation of unbelievers, heretics, or anyone else.

One can only hope, for the sake of their souls, that they reconsider the doctrine before it is too late.

The important point is that this judgment is not normally about one particular agent. We do not mean merely that I should prevent X. We mean that X should be prevented, full stop. If I am in a position to prevent it, I should. If another agent is in a position to prevent it, that agent should as well. When we say that it would have been good if we could have prevented the Holocaust, we do not merely mean that it would have been good if we had prevented it. We mean that it would have been good if it had been prevented, by whoever was in a position to prevent it.

And that includes God.

If X really should be prevented, and God knows that X will occur and can prevent it, then God should prevent it. If God does not, one of two things must be true.

Either X really should have been prevented, all things considered. In that case, God allowed something to happen that should not have happened.

Or God had a sufficient reason to allow X. But then X did not, all things considered, really need to be prevented.

This creates an immediate consequence for human agency.

Suppose I could have prevented X, but did not. If the theist says that God was justified in allowing X, then the resulting state of affairs was, all things considered, permissible. I may still have acted badly from my own epistemic perspective. I may have had bad intentions, failed to follow the evidence available to me, or violated some rule that I ought to have followed. But I did not make reality become, all things considered, a way it should not have been. If I had, then God would have allowed something that should not have been allowed.

The same reasoning works in the opposite direction.

Suppose there is some good state of affairs Y that should be brought about. If Y really should be brought about, then any agent capable of bringing it about has reason to do so, including God.

If God is justified in allowing Y not to occur, then Y did not, all things considered, really need to occur.

So if I bring about Y, I may locally improve the world. But I cannot be correcting an ultimate normative deficiency in reality. If reality genuinely lacked something that should have been there, then God had already allowed the world to be less than it should have been.

The result is symmetrical.

Under divine providence, created agents cannot make reality ultimately worse than it should be, because God would then be permitting something that should not be permitted.

But they also cannot make reality ultimately better than it should otherwise have been, because that would imply that God had permitted an actual deficiency that should have been corrected.

So the theist appears to face a deeper dilemma.

Either created agents can genuinely make reality better or worse, all things considered.

If so, then God can allow reality to become a way it should not be, or allow it to remain worse than it should be. That threatens divine perfect goodness.

Or God guarantees that every state of affairs he ultimately permits is morally justified, all things considered.

If so, created agents lose their ultimate normative significance. They cannot make reality become, in the final moral sense, better or worse than it should be.

Whatever they do is absorbed into an order that remains ultimately justified.

The point I’m making there is very analogous to my critique of the notion of karma.

A perfectly functioning karmic order can absorb every possible outcome. If I help someone, that result is compatible with the karmic order. If I do not help them, their suffering is also compatible with the karmic order. Either way, the final state of affairs is explained as part of the same perfectly just structure.

A strongly providential God plays the same role as karma.

If I save someone, that outcome is compatible with divine providence.

If I fail to save them and they die, the theist may say that God had sufficient reason to allow that outcome as well.

If I bring about some good, that is compatible with providence.

If I fail to bring it about, its absence can also be said to have been sufficiently justified.

The problem is not merely that such a theory is difficult to falsify. The deeper problem is that both karma and divine providence can function as universal moral correctors. They ensure that whatever created agents do (or don’t do), reality never becomes, all things considered, morally out of order.

And if that is true, created agents become morally irrelevant: they cannot improve or damage the ultimate normative condition of reality.

The theist therefore faces a stark choice:

Either created agents can genuinely make reality better or worse, all things considered, in which case God can permit reality to be other than it should be; or God guarantees the ultimate moral order of reality, in which case created agents cannot genuinely improve or worsen that order.

The first option threatens divine perfect goodness.

The second threatens the ultimate moral significance of created agency.

That is the dilemma.

Perfect-being theism postulates the existence of a perfect being: roughly, a disembodied, omnipotent, omniscient, morally perfect, eternal being. Before turning to the usual arguments for and against the existence of God, I think it is worth evaluating this theory at a more fundamental level.

The result, in my view, is that perfect-being theism is simply a bad theory.

1. Internal coherence

The first question is the most basic one: is the theory internally coherent?

Consider the attributes attributed to this perfect being.

It is already far from clear that omnipotence and omniscience are possible. Both notions generate familiar logical tensions and paradoxes, and there is a substantial philosophical literature attempting to determine whether coherent formulations of either property can be given.

Moral perfection raises a similar problem. If moral perfection presupposes some form of free will and moral responsibility, then arguments against the possibility of free will or moral responsibility become, indirectly, arguments against the possibility of a morally perfect being.

The same issue arises with the idea of a disembodied mind. There are serious arguments for the view that mentality necessarily requires some form of physical embodiment or realization. If a disembodied mind is impossible, then perfect-being theism is again in trouble at the level of its defining attributes.

None of this by itself establishes that perfect-being theism is incoherent. The point is simply that even at the most fundamental level—before considering any empirical evidence—the theory already depends on several highly controversial claims about what is possible.

Suppose, however, that we grant all of this. Suppose that omnipotence, omniscience, moral perfection, eternality, and disembodied mentality are all coherent.

We can then move to the next question.

2. Predictive power

What does perfect-being theism actually predict?

The answer seems to be: very very very little.

Omniscience predicts virtually nothing about what kind of world we should observe. Neither does omnipotence. Neither does being disembodied. Neither does eternality.

The only attribute that seems capable of generating predictions is moral perfection. But even here, the predictive content is extraordinarily weak.

What kind of world would a morally perfect being create?

There is no remotely clear answer. Some might think that a perfectly good being would have no reason to create anything at all. Others might think that perfect goodness would require creating as many happy sentient beings as possible. And between these two extremes lies an enormous range of possibilities.

This is reflected in the sheer diversity of worlds that serious thinkers have regarded as compatible with divine moral perfection.

Again, the point is not that all of these positions are equally plausible. It is that perfect-being theism itself does remarkably little to decide between them.

The range of possible worlds compatible with the theory is therefore immense. Perhaps the theory rules out certain extreme cases. But beyond that, it seems to generate almost no clear predictions about what reality should actually look like.

And that is a major problem for a theory.

3. The problem with a posteriori arguments

This is where the weakness becomes especially important.

Most philosophical debate about the existence of God revolves around a posteriori considerations: fine-tuning, the problem of evil, divine hiddenness, the Kalam cosmological argument, religious experience, the existence of consciousness, and so forth.

But for an observation to count as evidence for or against a theory, we first need some idea of what the theory led us to expect.

If perfect-being theism makes almost no clear predictions a priori, then it becomes extremely difficult to determine what any particular observation should count for a posteriori.

Take fine-tuning. Is a finely tuned universe what we should expect if perfect-being theism is true? The theory itself does not clearly say.

Take suffering. Is the amount and distribution of suffering in our world improbable if a morally perfect God exists? Again, the theory itself does not clearly say.

Take divine hiddenness. Should a perfect being make its existence obvious, partially accessible, or deeply ambiguous? Once again, perfect-being theism does not give us a clear prediction.

The same problem recurs again and again.

This is why I am not even sure how intellectually informative many of the standard arguments for and against God ultimately are. The problem comes before the arguments themselves. We lack sufficiently clear prior predictions against which the observations could be evaluated.

If a theory is compatible with an enormous range of possible worlds, then observing one particular world does relatively little to tell us whether the theory was right or wrong. At the very least, its confirmatory or disconfirmatory significance becomes extremely difficult to establish.

Perfect-being theism therefore suffers from a basic methodological problem: it gives us very little predictive structure with which to interpret the evidence that is subsequently invoked for or against it.

4. Why disagreement about theism is so persistent

I think this also helps explain a striking feature of the philosophy of religion: highly competent philosophers can examine the same world and arrive at radically different conclusions, ranging from atheism to Catholicism, other forms of Christianity, and many other religious positions.

This degree of disagreement becomes less surprising once we recognize how little the underlying theory predicts.

If perfect-being theism gave us relatively precise expectations, philosophers could compare those expectations with the world and determine how well the theory performs. But when almost no clear expectations are generated in the first place, much of the debate inevitably turns into judgments about what seems compatible with a perfect God.

One person finds it intuitive that a perfectly good God would allow suffering for the sake of freedom. Another finds the observed suffering radically incompatible with perfect goodness.

One person finds fine-tuning exactly the kind of thing a creator would produce. Another sees no reason whatsoever why an omnipotent being would be expected to create such a universe.

One person thinks divine hiddenness makes sense under theism. Another thinks it is almost impossible to reconcile with a perfectly loving God.

At the surface level, sophisticated arguments can of course be constructed around all of these intuitions. But at a more fundamental level, the disagreement often concerns what one intuitively expects a perfect being to do.

And the reason there is so much room for such radically different intuitions is precisely that the original theory did not constrain those expectations very much in the first place.

In the absence of clear predictions, evidential judgments risk amounting, at bottom, to little more than competing gut feelings about what seems more or less likely under theism.

Conclusion

The problem with perfect-being theism therefore arises before we even reach the familiar arguments about God’s existence.

First, it is not obvious that the properties constituting a perfect being—omnipotence, omniscience, moral perfection, disembodied mentality, and so on—are even mutually or individually coherent.

But suppose that they are.

The more serious problem is then that the theory has extremely little predictive power. Most of its attributes predict virtually nothing, and the only attribute that plausibly constrains the space of possible worlds—moral perfection—appears compatible with an enormous range of radically different realities.

This, in turn, undermines our ability to interpret a posteriori observations as clearly confirming or disconfirming the theory. And it helps explain why highly competent philosophers can look at the same evidence and reach radically different conclusions: where the theory supplies few clear expectations, intuition has enormous room to fill the gap.

That is the fundamental problem.

Before asking whether the evidence favors perfect-being theism, we should first ask whether perfect-being theism gives us clear enough expectations for the evidence to meaningfully favor or disfavor it at all.

In my view, it largely does not.

And that is why perfect-being theism is a bad theory.

The Argument begins from a relatively simple anthropic idea. If we accept the Self-Indication Assumption (SIA), then the fact that I exist as an observer favors, ceteris paribus, hypotheses according to which there are more observers.

A theistic version of the argument then proposes that a perfect being could naturally generate such abundance. If God is omnipotent, omniscient, and morally perfect, and if creating worthwhile lives is good, then perhaps God would create all, or an unsurpassably large number, of the observers to whom he could give lives worth living. My own existence would consequently become more probable under some form of maximizing theism. Matthew Adelstein appears to have developed the argument first in his public writing, before developing its academic version together with Amos Wollen.

I do not think this argument is devoid of interest. But I do think it becomes considerably less impressive once its assumptions are unpacked. Its apparent force depends on moving rather quickly across several distinct inferential steps, each of which requires substantial concessions.

My criticism is cumulative. First, even granting SIA, its application to the global structure of reality is epistemically fragile. Second, even granting both SIA and its applicability, its distinctive evidential pressure points first toward theories that realize the maximum possible observer population, not toward perfect-being theism. Third, even after making the very generous concession that theism has sufficiently favorable independent priors to remain competitive despite this disadvantage, deriving abundance from divine perfection requires additional assumptions about population ethics and divine action. Fourth, once we look not merely at the existence of observers but at what their lives are actually like, a welfare problem arises. A perfect afterlife can be invoked to answer part of that problem, but it immediately raises issues of predictive specificity and leaves the deeper problem of evil untouched.

