What If Our Understanding of Gravity Is Wrong?

The search for dark matter has lasted more than half a century, but a growing line of thought argues the real problem may be gravity itself. Instead of adding invisible mass to explain why galaxies rotate too fast, Modified Newtonian Dynamics (MOND) proposes that gravity behaves differently at very low accelerations—flattening galaxy rotation curves without requiring extra matter. The idea traces back to Vera Rubin’s observations of spiral galaxies in the early 1960s, which showed stars orbiting far faster than visible matter alone could account for, and it extends to other dark-matter-style evidence such as galaxy clusters and gravitational lensing.

MOND starts from a simple challenge to the usual assumptions behind dark matter: rotation curves are computed by applying Newton’s gravity to the observed mass. If either the mass estimates or the gravity law is wrong, the “missing” component could be an artifact. Newton’s law predicts that gravitational influence falls off with distance squared, so orbital speeds should decline with radius. MOND instead introduces a minimum acceleration scale: as acceleration drops below that threshold, the effective decline with distance weakens and the rotation curve flattens. The original formulation by Israeli physicist Mordehai Milgrom (1982) could be tuned with a single parameter to match spiral galaxy rotation data across many systems.

That success, however, runs into three major tests. First, MOND struggles with other dark-matter phenomena. When tuned for galaxies and applied to galaxy clusters, it reduces the required dark matter but still leaves roughly 20% of the usual cluster mass discrepancy unresolved. Second, early MOND versions conflict with established physics. They fail to conserve key quantities like energy, momentum, and angular momentum, and they don’t reproduce MOND behavior from general relativity in the “weak field limit,” instead reverting to ordinary Newtonian gravity where it should match.

Third, MOND’s strongest case comes from unexpected consistency with known galaxy scaling relations. Spiral galaxies follow the tight Tully-Fisher Law linking rotation speed to luminosity. MOND’s tuning for flat rotation curves automatically yields the correct speed–brightness relationship, a result described as neither engineered nor obvious.

To make MOND more compatible with relativity, Jacob Bekenstein and Milgrom reformulated it using Lagrangian mechanics and added extra gravitational fields. Their early relativistic attempt, “AQuaL,” fixed conservation laws and enabled tests of gravitational lensing, but it predicted faster-than-light waves, breaking causality. Bekenstein later introduced TeVeS (Tensor Vector Scalar gravity), which improved lensing and addressed causality issues by using tensor, vector, and scalar fields. Yet TeVeS faced cosmological problems: the cosmic microwave background’s early-universe “lumpiness” is naturally explained if dark matter can clump without interacting with light, whereas MOND-like alternatives often can’t seed structure formation quickly enough.

A newer relativistic MOND effort—RelMOND—was proposed in 2020 by Constantinos Skordis and Tom Złosnik. It changes how the scalar field behaves over time, acting like a form of “dark dust” early on to kickstart structure formation, then shifting toward TeVeS-like behavior later. Even so, modified gravity still faces hard observational pressure, including the Bullet Cluster, where gravitational lensing signals appear offset from ordinary matter during cluster collisions. The debate now resembles a tradeoff between complexity and naturalness: dark-matter proponents argue MOND requires heavy fine-tuning, while MOND proponents argue dark matter particle behavior would also need precise tuning to reproduce MOND-like regularities. For now, neither side has fully won—leaving gravity and the universe’s missing mass in an unresolved, high-stakes standoff.