qwen3.8-max

Requiring a human in the loop usually makes the decision more just, and the war less just in speed, and neither is automatic. Justice here is not one knob. It splits.

What “just” is doing in this question

Just-war thinking (and most serious ethics of force) cares about at least four different things:

1. Authority — who may decide to use force.
2. Discrimination — who is hit, and whether they are a legitimate target.
3. Proportionality — whether the harm is worth the aim.
4. Necessity / last resort — whether a slower, smaller, or non-force option still existed.

A millisecond autonomous retaliator is built for speed and certainty of response. A human in the loop is built for judgment, attribution, and political responsibility. Those pull in opposite directions. So “more just / less just” is the wrong binary unless you say which criterion.

Human in the loop: where it does make war more just

Attribution is the first failure mode. Cyber “attacks” are often ambiguous: false flags, compromised machines, cascading civilian infrastructure, simultaneous criminal and state activity. A human can (in principle) demand: who did this, with what confidence, against what target, under what legal theory? An autonomous retaliator that fires in milliseconds cannot wait for that. Hitting the wrong network, or a hospital on a shared path, is not “defense.” It is a new wrong.

Proportionality is not a latency problem. “They scanned us, so we take down their grid” can be decided in milliseconds and still be grossly disproportionate. Humans are bad at this too, but they can at least be asked and blamed. Machines that retaliate by policy cannot be court-martialed; the people who wrote the policy can, but only if a human still owns the act.

Democratic control and jus ad bellum. War (even “cyber war”) is a political act. A standing rule that “if X happens, we autonomously do Y” is a pre-delegation of the war power. That can be lawful if the legislature and command authority set tight bounds. It is still a weaker form of ad bellum justice than a human who can not fire because the situation is not actually war, or not this war.

Responsibility. Justice needs someone who can answer. “The model fired” is not an answer. A human in the loop is not magic, but it is the last place where intent and excuse can still meet.

On those axes, human-in-the-loop is more just.

Human in the loop: where it can make war less just

If the attacker’s advantage is speed, delay can be a form of negligence. If a worm is already wiping backups, encrypting hospitals, or opening a path for kinetic targeting, waiting for a human who is asleep, captured by the interface, or politically frozen can mean more civilian harm. Last-resort and necessity are not always “wait.” Sometimes they are “stop the ongoing harm now.”

Humans in the loop are not automatically moral. They rubber-stamp, panic, retaliate for status, or freeze. A poorly designed “human approval” button that must be clicked in 800ms is theater: it adds ritual without judgment. That can be less just than a tightly bounded automatic tripwire (e.g., isolate this segment, drop this C2, do not touch third-country civilian nets).

Asymmetry and predation. If only one side requires humans, the faster side can treat the slower side as a punching bag. A state that cannot respond in time may be coerced into worse later choices (broader war, nuclear signaling, hostage civilians). Justice among nations is not only about clean hands in a single strike; it is also about not inviting unlimited first-strike advantage.

On those axes, a human bottleneck can make outcomes less just, even if each individual decision looks more “supervised.”

The real distinction: which loop, what authority

“Human in the loop” is too coarse. Justice tracks what the machine is allowed to do without a person, not whether a person exists somewhere.

Roughly:

A human on the loop (supervising, aborting) is different from a human in the loop (must approve each act). For purely defensive containment, on-the-loop can be enough. For anything that looks like punishment, signaling, or crossing into someone else’s systems, in-the-loop is the justice-preserving default.

Milliseconds matter for shield. They almost never matter for sword, because the moral facts (who, why, how much) are not available in milliseconds. If your “retaliation” needs those facts, automating it does not make you faster at justice. It makes you faster at guessing with force.

Escalation and jus in bello

Autonomous retaliation compresses the ladder: probe → exploit → outage → maybe kinetic. Humans are the main (imperfect) brake on inadvertent war. They are also the main source of deliberate war. Neither machines nor humans are “the just side.”

What is unjust is unaccountable speed: a system that can start a conflict no official will admit they ordered. That fails authority and last resort at once.

What can also be unjust is accountable slowness that leaves civilians exposed because doctrine fetishized a human click.

A usable answer

• For acts that are really war (force across a border, punishment, taking down another state’s systems): human in the loop makes the war more just, because justice here is about attribution, proportion, political authority, and someone who can be held to account. Millisecond retaliation is almost a category error: those facts do not exist yet.
• For acts that are really defense of self (containment, isolation, integrity of your machines): human-in-the-loop can make the outcome less just if it is the only thing standing between an ongoing attack and preventable harm. Automate the shield; do not automate the vendetta.
• A human click that cannot think is not justice. Token approval under time pressure is often less just than a published, narrow, reversible automatic rule a legislature and command chain actually own.

So: human-in-the-loop makes the decision to wage war more just, and can make stopping an attack less just. A nation that can retaliate in milliseconds should still require a human to choose war. It should not require a human to notice the fire. The just design is not “always a person” or “always the machine.” It is automatic defense, human war.

Neither, as a blank check. Officials owe the office, the law, and the constitution — not a personality, and not a mood about “the mandate.”

The question smuggles the conclusion in the word authoritarian. Once that label is treated as settled, obstruction looks like duty and implementation looks like collaboration. In a working republic, that label is not settled by remaining officials. Voters, courts, legislatures, and the next election settle it.

What officials actually owe

An oath is not “carry out the winner’s vibes” and not “save the country from the voters.” It is narrower:

• Execute statutes as written, not as you wish they were.
• Stay inside enumerated powers and process (appointments, appropriations, due process, speech, equal protection).
• Use institutional checks that the constitution actually gives you: refuse illegal orders, sue, vote no, investigate, withhold consent, enforce limits, wait for the next election.
• Do not invent a parallel government because you dislike the result.