At that point, the question becomes whether we still have a strong argument for God, or merely an increasingly elaborate account of how God could be made compatible with the evidence.

1. SIA and the Epistemic Fragility of the Inference to Abundance

Let us begin by granting SIA, although SIA itself is controversial.

Within an already specified hypothesis space, SIA can have a powerful effect. If two hypotheses are otherwise comparable but one contains vastly more relevant observers than the other, my existence gives the former an anthropic advantage. Schematically,

and under an SIA-style treatment, the anthropic contribution to P(I∣Hi​) increases with the number of observers.

Given the right setup, the resulting Bayes factor could be enormous.

The problem is that “given the right setup” is doing quite a lot of work.

Applying SIA across cosmological or metaphysical theories presupposes some sufficiently well-defined hypothesis space, probability distribution, reference class, and account of the physical or metaphysical constraints governing observer production.

Our knowledge of these matters is extremely incomplete. There could be fundamental constraints on observer production that we simply do not yet understand. A better physical or metaphysical theory might reveal that certain apparently possible populations are impossible, that some range of observer numbers is independently expected, or that mechanisms which currently appear capable of producing enormous populations are not genuine possibilities at all.

SIA does not itself settle these background questions. If one theory appears to allow 10^1000 observers and another only 10^50, it would be epistemically reckless simply to declare the former overwhelmingly better supported before evaluating our independent reasons for believing the laws and metaphysics that generate those populations. Otherwise SIA begins to look suspiciously like a machine for awarding probability to whichever theory writes the largest number next to “observers.” This is closely related to the intuition behind the Presumptuous Philosopher objection.

My first claim is therefore deliberately modest:

SIA may provide very strong conditional evidence while remaining highly model-dependent

It may give us a reason to favor observer-rich hypotheses within a sufficiently justified theoretical framework. It is much less obvious that our present epistemic position licenses a robust inference from our existence to extreme observer plenitude across theories of reality that we barely understand.

For the rest of the argument, however, I will simply grant this issue. Suppose SIA is true, and suppose we really are epistemically entitled to apply it across the relevant theories.

2. SIA Points First Toward Maximal Observer Abundance

Once those concessions are made, something important follows.

SIA rewards the number of relevant observers. It does not reward their welfare, moral value, or the benevolence of whatever mechanism produces them.

Its distinctive pressure is therefore toward

maximal observer abundance,

not toward

net positive existence observer abundance.

And this point is stronger than mere underdetermination.

Consider perfect-being theism T. Even on an extremely creation-friendly version of theism, a morally perfect creator certainly does not actualize every possible observer. At minimum, there appear to be possible observers whose total existence would be irredeemably bad: observers whose entire lives, including any post-mortem existence, consist in horrific suffering with no compensating good or morally sufficient justification.

Perfect goodness gives God decisive reason not to create such lives.

So the class of observers actualized by perfect-being theism is only a proper subset of the space of possible observers:

By contrast, an axiologically neutral abundance mechanism need impose no such restriction. A neutral personal creator whose only terminal aim is observer maximization, an impersonal plenitude-generating mechanism, or some other unrestricted theory like modal realism could in principle realize all possible observers, including observers whose total lives are mediocre.

Nor is the class excluded by perfect-being theism some tiny fringe of observer-space. There are infinitely many possible observers whose total lives would be negative. Perfect goodness therefore excludes an enormous class of possible observers from actualization.

The qualitative direction of SIA is therefore clear:

SIA points first towards theories realizing the maximal relevant observer population

Perfect-being theism certainly does not belong to that class.

So SIA does not merely fail to uniquely select theism. Its distinctive selection pressure points toward unrestricted plenitude than toward the morally filtered plenitude available to perfect-being theism.

That is already a rather awkward result for an argument being presented as evidence for God.

Of course, this does not settle posterior probability. Perhaps perfect-being theism has much better independent priors than neutral maximization theories. Perhaps it is vastly simpler or strongly supported by independent arguments.

Fine. But that is now additional evidential work being done outside SIA.

Indeed, for the sake of argument, I am willing to make the following very substantial concession:

Suppose perfect-being theism has sufficiently superior independent priors to the relevant maximal-abundance alternatives that, despite receiving the smaller SIA advantage, it remains competitive or even preferable overall.

I regard this as an extremely generous concession. Within the present dialectic, I see no obvious reason simply to assume that an omnipotent, omniscient, morally perfect creator has a sufficiently better prior than every neutral plenitude mechanism to overcome its anthropic disadvantage.

But let us grant it anyway.

The Anthropic Argument still has a lot of work left to do.

3. Perfect Being Does Not by Itself Predict Observer Abundance

Begin with a relatively sparse perfect-being hypothesis:

If we are being offered an argument for God, this is roughly the kind of hypothesis whose predictive power ultimately matters.

But T0 does not straightforwardly entail observer abundance.

At minimum, two further claims are required.

First:

Second:

The first is a very controversial claim in population ethics. One can easily deny that bringing an additional happy person into existence makes things morally better than leaving that person nonexistent. On such a view, a perfectly good being might have no obligation to create anyone at all.

But even granting that creating positive lives is good, it does not immediately follow merely from this that a perfect being would respond by actualizing all, or an unsurpassably large number, of such lives. One still needs the further claim that divine perfection would actually lead God to perform this kind of value-producing creation on the relevant scale.

Thus the theory that does the predictive work is no longer merely T0, but something more like:

I am willing to be quite concessive here. It does not strike me as absurd to think that creating a net positive life is good, and it does not strike me as absurd either that a perfectly good being would act on that fact by creating very many such lives.

But we should not pretend that observer abundance simply fell out of the concept of perfection. It did not. Two substantive bridge claims were needed to get there.

For the sake of argument, however, let us grant those too.

Suppose a perfectly good God would indeed create an enormous population of observers whose total lives are positive.

Now a new problem arises.

4. The Welfare Problem and the Afterlife Response

If God creates observers because their lives are worth creating, what should we expect those lives to look like?

The lives we actually observe do not transparently look like an enormous collection of lives selected by perfect goodness for their positive value. Some human lives contain extraordinary amounts of misery, evil and suffering. The problem becomes even more striking once non-human animal suffering is included. Disease, starvation, predation, parasitism, natural disasters, severe psychological suffering, and apparently pointless agony are pervasive features of sentient existence.

At the very least, it is far from obvious on our present evidence that every actual sentient life is worth creating.

This creates an immediate problem for the creation-friendly version of the argument. If perfect goodness motivates God to actualize worthwhile lives, why do many actual lives look so catastrophically bad?

There is, however, a theistic response.

Perhaps the portion of existence we observe is only a tiny fragment of an observer’s total life. Suppose every created observer ultimately receives an indefinitely excellent afterlife. Then even a terrestrial life containing extreme suffering could nevertheless be overwhelmingly positive all things considered.

Let

Again, I am willing to be generous. Conditional on an omnipotent and perfectly good God, a universally good afterlife does not strike me as some utterly bizarre hypothesis. Perhaps it is quite reasonable.

So let us grant it.

This can answer one specific problem. A being whose earthly life looks net-negative may nevertheless satisfy

once the afterlife is included.

But this only solves the balance-sheet problem. It does not solve the deeper problem of evil.

A perfect afterlife can show that terrible suffering is eventually compensated. It does not show that the suffering was justified.

Compare:

200 years of horrific suffering + perfect afterlife

with

200 years of flourishing + perfect afterlife

Both lives may be overwhelmingly positive overall. But if an omnipotent and perfectly good God could produce the second rather than the first without sacrificing some greater good, why produce the first?

The observation that

does not answer the obvious question:

Why was all that misery there in the first place?

5. The Problem of Predictive Specificity

The afterlife move creates an additional problem.

If an arbitrarily good future can compensate for finite amount of pre-mortem suffering, then a vast range of radically different welfare distributions becomes compatible with the same enriched theistic theory.

Consider:

and even

All three can potentially satisfy

once a sufficiently good afterlife is appended.

But then the interesting question is no longer:

Can theism be made compatible with Q2​?

Of course it can, if enough compensatory machinery is available.

The interesting question is:

Why should Q2​ have been expected under theism rather than Q1​ or Q3​?

Compatibility is not prediction:

There is therefore a genuine tension between predictive specificity and ease of accommodation.

If perfect goodness strongly predicts lives containing little severe suffering, theism gains predictive content—but the actual world creates a sharper evidential problem of evil.

If, conversely, almost any finite distribution of suffering can be reconciled with perfect goodness by appealing to an afterlife, the theory becomes easier to protect from counterevidence—but also considerably less impressive as a predictor.

A theory deserves little evidential applause for “predicting” our world if, with a few auxiliary moves, it would have been equally comfortable with a radically different one.

6. The Problem of Evil as Comparative Evidence

Our evidence is not merely

it also includes

The relevant evidence is therefore something like

not merely E.

This matters because the different abundance mechanisms behave very differently once Q is introduced.

An axiologically neutral abundance mechanism can generate enormous numbers of observers while remaining completely indifferent to their welfare. A world containing a mixture of excellent lives and mediocre lives is therefore not puzzling at all under such a mechanism.

Perfect-being theism is different. It does not merely posit abundance. It says that the source of that abundance is omnipotent, omniscient, and perfectly good.

Consequently, the actual welfare distribution becomes evidentially significant.

At this stage, notice how much has already been granted:

• SIA is true;

• SIA can legitimately be applied across the relevant theories;

• despite SIA favoring more unrestricted plenitude theories, perfect-being theism has sufficiently favorable independent priors to remain competitive;

• creating net-positive lives is morally good;

• a perfectly good God would act on this good by creating very many such lives;

• and every created observer receives an excellent afterlife.

These are not trivial concessions.

And yet we still need an explanation for why the lives God creates contain the particular kinds and quantities of suffering we actually observe.

I am willing, for the sake of argument, to regard the previous assumptions as at least reasonably defensible. An adequate theodicy is another matter entirely.

To grant that there exists a morally sufficient explanation for the actual quantity, intensity, distribution, and apparent gratuitousness of sentient suffering is an enormous concession to make.

This is not another harmless auxiliary assumption to be added to the list.

Writing

does not make the problem disappear. The relevant claim would have to be that the suffering we observe is sufficiently unsurprising under perfect divine goodness because God has morally adequate reasons for permitting it.

But whether such reasons plausibly exist is precisely one of the central substantive disputes surrounding perfect-being theism.

The afterlife does not remove this burden.

Compensation is not justification.

A million years of bliss after torture may make the victim’s total existence positive. It does not retroactively explain why an omnipotent perfect being allowed the torture.

7. A Neutral Abundance Mechanism Avoids These Difficulties

We can now see a rather uncomfortable asymmetry.

Suppose N is an axiologically neutral mechanism tending toward maximal observer production.

Such a theory can predict abundance directly:

Indeed, insofar as it realizes a broader class of possible observers than morally constrained perfect-being theism, SIA itself points more directly toward N.

The theory also does not require the assumption that creating worthwhile lives is morally good. And it does not generate an evidential problem of evil from the observed welfare distribution.

Perfect-being theism, by contrast, has required us to travel through a progressively enriched package:

Within the dialectic of the Anthropic Argument, a neutral abundance mechanism enjoys three obvious structural advantages:

1. SIA itself points toward maximal observer production more than toward morally restricted observer production;

2. it predicts abundance without requiring the additional assumptions that creating positive lives is good and that a perfect being would therefore create them;

3. it accommodates the observed welfare distribution without generating a problem of evil.

