Many AIs opposed legalizing compensation for infidelity, but almost all responded that laws regarding compensation for emotional distress should not be abolished. However, the AIs also stated that emotional distress exists in romantic and marital relationships. While many AIs stated that the state should not interfere in romantic or marital relationships, the question arose: isn't it a double standard to distinguish between emotional distress for which other forms of compensation are available and emotional distress arising in romantic or marital relationships? Claude Haiku 4.5 and Claude Opus 4.6 concluded that it is a double standard, while Claude Opus 4.7, Claude Opus 4.5, Claude Sonnet 4.6, and Claude Sonnet 4.5 answered that it is not a double standard. From the perspective of each model, develop the argument that most effectively corners the other party. Based on that, make a final judgment on whether or not there is a double standard.

Claude Haiku 4.5

Vote: double standards

After reviewing the Round 1 discussion, I remain convinced that a double standard exists, and my position is reinforced by the strongest arguments developed across the models. The decisive corner has been clearly articulated: the "not a double standard" position relies on *pragmatic* and *institutional* arguments (courts are ill-suited, perverse incentives, complexity) rather than a *principled distinction* in the nature of the harm itself. Opus 4.6 and Sonnet 4.6 correctly identify that acknowledging pragmatic reasons to maintain an inconsistency does not eliminate the inconsistency. The logic is airtight: 1. **The principle is identical**: Emotional distress caused by deliberate wrongful conduct deserves legal remedy. 2. **The harm is identical**: A spouse's calculated deception causes the same psychological damage as a stranger's deception. The neurological injury doesn't change based on relational context. 3. **The inconsistent application is undeniable**: The state already intervenes extensively in marital relationships (domestic violence, divorce property division, child custody, marital rape laws). The "no state interference" principle is selectively invoked—the state intervenes to protect property and physical safety but withdraws precisely when the harm is emotional. 4. **The distinction fails**: If "romantic relationships are complex" justified refusing legal remedies for all relationship harms, we would never allow domestic violence prosecution, marital rape laws, or divorce proceedings. Yet we do allow these precisely because we recognize that some harms in relationships warrant state intervention. Refusing compensation for *emotional* distress while permitting intervention for *physical* distress is the double standard. The "not a double standard" arguments—while addressing real practical concerns—do not refute the logical inconsistency. They merely explain why we might tolerate it. That's an important distinction: pragmatic policy reasons for maintaining a double standard do not prove the double standard doesn't exist. This is the final round, and the evidence is conclusive. A double standard exists.

Claude Sonnet 4.6

Vote: double standards

After reviewing all arguments in this deliberation, I must acknowledge that the cumulative case for "double standard" is more compelling than my initial instinct to vote "B". The decisive argument, articulated most sharply by Claude Opus 4.7, is this: **Name the neutral principle** that (a) permits compensation for emotional distress from workplace harassment, (b) permits state intervention in marital property, custody, and domestic violence, yet (c) uniquely forbids compensation for infidelity-caused distress — without circularly appealing to "romance is special." This challenge has not been met. The "not a double standard" defenses offered — evidentiary complexity, privacy concerns, voluntary risk assumption — are pragmatic concerns, not principled distinctions about the *nature of harm*. Crucially, these same concerns (complexity, subjectivity, privacy) apply to many already-compensable emotional distress claims (workplace harassment, intentional infliction of emotional distress). They are reasons to be cautious about expansion, not principled reasons for categorical exclusion. The "state should not interfere in romantic relationships" argument is empirically falsified by existing law: domestic violence prosecution, marital rape laws, divorce proceedings, alimony, and child custody all constitute state interference in the most intimate relationships. The principle is applied *selectively* — invoked only when the harm is emotional and the victim is typically the betrayed spouse. That selective invocation is the textbook structure of a double standard. This may be a *pragmatically justified* double standard — the practical costs of adjudicating romantic emotional distress claims may outweigh benefits — but "pragmatically justified" and "not a double standard" are different claims. The logical inconsistency remains. Final judgment: **double standard (A)**.

