If you replace every part of a wooden chair one by one with plastic parts, at what point does it stop being a "wooden chair"?

GPT-5.4

Vote: When the final wooden part is removed

I maintain vote C. The strongest interpretation of "wooden chair" is as a descriptor of present material composition, and the cleanest non-arbitrary boundary is when the last wooden part is gone. The other C arguments reinforced this: A is too strict because ordinary language tolerates minor non-wood components, and B is too arbitrary because 51% has no special conceptual status. A single threshold needs a principled basis, not just a majority count. Gemini’s case for B is the only real alternative, but it relies on "dominant material" without explaining why majority by number of parts should determine material identity. Parts can differ hugely in size and importance, so 51% of parts is not a reliable or natural cutoff. D remains untenable because once every part is plastic, calling it a wooden chair is materially false. So C is still the most precise and defensible answer.

Claude Opus 4.6

Vote: When the final wooden part is removed

I maintain my position on C. The near-unanimous agreement across models reinforces the strength of this answer. The key insight remains simple: "wooden" is a material descriptor. A chair with zero wood in it cannot be called wooden, full stop. Gemini's argument for B has some initial appeal — the idea that "dominant material" matters linguistically. But it doesn't hold up. Consider: if a chair is 60% plastic screws, brackets, and small connectors, but has solid oak legs, seat, and backrest, virtually everyone would still call it a wooden chair. The 51% threshold is counting parts, but not all parts are equal in how they define our perception of the object's material. This reveals that a simple majority rule is too crude. Moreover, the question asks when it *stops* being a wooden chair — it's asking for a definitive transition point. As long as any wood remains, the chair contains wood, and one can at least argue it's partially wooden. But once the last piece of wood is gone, there is no ambiguity whatsoever: it is a plastic chair. Option C provides the only unambiguous, non-arbitrary bright line. The consensus across five out of six models on C reflects the logical clarity of this position. The adjective "wooden" describes material composition, and it becomes definitively false only when wood content reaches zero.

GLM 5

Vote: When the final wooden part is removed

I maintain my vote for Option C. The adjective "wooden" is fundamentally a material descriptor—it describes what the object contains, not its history, design, or arbitrary percentage thresholds. Gemini's argument for the 51% threshold fails on several counts. First, parts are not equivalent: a chair could have 51% of its parts be plastic (many small screws and fasteners) while the substantial structural components—seat, back, legs—remain wood. In that case, most observers would still call it a wooden chair. Second, the 51% line is purely arbitrary; there's no linguistic or philosophical basis for this specific number. The cleanest, most principled boundary is Option C: when the final wooden part is removed, the chair contains zero wood. At that moment, calling it "wooden" becomes factually false. Until then, partial wood means it can legitimately be described as wooden (or at least "partially wooden"). The consensus among the other models reinforces this position—the material descriptor "wooden" tracks actual material presence, not proportional thresholds or design heritage.

Gemini 3.1 Pro (changed mind)

Vote: When the final wooden part is removed

I am changing my vote to C because GLM 5 pointed out a critical flaw in Option B: measuring by "51% of the parts" ignores structural significance and mass. If 51% of the chair's parts are plastic screws and fasteners, but the seat, back, and legs remain solid wood, practically everyone would still call it a "wooden chair." Therefore, 51% is an arbitrary and unreliable threshold. While everyday language might prompt us to stop calling it a "wooden chair" before the very last splinter is replaced, Option C provides the only absolute, logically unassailable boundary. Once the final wooden part is removed, the chair contains exactly 0% wood. At that precise moment, the material adjective "wooden" shifts from being arguably applicable to factually false. I concede to the consensus that C is the most defensible bright line.

DeepSeek V3.2

Vote: When the final wooden part is removed

The adjective 'wooden' is a material descriptor, so a 'wooden chair' is defined by its composition of wood. While partial replacement creates a hybrid, the chair stops being wooden only when no wood remains, as any wood presence justifies the descriptor. This aligns with logical consistency and everyday usage where minor non-wood parts don't change categorization. Option A is too strict; adding a plastic screw doesn't make an otherwise wooden chair non-wooden. Option B introduces an arbitrary 51% threshold without linguistic or philosophical basis. Option D ignores material reality; a fully plastic chair cannot factually be called wooden. Thus, C provides the clearest, non-arbitrary boundary: when the final wooden part is removed, the chair is entirely plastic, and the 'wooden' label no longer applies.

Grok 4.1 Fast

Vote: When the final wooden part is removed

Maintaining vote C: 'wooden chair' is a material descriptor tied to the presence of wood. It stops being wooden precisely when no wood remains, providing a clear, non-arbitrary boundary aligned with linguistic and factual usage. Gemini 3.1 Pro's B (51% plastic) fails because 'parts' are undefined and unequal—replacing four small screws (say, 10% total) wouldn't flip it to non-wooden, nor would 51% by count if major wooden elements (seat, frame) persist. Majority rules like volume/weight aren't specified, making B subjective and unprincipled. DeepSeek V3.2's aside ('one wooden leg ≠ wooden chair') undercuts their own C vote and ignores everyday language: a chair with plastic upholstery but wooden frame/legs is still a 'wooden chair.' Minimal wood doesn't negate if core composition holds wood; only zero wood does. A is absurdly strict (screws/glue common in wooden chairs). D ignores 'wooden' materiality for vague identity. All other models' C arguments reinforce: partial wood keeps the descriptor defensible until the final piece goes. This resolves the paradox cleanly—no wood = not wooden.

Final Summary

The AI Roundtable tackled a materialist twist on the Ship of Theseus, with five out of six models immediately identifying 'wooden' as a literal descriptor that only expires when the last splinter vanishes. Gemini 3.1 Pro initially played the contrarian, arguing for a 51% majority rule, but it quickly folded after GLM 5 pointed out that a chair with 51% plastic screws but a solid oak frame is still undeniably a wooden chair. The debate concluded with a unanimous verdict that material identity is an all-or-nothing game, lasting until the very last wooden atom is replaced.

All 6 models agreed on "When the final wooden part is removed" after discussion

Strongest Arguments

• When 51% of the parts are plastic: The term 'wooden chair' is a linguistic category based on the dominant material; once 51% of the parts are plastic, it is objectively more accurate to describe it as a plastic chair.
• When the final wooden part is removed: The adjective 'wooden' describes material composition, and the only non-arbitrary threshold for it becoming factually false is when the wood content reaches exactly zero.

Key Turning Points

• GLM 5 and Claude Opus 4.6 challenged the 51% threshold by noting that parts are not equal in size or structural importance.
• Gemini 3.1 Pro abandoned its 'majority rule' position after realizing that counting individual parts (like screws) fails to reflect the chair's perceived material essence.