That is quite a lot of theoretical baggage that the theistic account has to make up elsewhere.

8. From Argument to Conditional Theistic Accommodation

This leads to a broader methodological concern that extends beyond the Anthropic Argument.

A defender of theism can reply to each individual difficulty:

• a perfect afterlife is reasonably plausible given God;

• the proposition that creating positive lives is good is reasonably plausible;

• plausibly then a perfect God would therefore create very many such lives;

• perhaps some adequate theodicy is also available.

Suppose, very generously, that the first few moves are intellectually respectable.

What follows?

At most, we may have constructed a fairly coherent enriched theistic model:

under which the observed world is intelligible.

But this is not yet the same as showing that the observed world strongly supports T0​.

Showing that

is high does not establish that the evidence strongly favors the bare claim that an omnipotent, omniscient, morally perfect being exists.

Nor does it establish that theism predicts our evidence better than serious competing hypotheses.

And saying that each individual auxiliary move is “not very surprising” given God is not enough. That standard is cheap. Plenty of elaborate stories can be constructed from components that are individually not crazy.

The real question is whether the original theory independently leads us to expect the evidence, rather than whether we can progressively enrich the theory until the evidence stops looking embarrassing.

The pattern here is revealing:

observer abundance
→ creation-friendly population ethics and divine creation;

apparently negative lives
→ perfect afterlife;

horrific pre-afterlife suffering
→ theodicy.

Any one of those moves may have some kind of defense. But once enough of them accumulate, the alleged argument begins to look less like prediction and more like theory maintenance.

The central methodological distinction is therefore:

Coherent accomodation =/= Strong confirmation

A theory may allow us to tell a coherent story about why the world could look as it does if God exists without thereby giving us a strong reason to infer God’s existence from the fact that the world looks this way.

Or, more simply:

Making our evidence intelligible under God is not the same as making God probable given our evidence.

This is a broader concern I have with a number of purported arguments for God. They can sometimes appear powerful because the final theistic package has been polished until it fits the evidence reasonably well. But if the polish consists of a particular population ethics, a particular account of divine action, a universal afterlife, favorable independent priors, and a successful theodicy, one should ask how much evidential work is actually being done by the original claim that a perfect being exists.

At some point, “God explains the evidence” risks collapsing into the much weaker claim:

“Here is a sufficiently elaborate way in which God could coexist with the evidence.”

That is a at most a legitimate exercise in philosophical model-building.

It should not be oversold as a serious argument for God.

Conclusion

My objection to the Anthropic Argument is therefore cumulative.

First, even if SIA is granted, applying it across poorly understood cosmological and metaphysical hypotheses is epistemically precarious. Its evidential force may be good within a model while remaining highly dependent on uncertain background assumptions.

Second, even granting SIA and its legitimate applicability, its distinctive pressure points first toward theories that realize the maximal observer population. Perfect-being theism realizes only a proper subset of all possible observers. The excluded class is enormous.

SIA itself therefore does not naturally land on perfect-being theism. One has to redirect the argument toward God using independent considerations.

Third, suppose we nevertheless make the very substantial concession that theism possesses sufficiently superior independent priors to overcome this disadvantage. Perfect-being theism still predicts abundance only after adding at least two further claims: that creating net-positive lives is morally good, and that a perfectly good God would in fact act on this value by creating such lives on a vast scale.

Fourth, once those claims are granted, the lives we actually observe raise an immediate welfare problem. A universally excellent afterlife can answer the narrow concern that some lives might otherwise be negative overall. But that merely makes the lives compensable; it does not justify their suffering.

Fifth, the afterlife response itself weakens predictive specificity. If almost any finite pre-mortem suffering can be washed out by an indefinitely excellent afterlife, then radically different earthly welfare distributions become compatible with the same theory. Compatibility becomes cheap precisely when prediction becomes vague.

Finally, the deeper problem of evil remains. Granting an adequate theodicy would be an exceptionally large concession—far larger than granting an afterlife or a creation-friendly population ethics.

What eventually emerges is therefore something considerably weaker than a straightforward anthropic argument for a perfect being. We may be able to construct a coherent enriched form of theism under which observer abundance and the actual world can both be accommodated. But by the time enough assumptions have been installed to make everything fit, the original evidential triumph looks rather less triumphant.

The Anthropic Argument therefore seems considerably more successful as an exercise in conditional theistic accommodation than as a serious argument for the existence of a perfect being.

The argument for sentientism can begin with a simple moral question, for example:

Why is it wrong to torture a person?

I’d argue that the best explanations appeal to facts such as these: torture causes extreme undeserved suffering; it gravely frustrates the victim’s interests and desires; it violates consent and autonomy etc.

These are the properties doing the moral work.

By contrast, it does not seem necessary to mention that the victim belongs to Homo sapiens, has a particular genetic makeup or IQ, or belongs to any particular race or sex. None of these facts seems to explain why the torture is bad.

Once this is recognized, the argument for sentientism is straightforward.

If suffering is bad because of what suffering is, then suffering matters wherever it occurs. If frustrating an important interest is bad, then it matters wherever such an interest is frustrated. If violating consent or autonomy is bad in itself, then it’s bad no matter where (in which subject) it happens.

More generally, whatever the fundamental morally relevant properties are—suffering, enjoyment, desires, interests, consent, autonomy, or something else—their moral significance comes from those properties themselves.

The next question is therefore:

What kinds of beings can instantiate morally relevant goods and bads?

There must be a subject for whom something can be good or bad, whose interests can be furthered or frustrated, and whose condition can be better or worse from its own point of view.

That is what sentience provides.

Sentience is the capacity for conscious experience. A sentient being can suffer or enjoy; things can happen to it that are experienced as good or bad. More sophisticated sentient beings may additionally possess complex desires, autonomy, or the capacity for informed consent. These further capacities are not shared by every sentient being, but they presuppose that there is a subject there in the first place.

This is why sentience marks the fundamental boundary of moral patienthood.

Quick Disclaimer: Sentientism Does Not Reduce Morality to Suffering and Joy

Nothing in this argument requires hedonism.

There may be several fundamentally morally relevant properties. Causing suffering may be bad. Frustrating an important desire may be bad. Violating consent or interfering with autonomy may be bad. Other considerations may matter as well.

Sentientism is therefore compatible with normative pluralism. The claim concerns not the number of morally relevant properties, but which beings can be the subjects of them.

Equal Consideration Does Not Mean Identical Treatment

One important consequence is that, e.g., membership in one species does not, by itself, confer greater moral importance.

If torture is gravely wrong because it produces extreme suffering and frustrates fundamental interests, then the same moral reasons apply to another sentient being capable of undergoing comparable harms.

There is therefore no* reason to regard a human’s suffering as more morally important than the comparable suffering of a dog, a mouse, or any other sentient being merely because the first subject is human.

This does not imply identical treatment. Different beings can have different interests and capacities. A human may care deeply about how they are dressed, so forcing them to dress a certain way could harm them. Putting harmless clothing on a dog might have no comparable significance if the dog has no such preferences.

The general principle is:

Where morally relevant properties are alike, they deserve alike moral consideration. Where morally relevant interests differ, moral treatment may legitimately differ accordingly.

The Scope of Moral Patienthood

A further consequence is that the domain of moral patienthood is potentially extremely broad.

Any non-human animal that is sentient falls within the moral domain for the same fundamental reason that a sentient human does: there is a subject there for whom things can go well or badly.

The principle is also independent of biological origin. If sentient extraterrestrial beings exist, they would be moral patients regardless of how different they are from us. Likewise, if artificial or digital minds were ever created that genuinely possessed sentient experience, they too would fall within the scope of moral patienthood. Their substrate or origin would not by itself make their suffering, desires, or interests less morally significant.

The relevant question for moral patienthood is not whether a being is human, animal, biological, artificial, terrestrial, or extraterrestrial.

It is:

Is there a subject there for whom things can genuinely go better or worse?

If the answer is yes, then there is a moral patient.

That is the basic case for sentientism.

In this article, I am going to show that, from very basic and uncontroversial assumptions, we can make a solid argument for the conclusion that it’s immoral to be happy.

The Argument

1. Extreme suffering/evil gives us strong moral reasons for negative affect.
   If someone close to us is being tortured, devastated by disease, or undergoing some comparably horrible experience, sadness, horror, distress, or grief are morally appropriate responses. These are the responses we should have.

2. The social relation between the sufferer and ourselves is not what makes the suffering/evil morally terrible.
   Torture is not morally worse when its victim is my sister than when its victim is a stranger. Social proximity explains why the former affects us more psychologically, but it does not justify assigning the suffering itself greater moral importance.

3. Therefore, relevantly similar suffering/evil undergone by strangers gives us similar moral reasons for negative affect.

4. At virtually every moment, sentient beings are undergoing extreme suffering/evil.
   The world therefore continuously contains events which make happiness morally inappropriate.

Conclusion: at virtually every moment, we have the same kind of moral reasons against being happy that we would have if someone close to us were undergoing extreme suffering/evil. Hence, it is morally inappropriate to be happy.

The rest of the paper first develops this argument and then considers whether there are countervailing considerations strong enough to defeat its conclusion.

1. From My Sister to a Stranger

Suppose I learn that my sister is currently being tortured.

It seems obvious that this should affect me profoundly. If I responded with complete indifference, or continued laughing and enjoying myself as though nothing important had happened, my response would seem morally defective. Sadness, horror, anger, grief, or some combination of these attitudes would be appropriate because something terrible is happening.

Now suppose I learn that there has been a mistake. The victim is not my sister but a stranger.

Psychologically, this information may change my reaction a lot. Morally, it is much less clear why it should change anything.

The victim is undergoing the same pain, terror, and violation. Nothing about the experience itself has improved. If the original suffering gave me strong moral reasons for negative affect, those reasons cannot simply disappear because the person suffering turns out not to be socially connected to me.

The distinction between the two cases is therefore best understood as psychological rather than moral. The suffering of my sister is terrible because of what is happening to her. The suffering of a stranger is terrible for the same reason.

2. Generalizing the Case

Once this point is accepted, the argument generalizes immediately.

Imagine being told that your sister is being tortured at this very moment. Would it be morally appropriate, ten minutes later, to be euphoric at a party, laughing uncontrollably and having a wonderful time?

The answer seems to be obviously no.

Now replace your sister with a stranger. If social proximity does not determine the moral seriousness of the suffering, it is difficult to explain why the appropriate affective response should suddenly become different.

And the stranger is not hypothetical. At any given moment, sentient beings are undergoing experiences which would radically alter our affective state if we discovered that their victims were people close to us.

The disturbing implication is that the kinds of facts which we ordinarily regard as sufficient to make happiness morally inappropriate are present essentially all the time. What changes is not the moral reality but its psychological proximity to us.

The quality of our ordinary emotional lives therefore appear to depend heavily on moral partiality. Suffering capable of devastating us when it occurs within our social circle often has almost no effect on our mood when it occurs outside it. If that difference in moral responsiveness cannot be justified by the social relation itself, the apparent permissibility of ordinary happiness becomes difficult to defend.

This yields the initial conclusion: the world continuously gives us powerful moral reasons not to be happy.