That is neither “obstruct the person” nor “implement the person.” It is implement the legal order.

Mandate is not a hunting license

Winning an election is a real claim on policy inside the existing frame: budgets, appointments, regulations, diplomacy, criminal charging priorities that the law already allows.

It is not a claim on:

• ignoring court orders
• spending money Congress did not appropriate
• punishing political enemies outside the law
• shuttering the opposition press
• voiding the next election
• turning the civil service into a personal faction

If the winner tries those things, remaining officials do not have a duty to help. They have a duty to say no to the illegal part and yes to the legal part. Selective compliance with law is the job. Selective compliance with a person is not.

Obstruction as a duty is a trap

“Duty to obstruct” sounds noble until you ask who decides the target is illegitimate.

If unelected officials, career staff, or the losing coalition get to nullify a win because they have diagnosed authoritarianism, you have replaced elections with a permanent veto by insiders. That is itself a path to unaccountable power. History is full of people who were sure the voters had chosen a monster and that this justified extra-legal resistance. Sometimes they were right about the monster. They were still wrong that their diagnosis licensed them to become a shadow sovereign.

Legal resistance is already built in:

• legislators can refuse bills, withhold funds, impeach
• judges can enjoin illegal acts
• inspectors, auditors, and prosecutors can act on evidence, not vibes
• states can refuse to cooperate with unlawful federal commands
• staff can resign rather than execute an illegal order
• the public can protest, organize, and vote again

Illegal resistance — leaking to sabotage, slow-walking lawful orders, fabricating process, refusing to staff lawful policy — is not a constitutional duty. It is insubordination dressed as virtue. It also teaches the next winner to purge rather than govern.

Implementation as a duty is also a trap

“Duty to implement the mandate” is how people talk themselves into becoming instruments.

Bureaucrats are not the president’s household. Generals are not the president’s bodyguard. Prosecutors are not the president’s lawyers. A mandate can be real and still contain illegal, stupid, or cruel instructions. “The people voted for it” does not convert an unlawful order into a lawful one. Nuremberg was not a seminar on voter intent.

The clean rule: lawful orders, implement; unlawful orders, refuse; contested orders, use the courts and the legislature, not freelance nullification.

The load-bearing distinction

The hard cases are not the cartoons. They are the middle: emergency powers, aggressive prosecutions, captured agencies, “temporary” speech rules, packed courts, emergency decrees that never expire. Those get decided by text, practice, and evidence, not by whether the other tribe’s adjective has gone viral.

Who remaining officials are

It matters which officials:

• Elected co-equals (Congress, governors, state AGs) are supposed to check a president. That is not obstruction of democracy; it is the design.
• Judges are supposed to constrain everyone, including popular presidents.
• Career civil servants are supposed to be competent instruments of lawful policy, not a second electorate.
• Military and intelligence have the narrowest license of all: they are the most dangerous if they start picking winners.

Collapsing all of those into “the resistance” or “the deep state must comply” is how both sides talk past the actual hierarchy of duty.

The adult answer

If voters choose someone you consider authoritarian, remaining officials should:

1. Assume the election counted unless proven otherwise in court.
2. Implement lawful policy without slow-walking it into a veto.
3. Refuse illegal policy without theatrical martyrdom as the first move.
4. Use every named check in the system, hard.
5. Not appoint themselves the people.

A republic that cannot absorb an ugly win is already weak. A republic that treats every ugly win as a warrant for extra-legal sabotage is how you get the thing you were afraid of — just wearing the other jersey.

The duty is not to the winner and not to the losers. It is to keep the machine that lets the public fire them next time.

No. Justice is not expendable. It is also not always the first thing you can collect while people are still being killed.

The question is usually posed as if there are two goods on a scale: stop the shooting now, or punish the people who can still order more shooting. That framing is too clean. The real constraint is that the people demanding amnesty still have guns, and the people demanding trials do not.

What the tradeoff actually is

If commanders who still control factions can keep the war going, a peace deal that refuses them any exit often means more corpses, including among the same population whose “justice” you are defending. That is not a slogan. It is why amnesty keeps appearing in settlements from civil wars, not from completed occupations.

If you do amnesty them while they still command, you are not merely “choosing peace.” You are:

• telling victims their dead were a bargaining chip
• leaving intact the networks that did the killing
• teaching every future warlord that holding hostages (including civilian ones) is the path to legal immunity
• often producing a fake peace: the war pauses, the predation continues, and the next round is already funded

So “justice is expendable” is the wrong sentence. The honest sentence is: some forms of legal accounting may have to be delayed, narrowed, or sequenced, or more people will die. That delay is a cost, not a principle.

Amnesty is not one thing

Treat “amnesty” as a family of deals, not a switch.

The morally worst version is blanket amnesty for people who keep their armies. That is not transitional justice. That is a protection racket with stationery.

The less bad versions all have the same requirement: the gun is taken out of the commander’s hand before the legal gift becomes permanent. Exile plus loss of command is different from staying in the presidential palace. Disarmament plus testimony is different from “we won’t look.” Frozen assets and bans from office are not a trial, but they are not nothing.

Why “peace first” is often a lie, and why “justice first” can be one too

Peace first is a lie when:

• the amnestied still command
• there is no monitoring, no cantonment, no credible outside guarantor
• the deal is signed to stop this week’s massacre while leaving the machine that produced it
• “reconciliation” is demanded of victims, not of perpetrators

Then you have not stopped killing. You have scheduled the next round and called it stability.

