Unfalsifiable, Undetected, and the JWST Surprise
Every theory needs a reliable, principled way to handle observations that contradict it. The honest way treats the observation as a genuine challenge to the theory. A less honest pattern has been standard practice in cosmology for decades, and it deserves the same scrutiny as the headline evidence for dark energy itself.
The Local Exception Defence
Andromeda, the nearest large galaxy to our own, is not receding from the Milky Way. It is approaching, at approximately 110 km/s, on a collision trajectory. This directly contradicts a model in which everything is supposed to be receding from everything else. The standard response classifies this as a "local exception": gravity briefly overpowering the universal expansion in our immediate neighbourhood. The Hubble Space Telescope and JWST have since documented hundreds of similar galaxy collisions across the observable universe, each one filed under the same exception, with the classification growing steadily less tenable as the count grows.
No spatial scale, mass threshold, or density criterion has ever been defined for where "local" ends and "universal" begins. A scientific theory must be falsifiable, meaning it must be possible, in principle, for some observation to prove it wrong, a criterion formalized by philosopher of science Karl Popper specifically to distinguish genuine scientific claims from claims that can absorb any outcome. A defence mechanism capable of absorbing any contradictory observation, indefinitely, without a defined boundary of applicability, fails that criterion by construction, regardless of how many individual predictions the surrounding theory gets right elsewhere.
Recession Faster Than Light, by Construction
Applying the Hubble law linearly, recession velocity reaches the speed of light, c, at a distance d_H = c / H0, approximately 14 billion light years, called the Hubble radius. The observable universe extends to roughly 47 billion light years. Everything beyond the Hubble radius is therefore assigned a recession velocity exceeding c.
The standard resolution, that galaxies are not moving through space faster than light but are instead carried by the expansion of space itself, which is not bound by relativistic velocity limits, was not independently derived and then confirmed by observation. It was constructed specifically to reconcile the mathematical consequence of the Hubble law with relativistic constraints, after the contradiction appeared. That is a post-hoc patch, not an independently motivated physical principle. And everything beyond the Hubble radius is, by definition, causally disconnected from us: no signal from that region could ever reach an observer here to confirm or deny any description of it. Every confident claim about that region is an extrapolation of a formula into territory the formula has never been checked against.
The Supernova Evidence, Reopened
The primary observational basis for cosmic acceleration, and therefore for dark energy, comes from Type Ia supernova observations published by Perlmutter et al. in 1999 and Riess et al. in 1998, work that was awarded the Nobel Prize in Physics. In 2019, a peer-reviewed paper by Colin, Mohayaee, Rameez, and Sarkar, published in Astronomy & Astrophysics, reanalyzed the Joint Light-curve Analysis catalogue of 740 Type Ia supernovae and found that the deceleration parameter carries a statistically significant dipole component, 3.9 sigma, aligned with the direction of the Cosmic Microwave Background dipole, which marks our own galaxy's motion through space.
*"The cosmic acceleration deduced from supernovae may be an artefact of our being non-Copernican observers, instead of evidence for a dominant component of dark energy in the Universe." - Colin, Mohayaee, Rameez &** Sarkar, 2019*
That finding has not been refuted in the peer-reviewed literature since. The observational foundation underneath one of the most celebrated results in modern cosmology, and the Nobel Prize awarded for it, rests on data that a peer-reviewed reanalysis suggests may reflect our own galaxy's local bulk motion instead of a genuine universal acceleration.
Ninety-Five Percent, Undetected
Dark energy and dark matter together are proposed to comprise approximately 95% of the total content of the universe, per Planck Collaboration 2020 figures. Neither has been directly detected. Dedicated dark matter searches, including the Large Underground Xenon experiment, the XENON programme, and the Cryogenic Dark Matter Search, have produced null results across decades of operation. Dark energy has no confirmed direct detection of any kind, of any type, at any point.