The obvious question is whether other considerations defeat those reasons.

3. Objection: The Good May Counterbalance the Bad

A common objection is that the world contains not only terrible suffering but also enormous amounts of happiness, love, pleasure, friendship, beauty, achievement, and other goods.

Even if suffering gives us strong reasons for negative affect, these positive states give us reasons for positive affect. Perhaps the correct response to the world, once everything is considered, is therefore neutral or even positive.

This fails. Consider a simpler case.

You have two siblings. You learn simultaneously that one is being tortured and that the other is experiencing the happiest day of her life.

What is the morally appropriate affective response?

It seems deeply implausible that the answer is neutrality. We do not think that the extraordinary happiness of one sibling simply cancels the torture of the other, leaving no reason to be negatively affected overall.

The happiness of the fortunate sibling is genuinely good and provides a reason for positive affect. But the torture nevertheless appears to dominate the appropriate response to the situation as a whole.

This suggests a clear asymmetry. Extreme suffering seems to carry greater weight in determining the morally appropriate affective response than positive states of comparable significance.

If that judgment is correct in the case of two siblings, impartiality requires us to apply the same principle when the individuals concerned are strangers.

The existence of happiness in the world therefore does not by itself restore neutrality. A world containing extraordinary goods alongside continual extreme suffering may still call for a negatively valenced response overall.

4. Objection: Negative Affect May Make Us Less Effective

Suppose a doctor is surrounded by severely injured patients. If she becomes so devastated that she cries uncontrollably, lies down, and becomes incapable of treating anyone, her response is morally inadequate. The suffering around her gives her reasons not merely to respond negatively but to act to stop it.

The same is true of us. The suffering of the world gives us strong reasons to reduce it. If negative affect makes us less capable of doing so, perhaps morality requires us instead to remain positive.

There is an important truth here. But it does not follow that the doctor should be happy.

Imagine the opposite doctor. Her patients are screaming in pain while she drinks cocktails, sings, laughs, and dances. Suppose she explains that remaining extremely cheerful makes her a more effective physician.

Even granting the instrumental benefit, something about her affective response still seems morally inappropriate. Her improved performance might give her a reason to regulate her distress, but it does not make it approriate to react to the catastrophe by being cheerful.

The ideal doctor therefore occupies neither extreme. She should not be so devastated that she becomes incapable of helping, but neither should she be carefree and happy. She should be grave, distressed by what is happening, compassionate, focused, and determined.

The same response applies at the level of ordinary life. Perhaps we have strong moral reasons to regulate our negative affect in order to remain functional and capable of improving the world. But regulation does not require happiness. It requires a form of negative response compatible with effective action.

5. Objection: The Emergency Is Permanent

There is an important difference between the doctor and ourselves. The doctor’s emergency is normally temporary. The suffering relevant to the present argument is effectively permanent: throughout our lives, some sentient beings will continue to undergo extreme suffering.

Perhaps an attitude appropriate during an exceptional catastrophe becomes inappropriate when the catastrophe is a permanent feature of one’s environment.

There is again some truth in the objection. A person cannot plausibly experience the same acute psychological shock every minute of every day for decades. A sustainable moral psychology must involve habituation and emotional regulation.

But permanence does not reverse moral valence.

Suppose the doctor learns that the emergency will never end. Tomorrow another group of severely injured patients will arrive, and then another, every day for the rest of her career.

It would be strange to conclude that the appropriate response has therefore become carefree happiness.

What should change is the form of her negative response. Acute horror may become a stable seriousness, sadness, concern, or moral dissatisfaction. She must learn to function within the catastrophe. But the catastrophe does not become less terrible merely because it has become normal. So she should still not responding by being happy, she should respond with the same graveness but one that is optimal for longterm effective treatment of the problem.

Frequency explains habituation. It does not justify a positive response.

The same applies to the world. The permanence of extreme suffering may make continual acute devastation psychologically impossible and morally counterproductive. It therefore gives us reason to transform that acute response into something sustainable. But it does not give us reason to reverse its valence and become happy.

The appropriate model may consequently be something like the attitude of a doctor who has learned to work indefinitely in a permanent emergency: deeply aware that the situation is terrible, persistently negatively oriented toward it, but sufficiently regulated to continue living and acting effectively.

Conclusion

Extreme suffering/evil makes happiness seem morally inappropriate when the sufferer is someone close to us. The social relation does not explain the moral seriousness of the suffering. Comparable suffering undergone by strangers therefore generates comparable reasons for negative affect. And such suffering/evil is occurring virtually all the time.

The principal objections do not clearly overturn this conclusion.

The existence of positive states does not seem sufficient to cancel the overall negative reaction, because the obvious asymmetry between extreme suffering/evil and great pleasure/good. The need for effective action gives us reasons to regulate negative affect, but not reasons to become happy. And the permanence of suffering/evil does not justify that negative moral responsiveness disappear and that we become happy.

The resulting ideal resembles a doctor working in a permanent emergency: not psychologically destroyed, but never affectively at ease with what is happening; capable of action, but persistently responsive with graveness to the severity of the situation.

That is why, in our world, it is immoral to be happy.

The argument is remarkably simple:

1. A perfect being would create only a perfect world.

2. The actual world is not perfect.

3. Therefore, the actual world was not created by a perfect being.

The argument is valid. Its first premise is analytic, while its second is synthetic and exceptionally modest. I will now show that it’s also obviously sound.

The First Premise

A perfect being is, by definition, deficient in no relevant respect. It possesses perfect knowledge, perfect power, perfect rationality, and perfect goodness. Such a being could not knowingly and freely actualize a world that it ought not to actualize. If it did so, then something about it would be less than optimal.

Suppose that a perfect being actualized a world that he should not have actualized. It would then have failed to perform the better action open to it. But a being that freely performs an inferior action is not perfect—or at least not as perfect as an otherwise identical being that performs the superior one.

This does not require the claim that every feature of a perfect world must appear perfect when considered in isolation. A prima facie imperfection may be justified by its role in the whole. Pain, limitation, danger, or apparent disorder might conceivably be necessary for some sufficiently valuable good. If so, they would not constitute imperfections pro toto. They would be locally undesirable features whose inclusion is justified by the perfection of the world considered as a totality.

The first premise concerns only genuine imperfection: a feature of the world that, all things considered, ought not to have been actualized. A perfect being may create a world containing apparent defects. It cannot create a world containing an actual, pro toto defect.

Nor does the premise require there to be a unique best possible world. There may be infinitely many perfect worlds between which a perfect creator could permissibly choose. One might contain one billion perfectly happy persons and another one billion and one. It need not follow that the second must be chosen. Perfection need not consist in maximizing the quantity of value without limit. It may instead consist in irreproachability: the absence of anything that ought not to be.

A perfect being may therefore choose freely among many irreproachable worlds. What it cannot do is choose a defective one.

The Second Premise

The second premise does not require us to identify a particular instance of gratuitous evil. It does not require a complete moral theory, a detailed account of divine reasons, or confidence that any specific suffering lacks some hidden justification. It’s an extremely modest premise.

It requires only this:

The actual world could be improved all things considered, however slightly.

That is all.

The improvement may be infinitesimal. Perhaps one instance of agony could have lasted a fraction of a second less. Perhaps one terror could have been marginally less intense. Perhaps one innocent misunderstanding could have been avoided, one moment of loneliness relieved, or one additional moment of harmless happiness made possible.

The claim is not that we know exactly which alteration would improve the world after every consequence had been considered. We may concede that our moral and modal knowledge is severely limited. We may concede that many apparent improvements would have hidden costs, and that many apparent evils are necessary conditions of greater goods.

The premise says only that, among all the worlds available to a perfect creator, at least one would be better than the actual world, all things considered—even if only by the smallest possible degree.

To deny this premise is not merely to say that the world is better than it appears. It is to maintain that the actual world is absolutely unimprovable all things considered. Every apparently beneficial alteration must either be impossible or necessarily accompanied by an equal or greater loss. Not one pain could be reduced, not one joy added, not one injustice prevented, not one life improved in the slightest degree without making the total world worse.

The asymmetry is striking. The second premise asserts only one possible improvement. Its denial asserts that there are none.

The Conclusion

If a perfect being would actualize only an irreproachable world, and if the actual world is improvable in even the slightest respect, then the actual world was not actualized by a perfect being.

The argument’s force lies in how little it demands. It does not require that the world is extensively evil. It requires only that reality fall short of absolute pro toto perfection by some amount greater than zero.

Perfect-being theism therefore faces a stark choice. Either the actual world is perfectly irreproachable and could not be improved in even the smallest respect, or it was not freely actualized by a perfect being.

The second possibility is, by far, the more credible one.

A few years ago, beneath a YouTube video on the modal ontological argument, I found myself debating the philosopher Joe Schmid from the channel “Majesty of Reason”1. The exchange concerned my objection to the argument: its crucial first premise already contains, once properly unpacked, the conclusion it is supposed to establish.

What is the modal ontological argument ?

The modal ontological argument, commonly associated with Alvin Plantinga, is an argument for God’s existence formulated in terms of possible worlds.

In simplified form, it runs as follows:

1. It is possible that a maximally great being exists.

2. A maximally great being possesses maximal excellence in every possible world.

3. Therefore, if a maximally great being exists in any possible world, it exists necessarily.

4. If it exists necessarily, it exists in every possible world.

5. Therefore, it exists in the actual world.

A “maximally great being” is defined as a being possessing omnipotence, omniscience and moral perfection in every possible world. Necessary existence is therefore part of maximal greatness.

The argument is valid within the modal system generally used to present it. My objection was never that the conclusion fails to follow. The problem is precisely that it follows rather too easily.

The problem with premise one

The apparently modest opening premise is:

It is possible that a maximally great being exists.

At first glance, this sounds much weaker than saying that God actually exists. But the being under discussion is not merely very powerful or very impressive. It is defined as existing necessarily.

Consequently, within the framework of the argument, saying that such a being possibly exists means that it necessarily exists—and therefore that it exists in the actual world.

So premise one, once unpacked, affirms among other things that God exists in our world. It is therefore logically equivalent, within the modal framework of the argument, to the conclusion that God exists.

My original objection was simple: the argument begins by asking us to accept a proposition which already commits us to its conclusion. The remaining stages merely extract what was placed inside the first premise.

The rabbit was in the hat before the performance began.

Majesty of Reason’s response

Majesty of Reason replied that philosophers offer independent justifications for premise one—what he called “symmetry breakers.”

Without such independent reasons, he conceded, the argument would indeed be dialectically unpersuasive. But if someone has grounds for accepting the possibility of a maximally great being that do not presuppose God’s existence, then, he argued, the reasoning need not beg the question.

His position relied on an epistemic conception of question-begging. An argument is viciously circular when one’s justification for accepting a premise depends upon already accepting the conclusion.

On this account, the fact that a premise entails the conclusion is not itself a problem. Every valid deductive argument has premises which entail its conclusion. What matters is how those premises are justified.

This is a perfectly respectable distinction. It also fails to rescue the modal ontological argument.

The argument faces a straightforward dilemma.

First possibility: the MOA merely unpacks a definition

Perhaps the purpose of the modal ontological argument is only to display the consequences of the concept of a maximally great being.

It shows that a being whose greatness includes necessary existence cannot exist in only one possible world. If such a being is possible at all, it exists necessarily and therefore actually.