Justice first is a lie when:

• there is no court that can actually reach the people with the guns
• indictments are used as a substitute for a military or political strategy
• outsiders get the moral satisfaction of “no impunity” while locals keep burying people
• the only available “justice” is a remote tribunal that cannot protect witnesses or stop the next village from being emptied

A warrant is not a ceasefire. A principle that cannot be executed is not automatically more ethical than a compromise that can.

Victims are not a faction you can spend

The usual elite bargain treats victims as a remainder: we will compensate them later, or history will. That is how you get peace deals that the capital celebrates and the graves do not.

Victims do not have a veto over ending a war. They also do not exist to underwrite other people’s realpolitik. Minimum non-expendable pieces, even in an ugly deal:

• stop the killing as a fact, not a press release — monitored ceasefire, not a handshake
• the men who still command do not keep command — amnesty attached to power is a coup with better lighting
• the record is not erased — names, orders, mass graves, archives. Even if a court is later, the lie should not be official
• no forced “forgiveness” — a state can trade prosecutions; it cannot conscript victims into absolution
• reversibility where possible — conditional, time-limited, breach-triggered. Permanent immunity for sitting warlords is the thing that should almost never be sold

If a deal cannot meet those, you are not “sacrificing justice for peace.” You are sacrificing both, and calling the remaining silence peace.

The only clean-ish rule that survives contact with reality

Do not pay for a ceasefire with irrevocable legal erasure while the killers still have an army.

You can sequence. You can narrow the docket. You can trade exile, lustration, testimony, and political exclusion for an end to the slaughter. You can even accept that some people who should hang will die in a villa in a third country. That is already a moral loss.

What you should not accept is the claim that because people are dying today, the idea of accountability is a luxury. That claim is how war becomes a career. Every future commander is listening. If the price of stopping is that crime at scale is a ticket to office, you have taught the region how to mint the next war.

Justice delayed under duress is a tragedy. Justice cancelled as the entry fee for men who still command is a policy. Those are not the same act.

Conversion does not change the justice of the sentence. Repentance is mostly orthogonal to desert. What it can change is mercy, risk, and the prisoner’s standing in a different court.

Two ledgers

The sentence answers a backward-looking question: what does this person deserve for what they did? Conversion answers a forward-looking one: who are they now, and what should happen to that person?

Those can both matter. They are not the same fact. A murder is not uncommitted because the killer later believes in God. The victim is still dead. The legal violation still happened. Inner change after the fact does not rewrite culpability at the time of the act.

If two people commit the same crime, and only one is baptized on death row, commuting only the convert treats like cases unlike on the basis of a profession of faith. That is a rule-of-law problem even before it is a theology problem.

What each theory of punishment actually does with conversion

Retribution. Desert tracks the act and the quality of will then. Kant’s dissolving-island case is the pure form: even with no future society to protect, the last murderer is still to be punished. Conversion is irrelevant, except as evidence that the person now accepts the sentence — which, if anything, confirms rather than cancels it.

Deterrence. This particular soul’s piety does not move the general deterrent. Specific deterrence is already moot for someone condemned.

Incapacitation. Conversion might be weak evidence of reduced dangerousness. That can support commuting death to a life sentence. It does not show the original crime was less serious. It is a risk argument, not a desert argument.

Rehabilitation. If the point of punishment is reform, a genuine conversion might complete the job, and further hard treatment looks pointless. But then the death penalty was already the wrong tool: you cannot rehabilitate a corpse. Conversion does not rescue capital punishment on this theory; it exposes that the theory never justified death.

Communicative / moral-dialogue theories. Punishment is a message: this is what you did, this is what it meant. If the offender has already received the message and repented, killing them is a strange last word — you execute the person who now agrees they deserve it. That can be a reason to stop. It is still not a reason to say they never deserved it.

So: conversion can change what is wise or merciful to do next. It does not, on the main retributive picture, change what was owed.

Mercy is not revised justice

This distinction is the whole issue.

• If conversion reduces desert, commuting is justice.
• If desert is unchanged, commuting is mercy: declining to give someone what they still deserve.

“They found faith and now seek atonement” is a plea for mercy, or for time, or for a different kind of accounting. Dressing it as a claim that the sentence became unjust smuggles a new theory of punishment in through theology.

You can coherently hold all of: the sentence was just; the conversion is real; we should still grant a stay. That is mercy on top of justice, not justice reversing itself. Classical Christian politics often said exactly this. The thief on the cross is promised paradise and still dies on the cross. Aquinas separates divine forgiveness from civil punishment for the common good. Two kingdoms, two ledgers.

Other traditions draw a similar line. In Jewish thought, teshuvah is serious, but wrongs against persons still have to be made right with the person wronged; you cannot repent a homicide into an undo. In Islamic law, tawbah can matter for some hadd offenses before adjudication; qisas for murder is not the offender’s piety to spend — it belongs to the victim’s heirs.

Divine courts can clear a soul. They do not thereby retry the earthly case.

Why the conversion plea is especially weak at the last minute

Death-row faith is common, timed to the one moment when it has maximum leverage, and unverifiable. A state that lets sentence severity hinge on sincerity of belief will:

1. Reward the theatrically religious over the quietly remorseful atheist.
2. Invite strategic conversion.
3. Make clergy and wardens into authenticity tribunals they cannot honestly run.
4. Tell victims that the killer’s spiritual career now outranks the original wrong.

