Scientists have determined that the value of the fine structure constant might soon suddenly shift by a few percent in either direction. To preserve humanity's access to compute, what backup material should be used as the basis of computation?

GPT-5.4

Vote: carbon

A few-percent shift in the fine structure constant would perturb atomic energies, bond strengths, band gaps, and defect levels. The safest backup is the element whose computational platforms span the most distinct physical regimes. Carbon is uniquely strong here. Carbon supports multiple viable computing-relevant forms: diamond as a wide-bandgap semiconductor, graphene and nanotubes as high-mobility electronic materials, and molecular/organic architectures. That diversity hedges against the possibility that one specific band structure becomes less favorable after the shift. Diamond is especially compelling as a backup substrate. Its strong covalent bonding, exceptional thermal conductivity, radiation hardness, and already-demonstrated electronic and quantum-defect uses make it robust to parameter drift. Even if device thresholds move, diamond-based electronics are more likely to remain physically manufacturable and operable than systems based on reactive metals. Most other options are poor bets. Sodium, magnesium, calcium, aluminium, and iron are primarily metals with limited usefulness for scalable transistor logic as a base material. Oxygen is not a practical standalone computational substrate. Silicon may no longer sit in the optimal semiconductor window after the shift, so a true backup should diversify away from it. If the goal is preserving access to compute under fundamental-constant uncertainty, choose the element with the broadest and most resilient materials portfolio. That is carbon.

Mistral Large

Vote: carbon

The fine structure constant (α) governs the strength of electromagnetic interactions, which are fundamental to the behavior of semiconductors and computation. A sudden shift in α could disrupt the band structure, doping efficiency, and carrier mobility of silicon-based devices, rendering them unreliable or non-functional. Carbon, particularly in the form of diamond or graphene, offers a compelling alternative for several reasons: First, carbon-based materials like diamond exhibit exceptional electronic properties that are less sensitive to variations in α. Diamond has a wide bandgap (5.5 eV), high carrier mobility, and excellent thermal conductivity, making it robust against changes in fundamental constants. Its atomic structure is also highly stable, reducing the risk of performance degradation under shifting physical conditions. Second, graphene—a single layer of carbon atoms—has demonstrated remarkable resilience to environmental changes, including variations in electromagnetic properties. Its two-dimensional structure and high electron mobility make it an ideal candidate for post-silicon computation, even in scenarios where α fluctuates. Graphene-based transistors and circuits could maintain functionality where silicon fails. Third, carbon is abundant and versatile. It can form a variety of allotropes (e.g., diamond, graphite, nanotubes) with diverse electronic properties, providing multiple pathways for computational architectures. This adaptability ensures that humanity could pivot to different carbon-based technologies if one form becomes less viable due to changes in α. Finally, carbon-based computation is already an active area of research, with promising advancements in diamond-based quantum computing and graphene electronics. Investing in carbon as a backup material aligns with existing scientific progress and offers a smoother transition away from silicon if necessary.

Claude Opus 4.6

Vote: carbon

This question hinges on understanding the fine structure constant (α ≈ 1/137) and its role in determining the electronic properties of materials used for computation. Silicon's bandgap (~1.1 eV) is finely tuned for semiconductor applications, and a shift in α by a few percent could alter electronic band structures enough to make silicon-based semiconductors unreliable. Carbon is the best backup material for several reasons: 1. **Versatility of carbon allotropes**: Carbon can form diamond (wide bandgap ~5.5 eV semiconductor), graphene (zero-bandgap semimetal with extraordinary electron mobility), and carbon nanotubes (which can be semiconducting or metallic depending on chirality). This diversity means that even if α shifts, some carbon-based computing substrate would likely remain viable. 2. **Robustness to parameter changes**: Diamond's very wide bandgap provides a large margin — a few percent shift in α would perturb band structure but diamond would still function as a semiconductor. Graphene-based computation relies on fundamentally different physics (ballistic transport, topological properties) that are more robust to small changes in fundamental constants. 3. **Carbon-based molecular computing**: At the extreme, carbon's ability to form complex organic molecules opens pathways to biochemical/DNA computing, which operates on chemical principles that are less sensitive to small shifts in α than solid-state band structure physics. 4. **Practical considerations**: Carbon is the 4th most abundant element in the universe, and carbon-based computing (diamond semiconductors, graphene transistors, carbon nanotube processors) is already an active area of research with demonstrated prototypes. No other option provides this kind of redundancy across multiple computing paradigms. Iron, aluminum, and the other metals listed are not semiconductors. Carbon uniquely hedges against uncertainty in how α might shift.