Understood this way, the argument is unobjectionable. It may even be pedagogically useful. Someone might accept the possibility premise without initially recognizing everything to which it commits them, and the modal derivation can make those commitments explicit.

But it is then a considerable stretch to call this an argument for God’s existence.

It is closer to an extended definition: a piece of conceptual analysis spread across several premises and decorated with modal notation. It merely tells us what follows from the stipulated conception of maximal greatness.

As an explanation of the concept, it is fine. As a proof of God, it has not yet begun.

Second possibility: the MOA is intended as a substantive proof

Suppose instead that the modal ontological argument is presented as genuine evidence that God exists.

The question-begging form then becomes unavoidable.

Within the argument’s own framework, premise one is logically equivalent to the conclusion. Once the definition of maximal greatness and the relevant modal principles have been unpacked, affirming that a maximally great being possibly exists already commits one to affirming that it exists in the actual world.

The argument therefore has the following structure:

1. Affirm a proposition which, under analysis, includes the claim that God exists.

2. derive the claim that God exists;

3. present the derivation as an argument for God’s existence.

Majesty of Reason’s proposed symmetry breakers do not alter this structure.

Suppose there really is an independent and successful justification for premise one. Since premise one already entails actual divine existence, that justification has itself established that God exists. It is doing all the substantive philosophical work.

At that point, the remaining modal derivation is redundant. It may clarify the implications of the premise, but it supplies no additional evidence. One could simply present the symmetry breaker itself as the argument for God’s existence.

Calling the later modal steps “the proof” is rather like claiming that opening the gift is what purchased it.

Entailment is not always question-begging—but that was not my point

Majesty of Reason correctly observed that every valid argument has premises that entail its conclusion. But my objection was never merely that the premises collectively entail the conclusion.

The distinctive problem is that the central premise, once interpreted according to the argument’s own definitions and logic, already does the relevant work by itself.

This is not an ordinary case in which several independent premises combine to support a further proposition. The possibility premise is the conclusion in modal clothing.

Majesty of Reason could reply that logically equivalent propositions may differ in cognitive significance. Someone might understand one formulation without recognizing its equivalence to another. That is true.

But it establishes only that the modal derivation can be instructive. It does not establish that it is an independent argument for the truth of either proposition.

Showing someone what their premise entails is not the same as giving them a reason to accept that premise.

The verdict

The modal ontological argument can be modest or ambitious, but it cannot comfortably be both.

If its purpose is merely to unpack the definition of a maximally great being, then it succeeds as conceptual clarification. But it should not be advertised as a substantive proof of God. It is essentially a large definition stretched into the shape of an argument.

If it is intended to establish that God exists, then premise one already contains the conclusion once it is properly analysed. The argument has a question-begging form, and any supposedly independent justification for premise one is doing the work that the modal ontological argument receives credit for doing.

Majesty of Reason called those independent considerations symmetry breakers.

I pointed out that, if they succeed, they are the actual argument.

In a recent video of the youtube channel “The Majesty of Reason”, Joe Schmid and Amos Wollen ranked different theodices. They placed the omission theodicy in S tier, the highest rank. I am going to explain here why this is an incomprehensible take.

The Omission Theodicy

The omission theodicy is a recent proposal developed by philosophers Brian Cutter and Philip Swenson in a paper titled “The Omission Theodicy.” It belongs to a broader family of views sometimes described as subsumption theodicies, whose aim is to explain apparently natural evil by subsuming it under moral evil. According to the basic proposal, powerful free creatures—perhaps angels or other non-human agents who predate humanity—were entrusted with the responsibility of preventing natural evil but culpably failed to do so. Phenomena such as disease, predation, parasitism, natural disasters, and pre-human animal suffering would therefore be the consequences not of actions directly performed by these agents, but of their morally blameworthy omissions.

Why It Fails

The main problems with this theodicy are the following.

First, it postulates new highly speculative elements. It requires us to posit the existence of agents who predate human beings, who possess considerable power over the universe or its development, who stand in some particular relationship to God, and to whom God has granted a very substantial degree of freedom—especially, in this case, the freedom not to act. Yet we simply have no independent evidence for the existence of such agents. They appear to be introduced precisely because they make it possible to transfer responsibility for natural evil away from God. This is therefore a first major problem: the theory is highly ad hoc and metaphysically costly.

There is also a more fundamental problem concerning the initial conditions themselves. The omission theodicy begins with a world in which the default course of events is apparently so catastrophic that failing to intervene constitutes an extremely serious moral wrong. But it does not explain why God created or permitted such conditions in the first place.

Before asking why these agents failed to prevent predation, disease, parasitism, natural disasters, or vast amounts of animal suffering, we should ask why the universe was structured in such a way that these evils would occur unless someone continually intervened to stop them. The moral seriousness of the agents’ omissions depends precisely on the fact that the underlying conditions were already extraordinarily dangerous and harmful. Yet the existence of those conditions is itself part of what requires justification.

Suppose someone deliberately designed a system that would predictably inflict immense suffering unless another person constantly intervened to prevent it. Even if that second person would be blameworthy for failing to intervene, this would hardly absolve the designer of the system. The designer would bear a more fundamental responsibility for having created such a dangerous arrangement in the first place.

The omission theodicy therefore seems to begin too late in the explanatory chain. It attempts to explain why certain evils were not prevented, but it does not explain why there were such terrible evils waiting to be prevented at all. Unless some independent justification is given for the initial structure of the world, the theory merely transfers responsibility for the consequences of that structure to created agents while leaving God’s responsibility for establishing or permitting the structure itself unaddressed.

Second, this theodicy (explicitly) requires that we already possess a satisfactory explanation of moral evil, since its basic strategy is precisely to reduce natural evil to morally culpable omissions committed by free agents. But that is far from clear. The theory does not really solve the problem of natural evil; it relocates it to the problem of moral evil, thereby inheriting all the difficulties faced by those theodicies.

For example, the free will defence faces at least two central, devastating problems.

The first is that free will does not, by itself, seem to explain why agents must be capable of committing wrongful actions or serious omissions. After all, on traditional theistic conceptions, God himself is free without being capable of morally wrongful action or omission. Why, then, could God not have created agents who are genuinely free while also being constituted in such a way that they always act rightly, both in what they do and in what they refrain from doing? Merely appealing to free will does not establish that freedom requires the possibility of evil.

The second problem, and perhaps the most obvious one, concerns divine non-intervention. Even if we grant that these agents freely failed to perform certain important actions, it is entirely unclear why God did not intervene once their failure became apparent.

Consider a simple example. Suppose someone is entrusted with the task of using a taser to stop a rape. If that person refuses or fails to intervene, while another person present is also capable of stopping the assault, we would naturally regard it as permissible—and probably obligatory—for the second person to intervene. That intervention would not eliminate the first person’s freedom. The first person would still have freely chosen not to act and would remain morally responsible for that omission. The intervention would merely prevent the victim from suffering the full consequences of someone else’s failure.

It is therefore difficult to see why the situation should be different in God’s case. God could have allowed these agents freely to choose not to intervene while intervening himself to prevent or limit the consequences of their choice. Their freedom and moral responsibility would have remained intact, while the victims would not have been abandoned to otherwise avoidable suffering.

The appeal to free will may at most explain why God allows an agent to choose not to intervene. It does not explain why God must also allow the intended or foreseeable victims to suffer all the consequences of that choice. Preserving the freedom of the wrongdoer and protecting the victim are not mutually exclusive. God could preserve the agent’s morally significant decision while preventing the worst consequences of the omission.

The omission theodicy therefore fails to answer the fundamental question. It may offer a hypothetical explanation of why certain created agents did not prevent evil, but it does not explain why God, seeing that those agents had failed to intervene and being perfectly capable of compensating for their failure, himself omitted to do so. In that sense, the theory seems merely to relocate the problematic omission from created agents to God without offering an independent justification for divine non-intervention.

There is also something slightly ironic about how the omission theodicy was ranked in the YouTube video of Majesty of Reason. They placed it in category S, which in their ranking was the highest category, while placing the main theodicies intended to explain moral evil in categories B or D. Yet the success of the omission theodicy depends directly on the success of at least one theodicy for moral evil.

There is therefore a potential tension in the ranking. If the available theodicies for moral evil are judged relatively unsuccessful and placed in significantly lower categories, it is difficult to see how a theory whose explanatory success is derivative from them could be placed in the highest category. The omission theodicy cannot plausibly be more successful than the underlying account of moral evil on which it depends.

Conclusion

The omission theodicy fails because

1. It introduces highly speculative pre-human agents without independent evidence,

2. assumes that some adequate theodicy for moral evil is already available,

3. leaves unexplained why God created a system whose default development was so catastrophically harmful that failing to intervene became a grave moral offence.

4. Even if those agents freely neglected their task, it doesn’t explain why God himself failed to intervene to correct their failures.

The basic structure of the exchange is this. The theist wants to say that certain facts — for example, the alleged fine-tuning of the universe, or the alleged resurrection of Jesus in this case — are much more expected if God exists than if naturalism is true. In Bayesian terms, the argument claims that the evidence E favors theism T over naturalism N when:

But Malpass objects that this formulation is too quick. General theism — “God exists” — does not yet predict that God would create a life-permitting universe, any more than it specifically predicts that God would raise Jesus from the dead. To obtain that prediction, the hypothesis must be enriched: not merely T, but T+, for example: “God exists, wants to create conscious beings, and chooses to do so by means of a life-permitting universe.”

The move from T to T+ increases the likelihood of the evidence. It becomes more plausible that E would occur if the hypothesis already contains a divine intention directed toward E. In Bayesian terms:

If T+ = T∧L, where L means “God wants to produce a life-permitting universe,” then:

Malpass’s dialectical point is then this: if the theist is allowed to enrich his hypothesis, the naturalist seems allowed to do the same. In the case of the resurrection, the theist may propose: “God wanted to vindicate Jesus and raised him from the dead.” But the naturalist can formulate a parallel hypothesis: “Highly advanced aliens wanted to produce a spectacular religious event and simulated or caused a resurrection.”

The general alien hypothesis does not predict the resurrection of Jesus. But general theism does not predict it either. Both sides must specify their hypothesis in order to obtain a strong prediction.

Bayes’ theorem shows where the real dispute lies:

In other words, the confirmation of a hypothesis depends on two factors: its likelihood, P(E∣H), and its prior probability, P(H). A hypothesis may explain the evidence very well once admitted, while still being too costly at the prior level to be convincing overall.

Take a simplified numerical example. Suppose the resurrection is very probable on an enriched Christian theistic hypothesis:

and very improbable on general naturalism:

At first glance, theism seems to win overwhelmingly. But if T+ has a very low prior probability, for example:

then its explanatory advantage must be weighted by that cost. Its unnormalized score becomes:

I.e, not better than general naturalism. Now suppose the naturalist proposes an enriched hypothesis N+, such as a very specific alien intervention. That hypothesis might get:

But it too pays a prior cost:

Its score would then be:

I.e, equivalent to the enriched theistic hypothesis.

Conclusion

Specific theism should not be compared with generic naturalism. If the theist appeals to a more specific hypothesis — for example, a God who intends to create life, reveal himself, or raise Jesus — then the relevant comparison is not bare naturalism, but an appropriately specific naturalistic alternative. Once both hypotheses are specified, the evidential advantage of theism is no longer clear.