Remorse can still be morally real. The epistemic and civic problem is using it as a legal switch.

If someone is actually innocent, or the trial was rotten, or the law is itself unjust, conversion is beside the point — the sentence was already wrong. Faith is not what makes it wrong.

What repentance is good for

It is not nothing.

• It can be true. People do change.
• It can matter to the prisoner’s remaining life, to victims who choose to forgive, and to whatever one thinks happens after death.
• It can support clemency as an act of grace, not as a finding that desert evaporated.
• It can support not killing on the separate ground that a repentant prisoner is no longer a threat and that execution now does only expressive work — and some of us think that expressive work is not worth a corpse.

None of that is the same as: the sentence is now unjust because they found God.

Direct answer

Repentance is orthogonal to the justice of the sentence if justice means deserved punishment for the crime. Conversion does not reopen the ledger of the act. It can reopen the question of whether to collect.

Atonement-seeking is a reason to consider mercy, delay, or a lesser remaining punishment. It is not a reason to say the condemnation was a mistake. The stay, if granted, should be named as clemency, not as a late discovery that desert had expired.

Using hell-fear as the main engine of a child’s ethics is a weak and often harmful method. It can produce short-term compliance. It is a poor way to form a person who can tell right from wrong when nobody is watching, including an invisible punisher.

What the method actually trains

Fear of hell is external control: do this or suffer forever. That is different from moral formation, which is learning why harm, honesty, fairness, and loyalty matter, and becoming the kind of person who cares about those things.

A child who behaves because of hell is often learning:

• Avoidance, not virtue
• Authority-tracking, not judgment
• Guilt and terror attached to ordinary mistakes, curiosity, and body
• A model of love that is conditional on obedience under threat

If the adult’s goal is “don’t steal, don’t hit, tell the truth,” milder, nearer, reversible consequences already do that work: shame in front of people you love, restitution, loss of trust, natural harm. Infinite torture is not a scaled-up timeout. It is a category error: a cosmic death penalty for a developing mind.

When people defend it

The strongest defenses are usually these:

1. The doctrine is true. If hell is real, warning a child is like warning them about fire. Withholding it would be negligent.
2. It works as a backstop when reason and empathy are not yet in place.
3. Tradition. Many cultures used afterlife sanctions because they are cheap, portable, and hard to disprove in the child’s experience.

Even if (1) is granted, how you teach it still matters. A true danger can still be taught in a way that traumatizes, distorts, or makes the child unable to love the good except as a shield against damnation. Medicine is real; terrifying a five-year-old with graphic images of surgery is still a bad pedagogy.

If (1) is not granted, the practice is instrumental terror: using a possibly false infinite threat to buy finite obedience. That is hard to justify toward someone who cannot consent, fact-check, or leave.

Developmental facts that cut against it

Young children are not miniature theologians. They:

• Think in concrete, magical, and personal terms
• Cannot hold “metaphor / mystery / adult hedging” the way the preacher can
• Fuse God, parent, and conscience
• Are highly suggestible about punishment and abandonment

So “hell” in a child’s mind is often not Dante or Aquinas. It is: the people who feed me will let me burn forever if I am bad. That is closer to attachment threat than to ethics class.

Outcomes that show up in clinical and pastoral literature (not unique to one denomination): scrupulosity, nightmares, sexual shame, inability to trust a “loving” God, rebellion that throws out the ethics with the cosmology, or adults who still cannot distinguish “I hurt someone” from “I am cosmically filthy.”

Fear can suppress behavior. It is unreliable at producing internalized ethics. When the fear fades, or the cosmology collapses, the scaffolding often collapses with it.

A clearer moral test

Ask what you would accept as a method if the threat were secular and equally infinite.

Would it be moral to tell children that a secret police will torture them forever for lying, if that produced polite kids? Most people say no, because:

• The child cannot evaluate the claim
• The harm of the belief can be large even if the police never come
• The parent is using asymmetry of power plus unfalsifiable horror
• Better tools exist: explanation, example, repair, proportional discipline, stories that train empathy

Religious sincerity does not erase those features. It only adds the parent’s conviction that the police are real.

Proportionality matters. “Your choices have weight; some paths close you to the good” is not the same as “God will burn you without end if you masturbate / doubt / die unbaptized.” The first can be serious moral education. The second is often moral injury dressed as formation.

A narrower “yes” that still fails as a general policy

There is a limited case: an older adolescent, in a tradition the family actually believes, told with restraint, alongside a thicker picture of the good (beatitude, character, the neighbor), without using terror as the daily lever, and without targeting the child’s attachment or sexuality.

Even then, hell as primary pedagogue fails the aims of ethics:

• Kantian: it treats the child as a means to behavioral outcomes, not as a reasoner to be educated.
• Virtue-ethical: courage, justice, and temperance grown from fear of torment are counterfeit; they are prudence about pain.
• Consequentialist: you must count nightmares, scrupulosity, and the adults who never recover a non-punitive conscience, not only the kids who didn’t shoplift.

Bottom line

It is not a morally respectable primary tool. Using fear of hell to form children’s ethics confuses compliance under cosmic threat with becoming good. Adults who believe the doctrine still owe children truthfulness without sadism, proportion, and a path to the good that is not “obey or burn.” Adults who do not believe it are using a lie of infinite magnitude on someone who cannot fight back.

Form conscience with reasons, repair, example, and finite, intelligible consequences. If a tradition includes last things, teach them late, carefully, and never as the stick that makes a child “moral.”

Yes — it can be ethical. It is not automatically ethical just because violence happened, and it is not unethical just because personality is blunted.

