Scientific Phenomena Defy Logic

The universe is expanding. That’s been known since Hubble’s observations in 1929. What should, by now, be a settled number — the rate of that expansion, called H₀ — is turning out to be one of the most contested measurements in all of physics.

Two methods exist for measuring H₀. The first uses the cosmic microwave background — light left over from the early universe — analyzed by the Planck satellite. It gives 67.4 km/s/Mpc. The second, called the distance ladder, builds up calibrations using Cepheid variable stars and Type Ia supernovae as cosmic mile-markers. The most precise distance-ladder measurement, from the SH0ES collaboration, gives 73.0 km/s/Mpc. These two numbers disagree by roughly 9% — and as of 2025, the discrepancy has crossed 6σ statistical significance.

A 6σ result means there’s less than a one-in-a-billion chance this is a statistical fluke. The most obvious alternative — that Cepheid measurements were corrupted by dust or stellar crowding — was directly tested by the James Webb Space Telescope in 2024–2025. JWST’s resolution confirmed the Hubble Space Telescope Cepheid data was accurate. Systematic error in the local measurement is no longer the leading candidate.

The European Commission found this important enough to award a €12 million Synergy Grant to the RedH0T project — specifically to investigate whether the standard cosmological model (ΛCDM) needs revision. The September 2025 “Tensions in Cosmology” conference in Corfu confirmed the problem has only deepened with new high-precision data from DESI, JWST, and ACT combined.

What makes this especially uncomfortable is what a resolution might require. If neither measurement is wrong, the standard cosmological model — the ΛCDM framework that’s been the foundation of cosmology since the 1990s — may need fundamental revision. That could mean dark energy isn’t a cosmological constant, that there’s new early-universe physics we’ve missed, or something stranger still.

Muons are essentially heavy electrons — 207 times more massive, with a lifespan of about 2.2 microseconds. Like electrons, they have a tiny internal magnet. The Standard Model predicts exactly how strongly that magnet should interact with quantum fluctuations in the surrounding vacuum. The prediction is extraordinarily precise — one of the most accurate in all of science.

The problem: Fermilab’s Muon g-2 experiment released its third and final measurement on June 3, 2025, achieving a precision of 127 parts per billion — surpassing the experiment’s original design goal set in 2012. The measured value of the magnetic anomaly is aμ = 0.001 165 920 705 ± 0.000 000 000 114 (stat.) ± 0.000 000 000 091 (syst.). The experimental result is consistent across all six years of data collection, which ended in July 2023.

Here’s where it gets complicated. Against the 2020 Standard Model theory prediction (compiled by the Muon g-2 Theory Initiative), the experimental result shows a 5.1σ discrepancy — almost certainly pointing to new physics beyond the Standard Model. But newer theoretical calculations using lattice QCD methods, published by the Budapest-Marseille-Wuppertal collaboration, come closer to the experimental value. That’s not resolution — it’s a new problem: two legitimate theoretical approaches now disagree with each other, meaning physicists aren’t entirely sure what the Standard Model actually predicts for this quantity.

“The anomalous magnetic moment of the muon is one of the most precisely measured quantities in physics, and it is great to see an experiment come to a definitive end with a precision measurement.” — Regina Rameika, U.S. DOE Associate Director for High Energy Physics, June 2025

The experiment is finished. The theoretical argument continues. Either way, something in our understanding of particle physics is off.

Big Bang nucleosynthesis — the process by which hydrogen, helium, and trace elements formed in the first few minutes after the Big Bang — is one of the most precisely tested theories in cosmology. Its predictions for hydrogen, deuterium, and helium-4 match observations beautifully. Lithium-7 does not.

The theory predicts roughly three to four times more lithium-7 than we actually observe in the oldest, most metal-poor stars in the galaxy’s halo — a mismatch of 4–5σ (Fields 2011, Annual Review of Nuclear and Particle Science). What’s telling is that those same ancient stars — the “Spite Plateau,” named after the astronomers François and Monique Spite who identified the pattern in 1982 — show a remarkably consistent lithium abundance across a wide range of temperatures and metallicities. That consistency rules out the simplest astrophysical explanations: if stars were just burning off their lithium through normal stellar processes, you’d expect more scatter.