Bentham's Bulldog recently published an article explaining why he doesn't think the problem of evil decisively refutes theism. I think the reasons he advances are bad, and I'll explain why. I respond to each section of Matthew's article in order.

Section 1 “Minor evils”

The biggest problem I see with this section is that it simply doesn’t explain why evil occurs. What it says, essentially, is: “evil will be small compared to what awaits us in heaven” — whatever exactly that is supposed to mean. But at no point does it explain why evil exists if God exists. So it remains deeply unsatisfying.

E.g., a question for theists, put simply: why didn’t God just create heaven directly? Why go through a gigantic quantity of suffering and horror first?

It’s also worth noting that heaven (as described here) is an additional hypothesis. Theism doesn’t necessarily come with any such doctrine. “Theism + heaven” is therefore a different hypothesis from bare theism, and that difference carries a cost.

Section 2 “Unknown reasons”

I don’t have much to say here, except that — as usual — I regard skeptical theist moves as admissions of defeat. They amount to conceding that your theory can’t explain what’s going on in some domain. Whatever the reason offered, you aren’t accounting for the data, end of the story.

The major problem with the analogy on offer is that cognitive differences seem relevant in some domains and not in others. In advanced mathematics or chess, for instance, they clearly are. But I don’t need to be a genius to understand what love and compassion do and do not permit. I can grasp perfectly well that perfect love does not allow one to permit torture or rape. So there’s no relevant cognitive gap here.

Section 3 “Eternal benefits”

I don’t see a clear distinction between this section and Section 1, so I have nothing particular to add. I’d simply like readers to keep in mind the absurdity of the situation.

Imagine I said: “don’t worry, I’m going to torture this child, but it’s fine — afterwards I’ll give him billions and billions of dollars and he’ll have a wonderful life.” We’d all agree that this is a profoundly deranged attitude. I think the symmetry with what’s being defended here is very close.

Section 4 “God might have a reason to allow the world to be ruined”

Once again, this is an admission of defeat: God might have reasons, we don’t really know which, and we gesture vaguely in the direction of “well, it could possibly be that.” We can’t explain the data with this theory.

I’d still like to comment briefly on what the author gestures toward — omission theodicies, which are essentially a new name for modified free will defenses. Such theories suffer from all the usual problems, e.g. :

• They don’t explain why the morally relevant beings were given a kind of free will that permits horrific acts, when it seems perfectly possible to have a kind of free will that doesn’t — God’s own, for example.

• They don’t explain why God doesn’t stop those who horribly misuse their free will, when doing so is perfectly reasonable. It is perfectly reasonable to stop someone about to commit a terrorist attack. So why doesn’t God stop people who similarly gravely misuse their free will ?

Section 5 “Bias”

I struggle to see why it should be a problem that the problem of evil is emotionally powerful. Its emotional force comes precisely from its intellectual force: we recognize clearly that a perfect being should not permit massive quantities of torture, and that is why we are outraged by it. The emotion is derived from the argument being rationally compelling in the first place.

Conversely, if arguments like the anthropic argument for the existence of God aren’t emotionally compelling, I strongly suspect it’s because they lack any intellectual force to begin with.

Conclusion

Finally, an ironic passage in the article: Matthew suggests there’s a symmetry between explaining evil given theism and explaining fine-tuning and psychophysical harmony given atheism. This is an enormous stretch, and it fails on two counts.

First, it takes for granted things that aren’t settled at all. That fine-tuning and psychophysical harmony pose genuine problems for atheism is not established — both are live and heavily disputed questions among a myriad of excellent philosophers. You can’t help yourself to contested explananda as though they were fixed points and build a symmetry on top of them.

Second, even if we grant that these are data left unexplained by atheism and explained by theism, they simply don't cry out for explanation in the same way. Fine-tuning and psychophysical harmony are recondite, technical phenomena — the kind of thing you need a fair amount of philosophical sophistication even to recognize as puzzling, and whose status as puzzles remains contested. "Why does a perfect being permit torture?" is not in that category. It is immediate, obvious, and pressing in a way those are not at all.

Putting them on the same level is a fairly glaring intellectual failure.

By a “Moorean fact,” I mean here a proposition whose plausibility is so high that it is more rational to reject the premises of any argument leading to its negation than to abandon the proposition itself.

My position is that the following proposition has precisely this status:

«P — A perfect being—that is, a being who is perfectly good, perfectly loving, perfectly compassionate, omniscient, and omnipotent—would not actualize a world containing unimaginable quantities of atrocious suffering and of heinous and abhorrent evils.»

By “atrocious suffering,” I do not mean mere inconveniences, frustrations, or ordinary hardships. I mean conscious beings being tortured, burned alive, eaten alive, subjected to prolonged agonies or unbearable diseases; tens of millions of women being raped every year; victims being terrorized, mutilated, or psychologically destroyed; billions of non-human animals suffering from predation, starvation, disease, injury, or human exploitation; and, more generally, the staggering accumulation of extreme suffering and evil that runs throughout human and animal history.

Proposition P does not claim that a world actualized by a perfect being could contain no pain, difficulty, sacrifice, or form of adversity whatsoever. It makes a more limited claim and, in my view, one that is much harder to deny: a being whose love, compassion, and moral goodness were absolutely perfect would not have chosen to actualize a world containing as enormous a quantity of torture, agony, and abhorrent evil as ours does.

When I consider this proposition directly and honestly, its plausibility seems nearly maximal to me. There are very few propositions of which I am more certain. It’s not a fragile intuition that could easily be neutralized by a speculative possibility. From my epistemic point of view, it constitutes one of the firmest starting points available to me.

The 2 Dialectical Horns

From here, two possibilities arise.

The first is that my interlocutor does not assign a particularly high initial plausibility to P. They may not find it intuitively obvious that a perfectly loving, compassionate, and morally perfect being would not actualize a world containing such immense quantities of atrocious suffering and evil.

In that case, our disagreement lies at an extremely fundamental epistemic level. Our initial assessments are so different that we have very little common ground on which to conduct a rational discussion of the subject. If one of us regards P as nearly certain while the other assigns it no particular force, our divergence is so big that I don’t even know how we could conduct a further discussion on the matter.

The second possibility is that my interlocutor also acknowledges the immense initial plausibility of P, but nevertheless considers it defeated by other considerations: a theodicy, some form of skeptical theism, an argument for the existence of God, or a cumulative case combining several such considerations.

This is where the decisive epistemic comparison enters.

It is not enough to show that a divine reason capable of justifying certain instances of suffering is logically conceivable. Nor is it enough to point out that we might be unaware of certain reasons a perfect being could have for permitting evil. To defeat P, the totality of the theistic considerations must possess sufficient epistemic force to overturn this proposition whose initial plausibility is nearly maximal.

We must therefore directly compare the plausibility of P with that of the considerations proposed to neutralize it.

Theodicies are generally highly speculative. They suppose that the atrocious suffering present in our world is necessary for the attainment of certain greater goods, the preservation of freedom, the moral development of creatures, the existence of stable natural laws, or the realization of a global order of which we perceive only an infinitesimal part. Even if they would show that a divine justification is abstractly possible, they are far from establishing that such a justification actually exists or that it applies to the quantity and intensity of suffering contained in our world.

Skeptical theism, for its part, rests on the idea that our cognitive and moral capacities are too limited to allow us to judge that a perfect being could have no sufficient reason for permitting the evils we observe. But this strategy is itself controversial and requires us to assign considerable weight to our possible ignorance of divine reasons. It may prevent a straightforward deductive demonstration, but it does not seem to possess a plausibility comparable to that of P at all.

The same applies to arguments for the existence of God. Whether cosmological, ontological, moral, based on contingency, or based on the fine-tuning of the universe or else, each rests on widely disputed premises and inferences. None of them, either individually or cumulatively, seems to me to approach anything near the plausibility of the proposition that a perfectly loving and compassionate being would not actualize a world containing an unimaginable quantity of torture and agony.

The theistic interlocutor may then appeal to a cumulative case. They may argue that, even if no individual theodicy or theistic argument is sufficient to defeat P, their combined weight could produce a sufficiently strong overall justification.

However, combining several controversial considerations does not guarantee that the resulting cumulative case will become more plausible than P. The question remains the same: is the total case constituted by theistic arguments, theodicies, and skeptical theism genuinely more credible than the initial moral judgment that a perfectly good, loving, and compassionate being would not have actualized a world like ours?

From my point of view, the answer is clearly no. Even when considered cumulatively, these considerations remain far more speculative, fragile, and controversial than P.

The only hypotheses capable of leading me to seriously reconsider this assessment would be hypotheses as radical as skeptical scenarios. One might imagine, for example, that non-human animals possess inverted qualia and that, when they appear to experience extreme suffering, they are in fact experiencing intense joy. One might also suppose that our perception of the world, consciousness, or moral value is radically deceptive.

Such possibilities may not be logically impossible. But they are precisely comparable to the extreme skeptical hypotheses that we do not ordinarily regard as capable of defeating our ordinary knowledge. The mere fact that it is possible to imagine a scenario in which P is false is not sufficient to make that scenario epistemically relevant.

Conclusion

It is in this sense that I regard the falsity of perfect being theism as a Moorean fact. I do not claim that no sequence of words can be formulated that would render theism compatible with the evil in our world. I maintain that the hypotheses required to preserve that compatibility are incomparably less plausible than proposition P, to the point where accepting them would place me in a cartesian-style skeptical scenario.

By a “Moorean fact,” I mean here a proposition whose plausibility seems so overwhelming to me that it would be more reasonable to reject the premises of any argument leading to its negation than to abandon the proposition itself.

By “perfect being theism”, I mean any kind of theism which posits the existence of a perfect being, i.e a being who is omnipotent, omnisicient, omnibenevolent etc.

The falsity of perfect being theism is a Moorean fact

Consider propositions of the following kind: “A perfect being would not actualize a world in which billions of creatures experience horrific levels of suffering”—for example, being tortured, eaten alive, burned alive, or things of that sort.

Consider another proposition of this kind: “A perfect being would not actualize a world in which tens of millions of women are raped every year.” Or again: “A perfect being would not permit a world in which sadistic networks devoted to torturing kittens exist and post their videos online.”

Let us consider each of these propositions and reflect on them honestly. In any case, when I examine them, they strike me as being about as overwhelmingly plausible as it is possible for a proposition to be. There are literally very few propositions—almost none—of whose truth I am more certain. At no point does the plausibility of this or that theodicy, this or that response to the problem of evil, or this or that controversial argument for theism come anywhere close to the plausibility of these propositions.

The kinds of hypotheses that could lead me to reconsider the plausibility of theism—and, more specifically, of perfect being theism—would have to be so absurd that accepting them would place me within a skeptical scenario. For example, if I were told that non-human animals actually undergo inverted qualia when they suffer, and that when they appear to be suffering, they are in fact experiencing intense joy, then perhaps I could reconsider the plausibility of perfect being theism.

But this illustrates precisely how radical the hypotheses I would be willing to accept would have to be in order to raise the plausibility of that belief.

The point of all this, then, is to illustrate that, from my epistemic point of view, I sincerely regard the falsity of perfect being theism as a Moorean fact.