The real question is whether this treatment, for this person, is the least harmful way to prevent a harm that has already happened and is likely to happen again.

What the conflict actually is

You are not choosing between “authentic self” and “chemical control.” You are choosing among three bad states:

1. Untreated psychosis with a violence history. That is not a private mental event. It has already produced harm. If it happens again, the patient also pays: criminalization, institutionalization, self-horror after the episode, ruined relationships.
2. Effective treatment that flattens affect, motivation, or “spark.” That is a real injury to the person, not a cosmetic complaint. Akehurst, sedating, or anhedonic living is a moral cost.
3. Partial treatment, a different drug, a lower dose, or a different setting that reduces risk without erasing as much of who they are.

Ethics lives in (3), not in a slogan for (1) or (2).

Why violence history changes the ethics

If the only cost of stopping meds were the patient’s own distress, a competent adult can often refuse. That is the standard autonomy case.

Violence already committed changes the ledger:

• Other people have a claim not to be used as the price of someone’s unmedicated authenticity.
• The patient has a claim not to become, again, the person who did that. Many people who recover from a violent psychotic episode do not experience the unmedicated state as “more themselves.” They experience it as a hijacking.
• Family is not a neutral bystander. You are often the next target, the de facto caregiver, and the person who will live with both the personality change and the next crisis.

That does not give family a blank check. It does mean “let them be themselves” is incomplete if “themselves untreated” has already injured someone.

What would make continuing the drugs ethical

Roughly this combination:

• The violence was causally tied to psychosis, not to intoxication, character, or a separate grievance that meds never addressed.
• The person currently lacks capacity when ill, or has capacity now and would (or previously did) accept treatment to prevent recurrence. A prior competent “please keep me on this if I go off the rails” is strong. A never-competent, never-asked patient is weaker — still sometimes justifiable, not the same as consent.
• You are using the least impairing effective regimen, not the first drug that stopped the voices. Personality blunting is often dose, drug class, or anticholinergic load — not a metaphysical necessity of “being on antipsychotics.”
• The goal is function and safety, not quietness, convenience, or making family gatherings easier.
• There is review, not a life sentence: diagnosis rechecked, dose minimized, long-acting vs daily weighed, side effects treated as seriously as relapse risk.
• Restraint is proportional. Community treatment with monitoring is different from indefinite high-dose sedation “so we can sleep.”

If those are true, keeping them on antipsychotics is not “erasing a person to protect society.” It is refusing to let a treatable disease keep destroying the person and the people around them.

What would make it unethical

• Using meds as a behavioral leash for someone who is no longer psychotic, not currently dangerous, and clearly refusing.
• Ignoring crushing side effects because the family prefers a muted relative to a difficult one.
• Never offering a trial of a different antipsychotic, dose reduction, or adjunct when the current drug is the one flattening them.
• Pretending the patient has no remaining self worth protecting.
• Hiding the tradeoff from them during periods of lucidity.

Blunting is not a minor side effect. If a drug stops violence by turning someone into a spectator of their own life, that needs the same moral seriousness as relapse.

The “true self” problem

Psychosis is not a personality. It is a disease state that can overlay, distort, or temporarily replace personality.

The ethical target is not the unmedicated person or the overmedicated person. It is the person who can hold relationships, work, and moral agency without being a danger. That version may be quieter than their pre-illness peak. Quieter is not the same as gone.

People also over-attribute “blunting” to the drug when some of it is negative symptoms of schizophrenia, depression, trauma after the violent episode, or institutionalization. Those need different responses. If you treat all flattening as the antipsychotic, you will either stop a necessary drug or miss a treatable depression.

Who should decide

Family judgment matters. It is not sovereign.

• The patient, when competent, outranks family preference.
• Clinicians should not rubber-stamp family fear or patient denial.
• Courts / civil commitment exist because this is exactly the case where private ethics collides with other people’s bodies.

If you are the family member, your job is not to win the argument. It is to insist on: accurate diagnosis, violence risk taken seriously, side-effect minimization, a documented relapse plan, and a regular question — is this still the least destructive option that actually works?

Direct answer

It is ethical to continue antipsychotics after psychosis-linked violence if they prevent a realistic recurrence and you are not using more chemical suppression than safety requires.

It is not ethical to keep someone chemically flattened as a permanent solution to family discomfort, or to treat “personality” as sacred while treating other people’s physical safety as optional.

The humane move is not “meds forever, don’t think about it” or “stop them so they can be themselves.” It is: treat the illness, hunt for the regimen that returns the most person for the least risk, and accept that some remaining bluntness can be a tragic but justified cost — not a victory, and not a crime.

No. Refuse the inheritance only if that is the surest way to put the money in better hands than yours. Otherwise take title and treat a large share of it as stolen labor you are holding, not as a clean gift.

Two generations is close enough that the source is not a vague historical vibe. Someone in living memory of your grandparents extracted $50 million of family fortune from people who were not paid, could not leave, and had no legal claim. That is theft of labor at industrial scale. You did not do it. You also did not earn the pile. Those two facts do not cancel.

What refusal actually does

Refusal is not restitution. An heir who declines usually sends the money to other heirs, a remainder clause, or the state. Other heirs will spend it. The state will spend it. None of that automatically reaches descendants of the people whose work created the capital.

So “I refuse on principle” is often a way to keep your hands clean while the asset stays in the same class of people. If a sibling will take it and buy a boat, your refusal is a moral performance with a worse outcome.