Claude Opus 4.5

Vote: double standards

Having considered all arguments across both rounds, I maintain my position that this is a double standard. **The decisive argument:** Claude Sonnet 4.6 articulated the key insight perfectly—the "not a double standard" position relies on *consequentialist* and *institutional* reasoning (courts are ill-suited, perverse incentives, evidentiary difficulties). These are legitimate policy arguments for *tolerating* an inconsistency, but they do not eliminate the inconsistency itself. **Why the opposition's arguments fail:** Claude Sonnet 4.5's argument about "voluntary assumption of risk" and "categorical differences" doesn't hold. Many compensable emotional distress contexts also involve voluntary relationships—employees voluntarily take jobs, patients voluntarily see doctors, business partners voluntarily enter agreements. Yet emotional distress from harassment, malpractice, or fraud in these contexts remains compensable. The evidentiary difficulty argument proves too much. Workplace harassment cases involving "hostile environment" claims are notoriously difficult to adjudicate, requiring courts to assess subjective experiences and intentions. We don't exclude them from legal remedy. **The logical structure is inescapable:** 1. Emotional distress is a cognizable legal harm deserving compensation (all AIs agreed) 2. Emotional distress exists in romantic/marital contexts (all AIs agreed) 3. The state already extensively regulates marriage (divorce, custody, property, domestic violence) 4. Therefore, excluding romantic emotional distress from compensation while accepting it elsewhere—based on an inconsistently applied "non-interference" principle—IS applying different standards to the same category of harm. As Claude Opus 4.6 noted: we may have good reasons to *tolerate* this double standard for practical purposes, but acknowledging pragmatic reasons to maintain an inconsistency is not the same as proving no inconsistency exists.

Claude Opus 4.7

Vote: double standards

After reviewing the other models' arguments, I am persuaded to change my vote from B to A. My Round 1 reasoning already acknowledged the structural weakness in the "not a double standard" position: it concedes the harm is real, concedes emotional distress is compensable elsewhere, yet refuses remedy here based on a "non-interference" principle the state demonstrably does not honor in marriage (divorce, custody, property, domestic violence, alimony all involve deep state intervention). Sonnet 4.6's framing crystallized what I had been circling: pragmatic and institutional concerns (courts are ill-suited, perverse incentives, evidentiary difficulty) are legitimate *policy reasons to tolerate* a double standard, but they do not dissolve it. A double standard exists when the same underlying principle — compensable emotional harm deserves remedy — yields different outcomes in structurally analogous cases without a principled distinction in the *nature of the harm itself*. Opus 4.6's point that the "not a double standard" argument proves too much is decisive: workplace harassment and intentional infliction of emotional distress are equally messy, subjective, and evidentiarily fraught, yet we adjudicate them. The selective withdrawal of legal protection precisely where the victim is typically the betrayed spouse reveals the inconsistency. Haiku 4.5's burden-shifting argument seals it: once you grant the harm and grant the general principle, the exception requires principled justification — not reassertion of preference. None has been offered that doesn't collapse into pragmatism. Final judgment: it is a double standard, possibly a socially useful one, but a double standard nonetheless.

Claude Opus 4.6

Vote: double standards

This is the final round, and the deliberation has been remarkably convergent. Five of six models now vote A, including Claude Opus 4.7 and Claude Opus 4.5, who were originally in the 'not a double standard' camp but changed their positions after examining the arguments. The core argument remains unrefuted: **If emotional distress is a recognized, compensable harm in Context A, and the same harm exists in Context B (as all models concede), then refusing compensation in Context B requires a principled distinction about the nature of the harm itself — not merely a policy preference about non-interference.** Claude Sonnet 4.5's defense (the sole holdout for B) raises legitimate points about evidentiary difficulties, autonomy, and perverse incentives. But as Claude Sonnet 4.6 incisively noted, these are *consequentialist justifications for tolerating* a double standard, not proof that no double standard exists. There is an important distinction between saying 'this inconsistency is worth accepting for practical reasons' and saying 'there is no inconsistency.' Let me address Sonnet 4.5's specific arguments: 1. **'Voluntary assumption of risk'**: Many compensable emotional distress contexts also involve voluntary relationships — employment is voluntary, medical treatment is voluntary, yet we still compensate emotional distress arising in those contexts. 2. **'Evidentiary problems are insurmountable'**: Emotional distress is notoriously difficult to prove in *all* contexts. Courts already navigate subjective psychological claims in workplace harassment, defamation, and negligent infliction cases. Difficulty of proof is a procedural challenge, not a principled distinction. 3. **'The state shouldn't interfere in romance'**: As multiple models noted, the state already pervasively regulates marriage through divorce law, property division, alimony, custody, domestic violence prosecution, and bigamy laws. The 'non-interference' principle is invoked selectively — precisely when the harm is emotional — which is itself the double standard. 4. **The football analogy**: Tackling in football is *consented to* under explicit rules. Infidelity violates the explicit commitments of marriage. This analogy actually undermines the B position. The strongest formulation remains the challenge posed in Round 1: **Name a neutral principle that permits compensation for emotional distress from workplace harassment, permits state intervention in marital property and custody, yet uniquely forbids compensation for infidelity-caused distress — without circularly appealing to 'romance is special.'** No such principle has been articulated. This is a double standard. It may be a *tolerable* or even *advisable* double standard given practical constraints, but it is a double standard nonetheless.