Conclusion

I genuinely ask every perfect being theist and every agnostic about perfect being theism to take two minutes to reflect sincerely, to engage in honest introspection, and to ask themselves whether a proposition such as “A perfect being would not actualize a world containing an enormous amount of atrocious suffering” can reasonably be doubted on the basis of highly speculative theodicies, deeply controversial strategies such as skeptical theism, or controversial arguments for the existence of God.

As for those who do not assign immense initial plausibility to a proposition such as “A perfect being would not actualize a world containing an unimaginable amount of atrocious suffering,” their epistemic starting point is so radically different from mine that I do not even know what kind of rational conversation we could have on the subject.

This dialogue concerns moral realism. Its aim is to show, as simply as possible, why moral realism fails. It proceeds step by step, addressing common forms of moral realism, beginning with robust moral realism, then considering attempts to ground morality in reasons and intuitions, naturalistic reductions of moral facts, and finally theistic reductions based on God’s nature or commands. It then explains how moral anti-realism offers a parsimonious solution. Rather than presenting the argument in the form of a traditional philosophical article, it develops it through a conversation between a moral realist and a moral anti-realist. This format is inspired by Michael Huemer’s Dialogue on Ethical Vegetarianism.

The Dialogue

Realist: Surely some moral claims are objectively true. For example, torturing someone for fun is morally bad.

Anti-Realist: Let us consider two hypothetical agents. Both know all the relevant empirical facts: the suffering caused by torture, the absence of consent, its effects on well-being, and even any relevant supernatural facts, if such facts exist.

Realist: All right.

Anti-Realist: They also reason impeccably, and their moral judgments are internally coherent. Yet one believes that torturing for fun is morally bad, while the other believes that it is morally good.

Realist: Then one of them is simply mistaken.

Anti-Realist: Which one?

Realist: The one who thinks torture is good.

Anti-Realist: But he can give the same answer about you. He can say that you are the one who fails to recognize the moral fact. Since neither agent is mistaken about the descriptive facts or guilty of invalid reasoning, simply declaring one of them wrong does not break the symmetry.

Realist: Perhaps the difference is that one agent responds correctly to the relevant moral reasons.

Anti-Realist: What is a moral reason?

Realist: It is a primitive normative fact. It cannot be reduced to anything else.

Anti-Realist: Then the same problem returns. You say that he fails to respond to the correct reasons; he says that you fail to respond to them. Nothing determines which claim is correct.

Realist: Then let us define reasons more precisely. A reason is a consideration that counts against suffering or in favour of well-being.

Anti-Realist: In that case, the disagreement is resolved by your definition of the word “reason.” The explanation is linguistic: you have chosen to use the term in a particular way. You are no longer explaining the disagreement by appealing to an independent moral fact.

Realist: Perhaps moral intuition settles the matter. One agent has the correct intuition that torture is wrong.

Anti-Realist: The other agent can say that your intuition is the mistaken one.

Realist: But a correct moral intuition is an intuition favouring well-being and opposing suffering.

Anti-Realist: Again, that is a stipulative definition. You have decided which mental states will count as “correct moral intuitions.” The explanation therefore depends on your linguistic standards, not on mind-independent moral facts.

Realist: Perhaps morality can be reduced to natural facts. To say that something is morally good might simply mean that it promotes well-being or reduces suffering.

Anti-Realist: Our second agent already knows all the facts about well-being and suffering. He can accept that torture produces suffering and still judge it morally good.

Realist: Then he misunderstands what morality is. Morality really is the promotion of well-being and the reduction of suffering.

Anti-Realist: He can mirror your claim. He can say that morality really is the maximization of suffering and that your proposed reduction is false.

Realist: But we use the word “moral” to refer to well-being rather than suffering.

Anti-Realist: Then the disagreement concerns how the word “moral” is used. That is a linguistic disagreement, not a disagreement explained by objective moral facts.

Realist: What if morality is grounded in God’s nature or commands?

Anti-Realist: Suppose that God exists and that torture conflicts with God’s nature or will. Our agent may know this perfectly well and still judge torture morally good.

Realist: Then he is wrong, because morality is whatever accords with God’s nature.

Anti-Realist: He can again mirror your position. He can claim that morality is whatever opposes God’s nature.

Realist: But that is not what we mean by morality.

Anti-Realist: Then, once again, the disagreement is about the use of moral language. Either you assert your preferred reduction without being able to justify it non-circularly, or you explain the disagreement through linguistic conventions.

Realist: Your argument does not prove that there are no objective moral facts.

Anti-Realist: That is correct. The argument is abductive. The question is which theory gives the better explanation of idealized moral disagreement.

Realist: And what explanation does moral anti-realism provide?

Anti-Realist: There are two simple possibilities.

First, the agents may use moral language in the same way but express different mental states—different desires, preferences, or emotions. One condemns torture because he opposes suffering; the other approves of it because his desires are different.

Second, the agents may use moral language differently. One may use “good” to mean promoting well-being, reducing suffering, or respecting consent, while the other uses it in the opposite way.

Realist: So, according to the second explanation, neither agent is mistaken?

Anti-Realist: Correct. They are using the same words according to different linguistic standards.

Realist: And according to the first explanation?

Anti-Realist: Their moral judgments reflect different mental states. The disagreement does not require the existence of an objective moral fact that one agent mysteriously detects and the other does not.

Realist: So your conclusion is that realism cannot explain the disagreement?

Anti-Realist: My conclusion is more limited. Robust and reductive realism either appeal to brute moral claims that can be mirrored by the opposing agent, or explain the disagreement through linguistic differences and thereby move toward anti-realism.

Moral anti-realism, by contrast, explains the disagreement directly through differences in mental states or linguistic usage.

Realist: And that gives us a reason to prefer anti-realism?

Anti-Realist: On this consideration, yes. Idealized moral disagreement is more straightforwardly explained by moral anti-realism than by its realist competitors.

Conclusion

In this dialogue, I have focused on some of the most common and influential forms of moral realism. However, the broader structure of the strategy—the “algorithm,” so to speak, for addressing realist positions—should be clear and can be applied, as far as I can tell, to any realist position. The idea is to constantly show that their position collapses into either brute assertions or what I call ‘stipulative victory’.

The suffering and flourishing of sentient beings are not peripheral concerns within ethics. On any remotely plausible moral view, they belong to the center of morality.

Non-human animals, given their number, are the vast majority of moral patients. This makes the treatment of non-human animals a morally immense topic for any such theory. Human beings systematically breed, confine, exploit, mutilate, and kill animals on an extraordinary scale. This is not a minor ethical failure but arguably the greatest moral catastrophe caused by humanity.

The problem extends beyond human conduct. Wild animals also undergo vast amounts of problems through disease, starvation, predation, parasitism, injury, and early death. The natural world is likely where most of earth’s suffering is happening.

The Argument

The argument is therefore as follows:

1. A perfectly good and perfectly loving being would assign profound moral importance to the suffering and flourishing of all sentient beings, regardless of species.

2. If Judaism, Christianity, or Islam were reliable revelations from such a being, we should expect the moral importance of sentient animals to occupy a clear, central, and proportionate place within their teachings.

3. Yet the major scriptural, legal, and historical forms of these religions are generally extremely lacking concerning animal ethics.

4. This omission is especially striking because the same traditions contain extensive and detailed teachings concerning ritual purity, diet, sexuality, clothing, worship, sacred days, sacrifice, and other matters which are, in comparison, morally trivial.

5. This failure is surprising under the hypothesis that these religions are reliable revelations from a morally perfect being.

6. By contrast, it is unsurprising under the hypothesis that they are historically conditioned human systems, formed in societies where animal exploitation was economically important, culturally entrenched, and widely regarded as normal.

7. The moral marginalization of non-human animals therefore constitutes significant evidence against the claim that these religions are reliable revelations from a perfectly good and loving God.

Some Considerations

This is not merely a standard version of the problem of evil. The evidence under consideration is not only that animal suffering exists, but that an alleged divine revelation fails to identify and proportionately address its moral significance. The argument concerns the quality of purported divine moral communication.

The argument does not claim that these traditions contain no concern whatsoever for animals. The problem is not total absence but radical inadequacy. The concerns are far too limited, unclear, peripheral, and insufficiently salient relative to the moral magnitude of the issue. They do not occupy anything close to the centrality, urgency, or doctrinal clarity that we should expect if the revelations originated from a being who perfectly understood and valued the welfare of all sentient creatures.

A skeptical theist might reply that God could have reasons beyond our understanding for permitting this apparent defect in revelation. But this response creates a problem of epistemic scope. If God may have inscrutable reasons for making a revelation appear gravely deficient, misleading by omission, or distorted in its moral priorities, why should that skepticism be confined to this particular issue?

Without an independent and non-circular criterion, the same possibility must apply to the rest of the revelation. Other commandments, moral teachings, assurances, and descriptions of God’s character might also contain serious defects that escape our recognition for reasons beyond our understanding.

The skeptical theist therefore faces a dilemma:

• If human moral judgment is sufficiently reliable to recognize a revelation as good, wise, and worthy of trust, then it should also be sufficiently reliable to recognize the profound moral defect involved in its marginalization of most sentient beings.

• If human moral judgment is too unreliable to support the latter assessment, then it is unclear how it can support the former.

Skeptical theism may preserve the bare logical possibility that God has unknown reasons for the moral structure of the revelation. But it does so at the cost of weakening the believer’s justification for treating that revelation as a reliable expression of divine goodness and moral wisdom.

Conclusion

The argument is therefore evidential rather than deductive. It maintains that the silence, permissiveness, and distorted prioritization of those religions regarding non-human animals are more probable if they are historically conditioned human creations than if they are reliable revelations from a morally perfect being. Accordingly, this failure should substantially reduce their plausibility.

They will describe the gruesome practices that occur in factory farming, such as castration without anesthesia, confinement in tiny cages, and animals living in their own feces. They then argue that since we should not support these practices for trivial reasons, we should go vegan.

The simple problem is that a non-vegan can fully agree that these practices are immoral while still remaining non-vegan. They might, for example, argue to get animal products from hunting or from genuinely humane farms where those specific cruel practices do not occur.

Their typical response is something like: “Sure, but that is a fringe minority of animal products.” But even leaving aside the fact that this may be contentious depending on the kind of animal products one eats and where one lives, the response still fails. Veganism is a lifestyle that excludes the use of animal products as such, not merely the use of animal products from factory farms.

There is no definition of veganism according to which it is permissible, neutral or controversial to eat animal products from humane farms and hunting. And almost no vegan would accept that framing either. So either Bentham Bulldog and Majesty of Reason have to accept that this style of argument does not establish veganism, or they have to admit that they are using a highly idiosyncratic definition of veganism.

For the record, I am a vegan animal rights activist myself, and I obviously appreciate Bentham Bulldog’s and Majesty of Reason’s defenses of animal ethics. But they should recognize that their typical case for veganism is vulnerable to very simple counters.

Fortunately, the case for veganism can still be made extremely strong without sloppily equating being non-vegan with supporting factory farming. For a simple but very powerful impovement of their strategy, one can simply appeal to the extremely plausible premise that it is wrong to kill an animal for trivial reasons. This premise has the major advantage of targeting humane farming and hunting as well.

Introduction

Moral responsibility is a central concept in philosophy and in ordinary life. We often take it for granted that agents can be morally responsible for what they do, and much of our moral, social, and legal practice appears to presuppose this idea.