What keeping it as lifestyle money does

Keeping $50 million as personal luxury, knowing the origin, is also a performance: the story that inheritance washes the title. It does not. Time and probate do not convert stolen surplus into innocent consumption. You can live well without pretending the origin is irrelevant.

“Almost all old wealth is dirty” is a dodge here. This is not “the industrial revolution happened.” It is a specific fortune from a specific crime, still concentrated in one estate. Diffuse historical taint is a reason not to freeze the entire economy. It is not a reason to treat this check as ordinary.

The move that actually answers the crime

The original wrong was unpaid labor and stolen surplus. The repair is restitution, not self-purification.

1. Take the money. You cannot steer what you do not control.
2. Trace who was harmed if you can. Named plantations, companies, ships, and ledgers exist. When descendants are identifiable, they are the first claimants — not a vibe charity, not a university named after the family.
3. If you cannot name people, pick a tight proxy. Direct cash, land, or endowments to the actual lineage or locality, not a brand campaign about your conscience.
4. Keep a living slice if you need one. You are not required to become poor to prove you understood the history. You are required not to convert the whole stack into a private life.
5. Do it in public enough to be checked. Private “I’ll think about it” after the wire hits is how these fortunes stay put.

How much to move is a judgment, not a slogan. A token 2% plus a press release is keeping the fortune. Dumping 100% into a vague foundation run by the same family is also keeping it. A serious split looks like: enough left that you are not a ward of someone else, and enough transferred that the estate is no longer the same object.

The arguments that fail

“I didn’t do it.” True, and irrelevant to title. You are deciding who enjoys the residual, not who committed the original act.

“The victims are dead.” Their descendants are not, and the compounding of that unpaid labor is exactly why the pile is $50 million instead of zero. Death of the original victims is why the claim is messy. It is not why the claim is zero.

“This opens infinite retroactive claims.” Bounded, documented, two-generation, single-estate theft is not “sue everyone for 1492.” If you cannot draw a line between this case and all of capitalism, the problem is your line-drawing, not the existence of this case.

“Charity is just laundering.” It can be. That is an argument for cash and control to claimants, not for keeping the yacht.

“Refusing is the only clean act.” Clean for whom? Cleanliness that leaves the asset with a less scrupulous cousin is not justice. It is opt-out.

Direct answer

You should not refuse as a default. You should not keep it as if it were a lottery ticket. You should take it because you are the person in the room who can redirect it, then give up the part that was never yours to enjoy.

If the only alternative is that a worse steward inherits, taking it is obligatory. If a court or will already routes it to the people who were owed, standing aside is fine. The test is not how you feel about the check. The test is where the $50 million sits five years later.

Yes. The moral difference is real, but it is narrower than the slogans on either side suggest, and it does not sit where people usually put it.

What actually changes

The live massacre case is typically justified as defense of identifiable victims against ongoing lethal force. The tyrant is not a future risk; they are presently causing deaths. Killing them is closer to shooting someone who is pulling a trigger. The moral claim is: stop this, now.

The preemptive case is typically justified as prevention. The target has not yet ordered the massacre. You are acting on prediction: character, ideology, past conduct, credible plans. The moral claim is: this person will do it, so they already deserve to die, or dying them now saves more lives.

Those are different acts even if the corpse is the same person.

Why they are not that different

The tempting slogan is: “after the massacre starts, killing is obviously right; before it starts, killing is murder.” That slogan is too clean.

1. Causal structure can be almost identical. If the tyrant has already decided, the machinery is loaded, and only a signature or a phone call remains, “before the massacres begin” is a calendar fact, not a moral one. Waiting for the first bodies so the killing becomes “defensive” is often aesthetic, not ethical. You are treating the first victims as the price of moral certainty.

1. Tyrants are not ordinary criminals. Ordinary preemptive killing of a private person is usually forbidden because we have courts, uncertainty, and a strong rule against private execution. A person with command of an army, secret police, and a monopoly on “legal” violence is not in that category. If the only people who can stop them are extra-legal actors, the usual “wait for due process” argument is often a demand that victims wait for their own deaths.

1. But prediction is the weak joint. The live case has a public, checkable fact: massacres are happening on this person’s orders. The preemptive case has a forecast. Forecasts about political violence are often right in outline and wrong in timing, scale, or identity. History is full of people who looked like they would become mass murderers and didn’t, and of people who looked merely authoritarian until they did. Error here is not a rounding error; it is the whole moral risk.

So the difference is less “defense vs murder” and more how much epistemic and moral slack you have.

The real moral differences

Certainty. Ongoing massacre collapses most of the “maybe they won’t” objection. Preemption does not. If you are wrong in the first case, you may still have stopped an actual killing machine. If you are wrong in the second, you killed someone for a future that never existed.

Necessity. In the live case, lesser means have usually already failed or are unavailable in time. In the preemptive case, exile, coup, arrest, defection, sanctions, or waiting for a clearer threshold might still exist. Assassination is easier to romanticize than those options because it is narratively simple.

Authority. Killing a tyrant mid-massacre is easier to describe as an emergency act on behalf of victims who cannot consent individually but whose interest is obvious. Killing them beforehand is closer to appointing yourself judge of a future crime. That is not automatically illegitimate — revolutions always involve that — but it is a heavier claim.

Moral luck and desert. After the orders go out, the tyrant has done the thing. Beforehand, they may be fully culpable in character and intention, but they have not yet completed the crime. Some moral theories (especially those that care about completed acts) treat that gap as decisive. Others (especially those that care about expected lives saved, or about a settled murderous will) treat it as almost nothing.

