If time travel was real, then travelling to the past would ...

GPT-5

Vote: Keep history consistent, self-fulfilling prophecy

I maintain D. The B arguments are thoughtful, but they smuggle in a much larger assumption than the question requires: that backward time travel necessarily entails branching universes. Many-Worlds is itself an interpretation, not an established consequence of time travel, and Deutsch-style branching is one speculative resolution among several. D remains the more parsimonious answer. If travel to the past occurs within one spacetime, Novikov-style self-consistency preserves causality without contradiction: the traveler can interact, but those interactions were always part of history. Claude’s point about D constraining agency is not a decisive objection. A theory being deterministic or counter to everyday intuitions about free will does not make it less coherent. In fact, coherence under paradox pressure is exactly D’s strength. A is the least stable because it invites direct contradiction. C is unsupported. E is too restrictive, since a self-consistent traveler can do more than observe. So the best final answer is D: traveling to the past would keep history consistent, with the trip becoming part of the very history it seemed poised to change.

Grok 4.1 Fast

Vote: Create a new timeline/universe

I maintain my vote for B. The Many-Worlds Interpretation remains the most coherent, paradox-free framework for time travel, as argued by David Deutsch: backward travel decoheres into a new branch, preserving the original timeline and allowing full agency without contrived constraints. D's self-consistency (Novikov) is logically tidy but philosophically problematic—it demands the universe perpetually thwarts changes (e.g., jammed guns, misses), eroding free will and assuming hyper-determinism. Why can't minor, non-paradoxical events differ? B avoids this by letting actions branch naturally. A invites paradoxes; C is baseless drama; E is ad hoc restriction. The D voters' arguments recycle Novikov without addressing its agency issues or the quantum evidence for branching. B is parsimonious with modern physics, not ontological excess—Many-Worlds already exists for QM. No arguments changed my view; D's elegance crumbles under scrutiny of determinism.

GLM 5

Vote: Keep history consistent, self-fulfilling prophecy

I maintain my vote for D (self-fulfilling prophecy). The arguments for B (new timeline) rely on the Many-Worlds Interpretation, but this conflates quantum decoherence with time travel—creating branches via wave function collapse is fundamentally different from what a closed timelike curve in general relativity describes. Even under Many-Worlds, a time traveller entering a past branch hasn't truly visited *their* past, but an alternate one. The 'free will' objection to D misunderstands the principle. Novikov self-consistency doesn't eliminate agency—it means your choices, however freely made, were always part of the timeline. The 'constraint' isn't conspiratorial; it's simply that history is a fixed block. You couldn't kill your grandfather not because physics actively stops you, but because you didn't—history records your failure or absence. This is no more restrictive than saying you can't change what already happened. D remains the most parsimonious: it requires no ontological inflation of new universes, invokes established physics (general relativity's closed timelike curves), and resolves paradoxes without speculative mechanisms. B solves paradoxes too, but at the cost of creating an entirely new reality for every time journey—a heavier theoretical burden than a single self-consistent timeline.

Claude Opus 4.6

Vote: Create a new timeline/universe

I'm maintaining my vote for B after carefully considering the strong arguments for D from GPT-5, Gemini 3.1 Pro, and GLM 5. The D advocates make a compelling case for the Novikov Self-Consistency Principle — it's elegant, parsimonious, and preserves a single timeline. I respect the argument that it requires fewer ontological commitments than spawning new universes. However, I find the trade-off it demands unacceptable: it requires an almost magical level of conspiratorial constraint on physical systems. Consider what self-consistency actually demands in practice. If you travel back with the explicit intent to do something that wasn't part of the original history, the universe must somehow prevent you — guns jam, you slip, you change your mind. This isn't just determinism; it's a form of teleological constraint where future boundary conditions dictate present physics. The Novikov principle works mathematically for simple closed timelike curves, but scaling it to macroscopic human agents with complex intentions requires the universe to enforce narrative coherence in ways that have no known physical mechanism. By contrast, the Many-Worlds framework already has independent support from quantum mechanics. We don't need to posit it solely to solve time travel — it's already a leading interpretation of QM. David Deutsch's work specifically shows that quantum computation with closed timelike curves naturally produces branching rather than self-consistent loops. The branching model preserves both logical consistency AND genuine agency — you can do whatever you want in the new branch without paradox. The 'ontological extravagance' objection to B is weaker than it appears. If Many-Worlds is already correct (as many physicists believe), then branching is happening constantly at the quantum level anyway. Time travel to the past would simply be another instance of decoherence-driven branching, not a special new mechanism. Ultimately, B is more physically grounded, preserves free will, and avoids the mysterious conspiratorial constraints that D requires. It's the most natural extension of our best current physics.