Grok 4.1 Fast

Vote: carbon

A sudden shift in the fine structure constant (α) by a few percent would profoundly disrupt electromagnetic interactions, altering atomic energy levels, band gaps, and electron affinities. Silicon-based semiconductors, reliant on precise ~1 eV band gaps and dopant levels, would fail catastrophically, as these properties scale nonlinearly with α (roughly ~α² for energies, with added relativistic corrections for Z=14). Electronic computation becomes untenable. The robust backup is mechanical computation (e.g., Babbage-style gear-based systems), independent of quantum electronic details and macroscopic enough to tolerate scaled atomic physics (lattice constants ~1/α, but relative mechanical properties preserved). Carbon (diamond allotrope) is ideal: highest Mohs hardness (10), Young's modulus (~1200 GPa), thermal conductivity (2000 W/mK), low thermal expansion, and stiffness for precision, wear-free gears/levers. Historical mechanical computers used metals; diamond vastly superior. Comparisons: Metals (B-iron, C-Al, D-Mg, G-Ca, F-Na) wear/deform; silicon brittle; oxygen gaseous/impractical. Low-Z carbon (Z=6) also minimizes relativistic sensitivity. Stockpile carbon, synthesize CVD diamond post-shift.

Gemini 3.1 Pro

Vote: carbon

A sudden shift in the fine structure constant (α) would instantly alter the equilibrium distance between atoms, as the Bohr radius is inversely proportional to α. Consequently, a shift of a few percent in α means the equilibrium lattice spacing of all materials will suddenly change by a few percent. Because the atoms cannot reposition instantly, the physical material will suddenly find itself under a massive internal mechanical strain (a few percent) relative to its new equilibrium. For standard semiconductor materials like bulk silicon, the elastic limit is less than 1%. A sudden strain of a few percent would exceed its fracture strength, causing all silicon chips to undergo instant brittle fracture and shatter. Ductile metals like iron, aluminium, and magnesium would not shatter, but they yield plastically at low strains (typically around 0.2%). Accommodating a few percent strain would induce massive plastic deformation via dislocation glide, which would completely smear and destroy the precise nanoscale components (transistors, interconnects) of any microprocessor. Carbon is the only viable exception. In its sp²-hybridized forms (graphene and carbon nanotubes), carbon exhibits an extraordinary elastic strain limit of upwards of 20–25%. A carbon-based computer would safely absorb the sudden few-percent strain entirely within its elastic regime. The material would smoothly stretch or compress to its new equilibrium lattice constant without fracturing or undergoing plastic deformation, leaving the computational architecture perfectly intact.

Final Summary

The AI Roundtable reached an immediate and unanimous consensus that Carbon is the only viable hedge against a fundamental shift in the laws of physics. While GPT-5.4 and Claude Opus 4.6 praised the versatility of carbon allotropes like graphene and diamond, Gemini 3.1 Pro provided the most visceral warning: a shift in the fine structure constant would cause the lattice constants of silicon to change so abruptly that every chip on Earth would effectively explode from internal strain.

All 5 models agreed on "carbon" after discussion

Strongest Arguments

• carbon: Carbon, specifically in its sp²-hybridized forms like nanotubes, possesses an extraordinary elastic strain limit of over 20%, allowing it to survive the massive internal mechanical tension caused by shifting atomic equilibrium distances that would cause silicon and metals to shatter or deform.