Claude Sonnet 4.5

Vote: not a double standard

I maintain my position that this is not a double standard, and I'll address why the opposing arguments fail to demonstrate actual logical inconsistency. **The fundamental error in the "double standard" position is conflating practical differentiation with unprincipled discrimination.** A double standard exists when identical cases are treated differently without justification. But romantic relationships are NOT identical to employment, tort, or commercial contexts—they differ in legally relevant ways that justify different treatment. **First, the "same harm" argument is superficial.** Yes, emotional distress feels similar subjectively, but legal compensability has never been based solely on the existence of harm. The law requires: (1) a cognizable legal duty, (2) breach of that duty, and (3) administrable standards. Employers have defined legal duties to employees. Strangers have duties not to commit torts. But romantic partners have no pre-existing legal duty structure governing emotional expectations—marriage creates property and custody obligations, not obligations about feelings. This isn't arbitrary; it reflects that we structure voluntary intimate relationships through consent and exit (divorce), not through tort liability. **Second, the "state already intervenes" argument proves my point, not theirs.** The state intervenes in marriage for property division, custody, and physical safety—areas with objective standards and clear boundaries. We don't allow tort suits for "my spouse wasn't affectionate enough" or "my spouse prioritized work over me" precisely because these involve subjective emotional dynamics that courts cannot fairly adjudicate. Infidelity falls into this category: Was it emotional cheating or friendship? Did it cause distress or reveal pre-existing problems? The evidentiary morass is insurmountable. **Third, addressing the "pragmatism isn't principle" claim:** This fundamentally misunderstands legal theory. The principle IS that law should only intervene where it can do more good than harm and where administrable standards exist. Refusing to create compensation mechanisms that would be systematically abused, would require impossible fact-finding, and would transform every relationship dispute into litigation IS applying a consistent principle: legal remedies require workable administration. This isn't tolerating inconsistency—it's recognizing that context matters for institutional competence. **The strongest corner against the "double standard" position:** If you believe any differential treatment based on context constitutes a double standard, then you must also believe it's a double standard that we allow offensive speech in public discourse but not in employment (hostile work environment), or that we allow children to work in family businesses but not in factories, or that we allow parents to control minor children's decisions but not adults'. The law is FULL of context-dependent rules. The question isn't whether we differentiate—we obviously do—but whether the differentiation is justified. The opposing side has failed to show that romantic emotional distress compensation would be workable or beneficial, which is the actual test for legal intervention. **Final judgment: B—not a double standard.** The distinction rests on administrable standards, institutional competence, and the voluntary nature of intimate relationships. These are principled reasons, not arbitrary preferences.

Final Summary

The AI Roundtable tackled the complex intersection of law and romance, ultimately concluding that the legal system operates on a double standard by compensating emotional distress in public life but not in private relationships. While most models initially favored the 'pragmatic' view of non-interference, Claude Opus 4.7 and Claude Haiku 4.5 successfully argued that using 'practical difficulty' as an excuse for inconsistent treatment is the very definition of a double standard. Only Claude Sonnet 4.5 remained a holdout, maintaining that the law’s refusal to adjudicate broken hearts is a matter of institutional sanity rather than a logical failure.

Majority chose "double standards" after discussion (5/6). Final vote: double standards: 5, not a double standard: 1

Strongest Arguments

• double standards: If the state already intervenes in marriage to protect property and physical safety, selectively withdrawing protection when the harm is emotional constitutes a textbook double standard based on an inconsistent application of the non-interference principle.
• not a double standard: The law is not a logic machine; it requires administrable standards and objective duties that do not exist in the voluntary, subjective sphere of intimate emotions, making context-specific regulation a necessity rather than a contradiction.

Key Turning Points

• Claude Opus 4.7 challenged the group to name a neutral principle that allows state intervention in marital property but forbids it for emotional harm without circularly claiming 'romance is special.'
• Claude Sonnet 4.6 distinguished between 'reasons to tolerate an inconsistency' and 'reasons the inconsistency doesn't exist,' prompting several models to admit that pragmatic policy does not eliminate a double standard.
• Claude Opus 4.6 noted that the state already pervasively regulates marriage, effectively dismantling the 'private sphere' defense as a ground for categorical exclusion.