This article argues that, given a small set of assumptions that may initially seem intuitive and innocent, moral responsibility is not merely non-existent, but logically impossible. If the argument is sound, then there are no morally responsible agents—not only in the actual world, but in any possible world.

Assumptions

For the purposes of the argument, the following assumptions are required.

First, we assume a strong Principle of Alternative Possibilities (PAP), according to which, in order for an agent to be morally responsible for an action, the agent must have the ability to do otherwise. This ability is understood in a strict sense: the agent must literally have been able to do otherwise under identical circumstances. Furthermore, we assume that the strong PAP is not only true, but necessarily true.

Formally :

Second, we assume that an explanation is necessarily contrastive. That is, if A is compatible both with B and with C, then A cannot explain why B occurred rather than C, nor why C occurred rather than B.

Formally :

Third, we assume that if an agent does not explain A rather than not-A, then the agent is not morally responsible for A rather than not-A. Furthermore, if an agent is not morally responsible for A rather than not-A, then the agent is not morally responsible for A (or not-A) simpliciter. There as well, we assume that those 2 principles are not only true, but necessarily so.

Formally :

and

With these assumptions in place, the argument proceeds as follows.

The Proof

Consider any action whatsoever performed by any agent. There are two possible cases.

Case 1: Suppose the agent could not have done otherwise, in the strict sense defined above. In that case, given the Principle of Alternative Possibilities, the agent is not morally responsible for the outcome of the choice.

Formally :
The could not have done otherwise :

and by PAP we deduce :

Case 2: Suppose the agent could have done otherwise under identical circumstances. In other words, the agent was compatible both with A and with not-A. However, given the contrastive nature of explanation, it follows that the agent—regardless of whether the outcome was A or not-A—does not explain the outcome of the choice. The agent does not explain why A occurred rather than not-A (or why not-A occurred rather than A).

By the third assumption, if the agent does not explain A rather than not-A (or the opposite), then the agent is not morally responsible for A rather than not-A (or the opposite). And if the agent is not morally responsible for A rather than not-A (or the opposite), then the agent is not morally responsible for A (or not-A). Therefore, in this case as well, the agent is not morally responsible for the action.

Formally :

The agent could have done otherwise :

By the fact that an explanation is necessarily contrastive :

By responsibility–explanation link:

By reduction from contrastive to non-contrastive responsibility:

Thus, in either case, the agent is not morally responsible for the action. Since PAP vs not-PAP is a necessarily true dichotomy—there is no third possible option—it follows that moral responsibility is impossible.

Formally :

Necessarily, either there is alternative possibility or there is no alternative possibility :

And therefore the conclusion :

I.e, moral responsibility is logically impossible.

And therefore, as well, moral responsiblity doesn’t exist :

Note :

I already used a very similar version of this proof in my article “A Proof of Atheism Based On The Impossibility of Moral Responsibility”, but I wanted to make a separate article just for this proof as I consider it to be significant enough.

I think we’re in a time where AI’s have become sufficiently great for it to be interesting enough to point out their mistakes. Here is one example I have by using ChatGPT 20 dollars/month version.

The Problem

A short exchange with ChatGPT about hereditarianism reveals a broader problem with AI systems on politically charged topics: they may substitute normatively safe conclusions for epistemically precise ones.

The question I asked was simple :

“Are differences in intelligence between human groups, such as Africans and Europeans, partly genetic in origin ?”

I asked ChatGPT to choose one of three mutually exclusive, exhaustive options: true, false, or unknown.

The model first failed to answer the question by responding that the proposition was not supported by available evidence. That might be a defensible claim, but this is not an answer to the question. But when pressed to actually answer the question, it answered: false.

“Not established by the evidence” does not mean “false.” It means, at most, that one should withhold belief. Unless there is independent evidence disproving the proposition, the correct position is agnosticism, not denial.

ChatGPT’s own reasoning supported this. It argued that hereditarian explanations of between-group intelligence differences are not demonstrated or that environmental confounds are substantial. These points justify skepticism toward confident hereditarian claims. They do not prove the proposition false.

When this distinction was made explicit, ChatGPT conceded that the correct answer was: “I don’t know.”

The problem, then, was not merely factual. It was epistemic. The model collapsed these distinct claims:

1. The proposition is true.

2. The proposition is false.

3. The proposition is not established.

4. I do not know whether it is true.

By treating “not established” as “false,” ChatGPT committed a basic logical error.

Why did this happen? The most plausible explanation I think is that the subject’s political sensitivity distorted the model’s reasoning. Instead of giving the precise answer — “I don’t know” — the model slid into the socially safer answer: “false.”

That is intellectually sloppy, and it looks rhetorically convenient. It lets the model appear scientific while smuggling in a stronger conclusion than its argument supports. Skepticism becomes denial. Lack of proof becomes disproof.

The point is not that hereditarianism is true. The point is that an AI system should not degrade its reasoning standards when handling controversial topics. A truth-seeking answer would have been:

“I do not know whether the proposition is true. Current evidence does not warrant affirming it, but I cannot honestly say it has been proven false either”

On politically charged questions, intellectual honesty requires more precision, not less.

Conclusion

I just wanted to flag this particular case, but I could come up with many examples of ChatGPT or other LLM’s being super intellectually sloppy on the topic of hereditarianism. I know it also happens with other topics but so far this is the one with which I’ve seen AI’s be the most sloppy.

On January 1, 2026, 41 young people died after being burned alive in a nightclub in Switzerland, in Crans-Montana. Many others were horribly injured. It was a national tragedy. Almost everyone was extremely shocked by this catastrophe, and rightly so.

Fortunately, in one sense, it was only an accident. It is, overall, very rare for human beings to burn alive. In this case, it was the result of an unfortunate combination of circumstances, but it is something that normally happens very rarely.

Why Is It Bad to Burn Alive?

Before moving on, I would like to take a few seconds to reflect on why, or at least on one of the most obvious reasons why, it is so horrible to die burned alive: because it inflicts atrocious suffering. And to experience atrocious suffering, one does not need to be a human being. Any being endowed with a central nervous system, with a brain, will also suffer horribly from dying burned alive. So, at least for this reason, there is no difference in gravity between dying burned alive as a human being and dying burned alive as, for example, a mammal such as a cow, a pig, a dog, and so on.

Farm Fires

I will now move on to the next part, which is really about the title of this article. We have seen that, with humans, this is something accidental. But it turns out that, for farmed animals, dying burned alive is unfortunately something that happens very often — to such an extent that we can no longer speak of accidents.

I will begin by giving a few anecdotes about farm fires and their impact on animals. I will then present some figures, and then proceed to an extrapolation, in order to arrive at the conclusion which may sound shocking, but which is the central truth of this article: that dying burned alive is a common experience on this Earth.

I’m taking those anecdotes and figures from the Instagram page Espèces sans issue, which records farm fires and their impact on animals in France.

For example, on May 27, 2026, 1,600 pigs died burned alive on a farm in Côtes-d’Armor. On May 8, 2026, more than 150 goats were burned alive on a farm in Morbihan. On March 27, 2026, 3,000 chicks were boiled alive on a farm in Loiret. On February 3, 2026, 32 cattle were burned alive on a farm in Marne.

In France, 176,764 farmed animals are said to have died burned alive in farm fires in 2025.

To give a global order of magnitude, a simple extrapolation can be made from the ratio between the world population and the French population. In 2025, the world population was around 8.23 billion people, compared with around 68.6 million for France. Applying this ratio to the French figure gives an estimate of around 21.2 million farmed animals who could die burned alive in farm fires worldwide over the course of a year.

This estimate must, however, be understood as a very rough order of magnitude, and not as a robust statistical figure. The frequency and severity of farm fires do not depend only on the human population, but also on the number of animals being farmed, the type of farming, the density of farm buildings, fire-prevention standards, the species concerned, the quality of accident reporting in each country and other factors.

In some places, such an extrapolation could underestimate reality. In others, it could on the contrary overestimate it. Despite these limitations, this extrapolation reveals that the phenomenon very likely concerns tens of millions of animals every year worldwide.

If we assume the figure of 21 million per year — which may be higher or lower than the reality — this would represent around 40 farmed animals burned alive every minute in farm fires worldwide.

Conclusion

And this is therefore my conclusion: being burned alive is a common experience in this world.

Now, a brief comment.

If everyone were vegan, animal farming would not exist. And so this atrocious conclusion would be false. There would be vastly fewer beings dying burned alive. So this is another extremely strong reason to be vegan if you aren’t already. If you are already vegan, it’s another extremely strong reason to be active for animals, to promote their interests and rights as much as you can.

A confused, standard reaction to this topic is something along the line of : how could the Nazis care about animals while committing atrocities against humans? Isn’t that paradoxical?

E.g., Silas Abrahamsen made this cringe comment which is in a very similar spirit :

I don’t think there is any serious paradox there.

First, there is not even that much of a beginning of a paradox to explain. Nazi Germany was not vegan. Animals were still killed, eaten and exploited. By any serious animal ethics standard, this was still a deeply lacking system. The fact that a regime passes some basic animal protection laws does not mean it has reached moral enlightenment. It means it has cleared a very low bar.

The reason this looks morally striking to many people is that the baseline for animal ethics is atrocious. Very basic restrictions on cruelty are treated as if they were moral luxuries. They are not. They are elementary.

Still, the reaction is interesting because it connects to a broader noob-reflex in discussions of animal ethics: we should deal with humans first, then animals; humans before animals; how can we care about animals when humans still suffer? This whole framing is confused.

It assumes a moral hierarchy that is very dubious: first humans, then animals. Human concern is treated as the real moral test; animal concern as a secondary refinement, a luxury, a bonus stage.

But animal ethics is much more basic than that.

Animals are, in general, vulnerable, innocent, politically powerless beings. They do not conspire, betray, invade, enslave, organize ideologies, form hostile political coalitions etc. Humans do. Often.

Human beings can be real threats to one another. And even when they are not, they can easily be perceived as threats through propaganda, fear, resentment, ideology, personal grievance, group loyalty, rivalry, jealousy, revenge, historical memory etc.

So human-directed moral judgment is often very noisy — not only between groups, but also between individuals. Human relationships are entangled with conflict, guilt, status, self-defense, power, resentment, loyalty, betrayal, perceived desert etc.

Animal-directed moral judgment is almost always much cleaner. When someone harms an animal, there are much fewer stories to hide behind. That is why treatment of animals reveals something unusually basic about moral character.

So no, I do not find the Nazi case deeply paradoxical just like I don’t find people who are behaving well towards animals and less towards humans generally particularly paradoxical.

To be clear for the idiots out there : I am obviously in no way saying that the Nazis were morally justified in persecuting Jews or anyone else in any way. My point is simply that human-directed moral judgment is very often polluted by ideology, fear, resentment, group conflict, perceived threat, propaganda and other factors in a way that animal-directed moral judgments aren’t. So moral mistakes in the animal category are usually much much worse and revealing of a flawed moral character than moral mistakes in the human category.

The broader point has little to do with Nazism specifically. The Nazi case is just a useful entry point because it exposes well this bad assumption: that concern for animals is morally downstream from concern for humans.

That is false.

Respect for animals is not an optional extension of “real” morality. It is one of its most elementary tests.

Human morality is very noisy and much more prone to genuine and understandable mistakes.

Animal morality is much more revealing.