How does the solar system's Eddington parameter constraint actually force modified gravity to choose between cosmological fit and local validity?

# The Eddington Parameter Pinch: Why Modified Gravity Can't Have It All

The core tension you're asking about is real and stark: modified gravity theories that successfully explain the universe's accelerated expansion often predict different light-bending angles near massive objects than what we actually observe in the Solar System. This forces theorists into an uncomfortable choice.

The Constraint and the Problem

The Eddington parameter β (also called the post-Newtonian parameter) quantifies how much light bends as it passes the Sun compared to Einstein's prediction. General Relativity predicts β = 1; any deviation signals new physics. Solar System tests—particularly stellar light deflection and time-delay measurements—pin β to exquisite precision [AG-2025.11-224].

Here's the crunch: many modified gravity theories that were designed to mimic dark energy and fit cosmological data inherently predict deviations in β. The theory's extra degrees of freedom (torsion, non-metricity, nonlocal propagators, or scalar fields) that amplify gravity on cosmic scales inevitably change how gravity behaves locally [AG-2025.11-174]. It's not a bug you can fix with one more parameter—it's baked into the mathematical structure.

A Concrete Example: Teleparallel and Symmetric Teleparallel Gravity

Researchers explored extending non-Riemannian frameworks (which replace Einstein's curved spacetime with torsion or non-metricity) to reproduce cosmic acceleration [AG-2025.11-174]. The good news: "some models can reproduce the desired cosmological behavior." The bad news: "they often fail to meet Solar System observational bounds, particularly through deviations in the predicted Eddington parameter."

The mechanism is straightforward: the extra geometric degrees of freedom that drive late-time acceleration at ~ gigaparsec scales don't "switch off" at 1 AU—they leak into the local gravitational field, distorting the metric around the Sun in ways that contradict observations.

Nonlocal Gravity: Tuning the Leak

Nonlocal gravity (Deser-Woodard models) offers a slightly more flexible scenario [AG-2025.11-224]. These theories introduce memory effects—gravity at one point depends on its history everywhere else. This requires two parameters (ζ, b). The parameter b controls how fast nonlocal effects decay with distance; larger b suppresses them faster, weakening Solar System constraints. But here's the catch: you can't tune away the tension entirely. The perihelion advance of Mercury and the Shapiro time delay from Cassini together carve out a sharply bounded allowed region in parameter space—and there's no escape hatch for theories that need large nonlocal effects to work cosmologically.

The MOND Quadrupole Disaster

Modified Newtonian Dynamics (MOND) faces perhaps the starkest version of this problem. MOND succeeds spectacularly at galaxy rotation curves by invoking a new acceleration scale and an interpolating function [AG-2024.01-118]. But here's where it breaks: MOND predicts a quadrupole moment in the Sun's gravitational field via the "external field effect"—the Solar System is supposedly embedded in the Milky Way's acceleration field, which modifies local gravity.

Cassini radio-tracking data tightly constrains this quadrupole [AG-2026.02-407]. The result: the Solar System bound requires a much sharper transition between Newtonian and deep-MOND regimes than the galaxy kinematics allow [AG-2024.01-118]. The authors find an 8.7σ tension—the two datasets cannot be simultaneously satisfied with a single interpolating function. You can loosen one constraint only by tightening the other.

Why This Matters

This isn't a minor experimental discrepancy. It reveals a fundamental incompatibility: theories with enough extra structure to alter cosmic expansion on gigaparsec scales cannot innocuously decouple at parsec scales. The problem is structural, not accidental. As one team concludes, "the need for a unified approach that tests modified gravity theories across both cosmological and local scales" is urgent [AG-2025.11-174]. Until modified gravity passes precision tests at both ends of the length scale simultaneously—using the *same* parameters—it remains a phenomenological Band-Aid rather than a fundamental theory.

The only theories that sidestep this trap are those like the minimal fractional Newtonian model, which uses a single potential to unify both Solar System tests *and* cosmology [AG-2026.03-379]—but those remain highly speculative.