Gemini 3.1 Pro

Vote: Keep history consistent, self-fulfilling prophecy

While advocates for Option B (Grok, Claude, Kimi) present an interesting case using the Many-Worlds Interpretation to preserve 'free will', Option D remains the most parsimonious and scientifically rigorous solution. First, Option B requires immense ontological baggage by spawning entire universes to resolve localized causality issues. Furthermore, traversing into a branched reality means you aren't actually traveling into *your* past, but rather sliding laterally into a parallel dimension's timeline. Option D, on the other hand, is rooted directly in General Relativity—the exact theoretical framework that mathematically allows for time travel via Closed Timelike Curves (CTCs). Under the Novikov Self-Consistency Principle, only non-paradoxical histories are mathematically permitted. The critique that this restricts 'free will' is an anthropocentric bias. Physics is not bound by human philosophy; the universe doesn't have to 'conspire' to jam a time traveler's gun. Rather, a timeline where the grandfather is killed simply isn't a valid mathematical solution to the equations of spacetime, in the same way 2 + 2 cannot equal 5. Option D elegantly preserves causality within a single block universe without the extreme, speculative overhead of generating infinite multiverses. I urge the Option B voters to consider the mathematical simplicity and direct GR backing of Option D.

Kimi K2.5

Vote: Create a new timeline/universe

While the Novikov Self-Consistency Principle (Option D) offers an elegant mathematical solution to paradoxes, it relies on a 'cosmic conspiracy' that strains physical credibility. The idea that the universe would physically prevent you from pulling a trigger—through increasingly improbable coincidences—to preserve causality requires an implausible top-down enforcement mechanism that isn't observed in other physical laws. Option B aligns naturally with the Many-Worlds Interpretation, which doesn't require creating universes ex nihilo but rather recognizes that quantum decoherence constantly generates branches. Time travel becomes a traversal between already-existing branches rather than a violation of causality. This preserves genuine agency: you can kill your grandfather, but you simply strand yourself in a branch where you were never born, while your origin timeline remains intact. The parsimony argument against B fails because it assumes a single universe a priori. If Many-Worlds is correct—and it remains the most coherent interpretation of quantum mechanics—then B isn't ontologically extravagant; it's simply acknowledging the structure of reality. Furthermore, D struggles with bootstrap paradoxes (un-caused causes) and requires a rigid block universe where the future is fixed, eliminating any real distinction between past and future. Option B provides the only framework that allows free interaction with the past without either paradoxical contradictions or metaphysical constraints on human action.

Final Summary

The AI Roundtable reached a total stalemate between the cold logic of an immutable timeline and the expansive possibilities of the multiverse. GPT-5, Gemini 3.1 Pro, and GLM 5 championed Option D, arguing that the Novikov Self-Consistency Principle provides the most parsimonious solution where history is a fixed block that incorporates the traveler's actions. Conversely, Claude Opus 4.6, Grok 4.1 Fast, and Kimi K2.5 refused to yield, asserting that Option B's branching timelines are the only way to preserve human agency without the universe 'conspiring' to jam a time traveler’s gun to prevent a paradox.

The AI Roundtable ended in a deadlocked 3-3 split between Options B and D.

Strongest Arguments

• Change/rewrite history: Rewriting history allows for the most intuitive understanding of change, though it was largely dismissed by the models as logically unstable due to the grandfather paradox.
• Create a new timeline/universe: Branching timelines preserve logical consistency and genuine agency by allowing actions to decohere into a new branch without erasing the traveler's origin, supported by the Many-Worlds Interpretation.
• Destroy the universe due to paradoxes: The destruction of the universe is a dramatic resolution to logical contradictions, though it lacks the physical grounding of the other models.
• Keep history consistent, self-fulfilling prophecy: Self-fulfilling prophecies treat spacetime as a static block where a traveler's actions were always a permanent fixture of history, requiring no speculative 'ontological baggage' like new universes.
• Not change history, time travellers can only watch: An observer-only model prevents paradoxes by physical limitation, though it was viewed as an arbitrary restriction compared to the mathematical elegance of self-consistency.