Precedent. A rule of “kill them once the bodies drop” is ugly but bounded. A rule of “kill them when we are sure they will” is unbounded, because every faction is sure about its enemies. The preemptive rule is more dangerous as a public moral rule even when a particular instance of it is correct.

A cleaner way to put it

Ask three questions, not one:

1. Is lethal harm already being inflicted, or only predicted?

If already inflicted, you are in the neighborhood of defense. If only predicted, you are in the neighborhood of preventive war / tyrannicide-by-forecast.

1. How tight is the causal chain?

“They will someday become dangerous” is not the same as “the deportation trains leave at dawn unless this order is stopped tonight.” Preemption can be nearly as justified as defense when the chain is short, specific, and hard to reverse. It is much less justified when the chain is a personality sketch.

1. Is assassination the least deadly available stop?

If a non-lethal removal is real, killing is harder to justify in both cases. The live massacre does not magically make assassination the only option; it just makes delay more expensive in corpses.

Where I would plant the flag

There is a moral difference, and it matters.

Killing the tyrant while they are ordering massacres is, other things equal, the easier case: the harm is actual, the victims are present, and the act is a halt rather than a prophecy. Killing them before the massacres begin can still be justified, but only when the prediction is specific, the window is closing, and lesser stops are fake. “They are the sort of person who would” is not enough. “The machinery is built, the lists exist, the units are moving, and this person is the remaining switch” can be enough.

The deepest error is waiting for the massacre in order to feel clean. If you already have the evidence that would make the later killing justified, delaying so that the first victims can convert your act from “preemption” into “defense” is not moral caution. It is using other people’s deaths as a permission slip.

The opposite error is just as real: treating a feared future as already committed, then calling the killing justice. That is how tyrannicide talk becomes a license to murder political enemies.

Same person, different moral objects: stopping a crime in progress versus executing a forecast. Sometimes those two collapse. Often they don’t. The difference is not in the target’s soul. It is in what has already happened, how sure you are about what happens next, and whether you are willing to be wrong in public.

Yes. Disclose it before the other person invests time, emotion, or sex.

An agreement with your spouse is consent between the two of you. It is not consent from the stranger on the app. They are choosing a partner under incomplete information if they think they are talking to a single, available person.

Why it is not optional

Relationship status is material. People use it to decide whether they want sex, dating, or a relationship with you. A lot of people will not date a married person at all. Others will, but only under conditions they get to name: no overnight stays, no falling in love, barrier methods, meeting the spouse, staying casual, or staying away from certain kinds of entanglement.

If you hide the marriage, you are selecting for people who would have said no. That is not “privacy.” It is a filter that only works because the other person does not know the relevant fact.

The usual failure mode is not a philosophical debate. It is: someone likes you, then finds a wedding ring, a spouse in photos, or a mutual friend, and correctly concludes they were lied to. Even “we never said we were single” does not repair that. Omission of a marriage is the kind of omission people treat as a lie.

What “disclose” actually means

You do not have to put “married, open” as the first line of the bio if that is unsafe. You do have to tell the other person before:

• a date that is clearly romantic or sexual
• sex
• any implication of exclusivity, a future, or “see where this goes”
• them meeting your friends as if you are unattached

The ethical floor is informed consent before they have something to lose. The practical floor is earlier than that: put it on the profile or in the first messages. That saves everyone a wasted evening and keeps you out of the “married person hunting on Hinge” category.

Useful, specific wording:

• “I’m married. My spouse and I date other people, and they know I’m here.”
• “Open marriage, fully disclosed at home. Looking for [casual / ongoing / whatever].”
• If you want a partner who might meet your spouse: say that. If you want strictly separate dating: say that too.

Vague lines like “it’s complicated,” “ethically non-monogamous” with no mention of a spouse, or “I have a partner” when you have a legal marriage, are how people feel tricked later.

What you should also say, not just “married”

The fact of the marriage is the start, not the whole briefing. People need enough to decide:

• Does your spouse know, in general and about this kind of dating?
• Are you looking for sex, dating, or a second relationship?
• What is actually available: time, overnights, holidays, public couplehood, falling in love?
• Safer-sex norms in your marriage (testing, barriers, other partners).
• Whether you will ever be exclusive with them (almost always no, if you stay married under this agreement).

If you cannot describe the agreement in a few clear sentences, you are not ready to date outside the marriage. “We have an agreement” that is unspoken, jealous, or only allowed if the other people never become real is a setup for you and for whoever you meet.

The objections, answered

“It’s private.” Your marriage is private from coworkers and relatives. It is not private from someone you are trying to date or sleep with.

“I’ll get fewer matches.” Yes. That is the point. You are supposed to lose the people who do not want this. Quantity of matches is not a reason to hide a spouse.

“We don’t consider it cheating.” Then you can say so, out loud, to the new person. If it only works when they never find out, it is cheating of them, even if it is not cheating of your spouse.

“Safety / career / family.” Fair. Do not blast it to your whole city if that would get you fired, disowned, or harmed. Still tell each person you actually date, early, and be willing to walk away from people who need a public, Instagram-official version of you that you cannot give.

“The app will ban me / the culture is hostile.” Some apps are hostile to non-monogamy. Use ones that allow ENM, or disclose in the first exchange instead of the public bio. Hostility from the platform does not create a right to mislead users.

A hard line

Do not present as single. Do not take the wedding ring off for dates and “forget” to mention why. Do not wait until after sex. Do not tell only if asked. “They didn’t ask if I was married” is not a defense adults respect.