Per the Planck Collaboration's 2020 release, the universe's total energy budget is apportioned as roughly 68% dark energy, 27% dark matter, and just under 5% ordinary baryonic matter, the matter that makes up every star, planet, and observer ever measured directly. The most recent XENONnT results pushed the exclusion limit on weakly interacting massive particle dark matter candidates to cross-sections below 10^-47 square centimetres, ruling out enormous swaths of the parameter space theorists had proposed over the preceding two decades, without a single confirmed detection event anywhere in that search.
Galaxies That Shouldn't Exist Yet
JWST observations have identified massive, morphologically mature galaxies at redshifts z greater than 10, corresponding to less than 500 million years after the proposed Big Bang. These galaxies show stellar masses, star formation histories, and structural complexity inconsistent with the timescales the standard hierarchical structure-formation model allows. Labbe et al. (2023), published in Nature, identified candidate galaxies at z of roughly 7 to 10 with implied stellar masses so large relative to the survey volume that, if confirmed, they would require an implausibly large fraction of all available baryonic matter to have already converted into stars within the standard model's timeline. Multiple independent authors, including Steinhardt et al. (2016) and Boylan-Kolchin (2023), have noted that these observations directly challenge the standard model's own timeline for structure formation, describing the problem as "too much, too early, too fast" relative to what the standard model's own growth-rate calculations permit.
Specific named objects sharpen the problem further. JADES-GS-z13-0, confirmed spectroscopically in 2022 at a redshift of roughly 13.2, corresponds to an observation just 325 million years after the proposed Big Bang, a window in which the standard model expects only the earliest, smallest proto-galactic structures, not the organized stellar populations actually observed. A separate, related tension sits alongside the age problem: the S8 tension, a roughly four-standard-deviation mismatch between how much large-scale matter clustering the standard model predicts by the present day and how much is actually observed through weak gravitational lensing surveys. Structure, by several independent measurements, appears to have formed both earlier and less abundantly than the standard model's own calculations expect, a combination that is difficult to resolve by adjusting a single parameter in either direction.
The Tally
- A starting premise that logically implies the thing it denies
- A "constant" that has never once behaved like one
- An unfalsifiable local exception with no defined boundary of applicability
- A faster-than-light recession result patched by a rule invented after the contradiction appeared
- Headline supernova evidence carrying a directional bias aligned with our own galaxy's motion, unrefuted since 2019
- 95% of the universe's proposed content, undetected after fifty years of dedicated search
- Galaxies too structurally mature for the model's own permitted timeline
None of this is speculation from outside the field. Every item here is documented in the peer-reviewed literature, authored by mainstream researchers, several of them Nobel laureates, publishing in the same journals that established the results now under question.
What makes this framework of problems different from the ordinary friction any successful theory generates is where the problems sit. They are not scattered at the model's outer edges, in exotic regimes far from everyday testability. They sit at the centre: the interpretation of the single dataset used to argue for cosmic acceleration, the census of what the universe is actually made of, and the timeline for when its largest structures came to exist. A theory can absorb friction at its edges indefinitely. Friction at its centre is a different kind of problem.
The 2011 Nobel Prize in Physics was awarded specifically for the discovery of the accelerating expansion of the universe through observations of distant supernovae, a discovery treated at the time, and largely since, as one of the most secure results in modern cosmology. A peer-reviewed reanalysis calling that same dataset's central interpretation into question, without being refuted in the years since its publication, is not a minor footnote to an otherwise settled picture. It is a direct challenge to the observational foundation underneath a Nobel-recognized result, sitting unresolved in the published literature for anyone to examine.
None of this requires assuming bad faith on anyone's part. Perlmutter, Riess, and Schmidt worked with the best statistical tools available at the time, and their original conclusion was a reasonable reading of the data as it stood. The dipole signal identified in 2019 required a larger combined catalogue and a specific statistical test designed to detect exactly that kind of directional bias, tools that were not the standard approach in the original 1998 analysis. Science proceeding this way, an earlier result later shown to carry an unaccounted-for bias, is normal and healthy. What is unusual is how rarely that reopening gets mentioned alongside the original result in public explanations of why dark energy is believed to exist.
All DOIs linked below.