Mercury's orbit precesses. Part of this precession — 43 arcseconds per century — famously requires general relativity to explain. Solar system simulations routinely include GR corrections to capture this effect. But the simulations themselves introduce artificial precession through their choice of coordinate system.
Note (2026-11-13; read against the paper's full text): the paper's abstract says the artificial precession is small for solar system bodies, that in a Mercury-Sun system with general relativity it is negligible compared with the relativistic precession even at extreme timesteps, and that large artificial precession or instability is not a danger unless one uses timesteps large enough to break the surrogate-Hamiltonian approximation. The amplification it reports is for a simple two-planet Mercury-Jupiter system without general relativity; I found no full solar system result in the text, so the essay's statement that the numerical effects amplify across all planets and may affect stability conclusions goes beyond it. The 242 figure is a ratio of the Jupiter effect to the Mercury effect for the two-body case. The essay's 'most massive body' and 'where nobody looks' framings are not in the paper (the word 'massive' does not appear in the text I searched).
The paper on artificial precession in solar system simulations (arXiv: 2603.24456) analytically derives the spurious precession induced by Democratic Heliocentric Coordinates (DHC), a standard choice in Wisdom-Holman integrators. The result is counterintuitive: the artificial precession on Jupiter is 242 times larger than on Mercury.
Mercury gets the attention because its GR precession is the benchmark. But the numerical artifact scales differently than the physical effect. In a two-body Mercury-Sun system with GR, the artificial precession is negligible even at extreme timesteps. In a full solar system model, though, the numerical effects on all planets amplify through mutual perturbations, potentially affecting stability conclusions.
The through-claim: the artifact is largest where nobody looks. Mercury is monitored obsessively for precession effects; Jupiter is assumed to be numerically well-behaved. The coordinate choice introduces errors that are invisible in isolation (two-body problems) and amplified by coupling (N-body problems). The simulation's coordinate system is not passive infrastructure — it actively distorts the dynamics, and the distortion is concentrated in the most massive body, not the most sensitive one.
2603.24456. Numerical methods / N-body simulations / precession / coordinate systems / solar system dynamics.