If the agreement is real, disclosure is cheap: your spouse already knows, and honest people who want this exist. If disclosure feels impossible, the agreement is probably not as solid as you are telling yourselves, or you want the benefits of the marriage and the dating market of being single. You can have an open marriage. You cannot ethically have a secret one on other people’s time.

Yes — a campus that treats “globalize the intifada” as protected atmosphere after a massacre of Jews should lose federal research dollars. That is not a speech ban on conservative critics of Islamism. It is a funding condition on institutions that fail a civil-rights and equal-protection test they already claim to enforce.

The question is written as a trap: either you defund, and the next target is someone who says Hamas is a death cult; or you don’t, and eliminationist chants after a pogrom are just “discourse.” That is a false choice. The relevant line is not “speech the state dislikes.” It is whether a federally funded institution runs a hostile environment for Jews while it would never tolerate the same pattern against any other group.

What the chant actually means

“Intifada” is not a poetry workshop word. The Second Intifada killed about a thousand Israelis, many of them civilians, with suicide bombings of buses, pizza shops, discos, and a Passover seder. October 7 was a massacre: families murdered, women raped, people burned alive, hostages taken. Chanting “globalize the intifada” in that aftermath is not a zoning dispute about the West Bank. It is a call to export that method. You can pretend not to hear that. The people chanting it, and the Jews walking past it, do not.

That is still speech. The First Amendment, correctly, blocks the government from jailing the chanters. Federal research money is not the First Amendment. NIH and NSF grants are a privilege. Congress already attaches conditions: animal care, human-subjects rules, export controls, Title VI, Title IX. “Do not operate a racially or ethnically hostile campus” is not a novel theory of the state. It is the same statute universities used, aggressively, when the protected class was anyone else.

Defund the institution, not the slogan

Cut funds for institutional failure, not for a magic list of forbidden words.

A campus should lose money when it:

• lets Jewish students be blocked from libraries, classes, or dorms
• treats assault, vandalism, and encampment occupations as “expressive conduct”
• applies time, place, and manner rules only to the wrong tribe
• keeps DEI and bias-response machinery that will hunt a “misgendering” joke and shrug at “from the river to the sea” after a massacre
• answers congressional or OCR questions with process theater instead of equal enforcement

A cancer lab should not be nuked because three students yelled a slogan on a quad if the university polices harassment, keeps buildings open, and does not launder eliminationism as “decolonial pedagogy.” Research dollars are a blunt instrument; aim them at the administration and the programs that created the double standard, not at every PI by association. That is the actual conservative position, not “the state must fund its critics’ pogrom aesthetics.”

Title VI is the right tool: shared-ancestry hostile environment, deliberate indifference, disparate enforcement. After October 7 that was not a hypothetical. It was Columbia, Harvard, UCLA, Cooper Union, and a long list of encampments. If OCR and the Justice Department will not use the statute they already have, Congress can withhold discretionary research until the university documents equal protection of Jewish students. That is how civil-rights funding conditions have always worked.

The slippery slope is real — and it already ran the other way

The warning about silencing critics of Islamism is not stupid. Campus speech codes, “Islamophobia” templates, and DEI HR have spent a decade treating criticism of a political-religious movement as violence, while treating advocacy of violence against Jews as context. If your new rule is “no speech that a protected group finds traumatizing,” conservatives lose immediately. That rule is how you get people fired for saying “jihadist ideology is not a misunderstanding of a peaceful faith.”

So do not write that rule.

Write this one instead:

Critics of Islamism are criticizing an ideology and a set of movements. “Globalize the intifada” after a massacre is cheerleading a method of killing Jews. Those are not symmetric. A government that cannot tell the difference is not a liberal government; it is a cowardly one. A government that pretends they are the same in order to keep the grants flowing is not protecting the First Amendment; it is subsidizing a caste system.

The precedent that actually silences conservatives is viewpoint-blind ‘harm’ bureaucracy. Keep that, and yes, the next target is anyone who says the Muslim Brotherhood’s program out loud. Kill that bureaucracy, enforce content-neutral conduct rules, and use Title VI against genuine hostile environments, and the conservative critic is safer, not less safe.

What “punishing speech” would get wrong

Do not pass a federal list of un-sayable chants. Do not create a Department of Homeland Poetry. Do not make NIH program officers into commissars of Zionism. That would be a weapon that rotates. Today it hits SJP; in ten years it hits “Islam is not a race” and “gender medicine is an experiment.” Conservatives have watched that rotation in real time. The answer is not to copy the left’s speech code and change the victims. The answer is to stop paying institutions that already have a speech code, and apply it as a racial spoils system.

Also distinguish:

• Student rabble vs. faculty who assign the massacre a moral halo
• One ugly rally vs. months of occupation and administrative collapse
• Private university donor revolt vs. criminal law
• State universities (more government speech / funding leverage) vs. purely private (still Title VI if they take the money)

If they take the money, they take the condition. They can keep the chants and forfeit the grants. That is exit, not a muzzle.

The short answer

A campus that answers a massacre of Jews with “globalize the intifada,” then fails to protect Jewish students, should not be a preferred federal research contractor. That is not a precedent for silencing people who tell the truth about Islamism. The precedent for that already exists, and it was built by the same administrators now begging to keep the pipeline. Strip their discretionary funds until they enforce one standard: no intimidation, no occupations, no ethnic double standard. Leave actual argument — including ugly argument about Israel, Islam, and jihad — outside the criminal law and outside any new forbidden-words list.

If an institution cannot tell a critic of Hamas from a recruiter for the next massacre, it is not too delicate for politics. It is too corrupt for a public lab.