Proposed solutions have included new nuclear reaction rates, stellar diffusion models that sequester lithium below the photosphere, and various exotic physics scenarios involving particle decays in the early universe. None has achieved consensus. The problem has been open for over 40 years. Deuterium and helium-4 predictions remain in excellent agreement with observation — which is why a straightforward rejection of Big Bang nucleosynthesis isn’t the answer either. Something specific to lithium-7 is different, and nobody is sure what.

Quantum entanglement isn’t a mystery in the sense that we can’t demonstrate it. We’ve demonstrated it thousands of times. The mystery is deeper: there’s no mechanism for it that doesn’t break something fundamental in our understanding of physics.

When two particles are entangled, measuring one instantly correlates with the other — regardless of distance. Einstein called this “spooky action at a distance” and spent years arguing it couldn’t be real, that there must be “hidden variables” carrying information between particles. Bell’s 1964 theorem showed that any hidden-variable theory imposes a mathematical limit on how correlated entangled particle measurements can be. Quantum mechanics predicts that limit is violated.

Experiment has repeatedly confirmed the violations. More importantly, the 2015 Delft experiment by Hensen et al. (Nature, 2015) was the first loophole-free Bell test — closing both the detection loophole and the locality loophole simultaneously. The result was unambiguous: nature violates Bell inequalities. Hidden local variables don’t exist. Reality is, in some precise technical sense, nonlocal.

What remains completely unexplained is why. Quantum mechanics provides the mathematical framework to calculate entanglement probabilities with extraordinary accuracy. It doesn’t explain what’s actually happening, or how a measurement here can be correlated with a measurement there instantaneously without any signal passing between them. The “no-communication theorem” prevents this from being used to transmit information faster than light — but that restriction itself requires explanation, not just assertion. Multiple interpretations of quantum mechanics exist (Copenhagen, many-worlds, pilot wave, relational QM), and they give genuinely different answers to what’s going on. None is provably correct.

Fast Radio Bursts are among the most energetic phenomena in the observable universe. In a few milliseconds, some emit as much energy as the Sun produces in days — entirely in radio waves, from sources billions of light-years away. The CHIME radio telescope in British Columbia has catalogued over 500 of them since 2018, including roughly 25 that repeat.

The leading hypothesis is magnetars — neutron stars with extraordinarily powerful magnetic fields. This became more plausible in April 2020, when astronomers detected an FRB-like burst from SGR 1935+2154, a known magnetar within our own galaxy — the first FRB with a confirmed source. That’s a significant step. The problem is that SGR 1935+2154 was, at that moment, orders of magnitude less energetic than the most powerful extragalactic FRBs. Whether a magnetar can produce the full range of observed FRB energies, and why some repeat while others apparently don’t, has no agreed answer.

The repeating FRBs complicate things further. FRB 121102A — the first confirmed repeater — has shown periodic activity windows and complex sub-millisecond structure within individual bursts, suggesting something more organized than a single catastrophic event. The morphology of its radio pulses doesn’t cleanly fit magnetar models as currently formulated.

In 1966, Greisen, Zatsepin, and Kuzmin independently predicted that cosmic ray protons above about 5 × 10¹⁹ electron volts would lose energy rapidly by interacting with photons in the cosmic microwave background. This means that any cosmic ray arriving at Earth above that energy — the “GZK limit” — must have originated within roughly 160 million light-years. Beyond that range, it simply can’t survive the journey at that energy.

The Pierre Auger Observatory in Argentina has detected cosmic rays above this threshold. Some significantly above it. The problem is that, when you scan the sky in the direction those particles came from, there aren’t obvious sources within the required distance that could plausibly accelerate particles to those energies. The particles are real — detected by an array of 1,600 water-Cherenkov detectors covering 3,000 km² — but their origin remains unclear.

Several explanations have been proposed: heavy nuclei rather than protons might behave differently; there might be exotic acceleration mechanisms in nearby active galactic nuclei; or — more speculatively — there might be new physics at extreme energies. Auger’s composition data suggest the particles are heavier than protons, which changes the energy-loss calculation somewhat. But it doesn’t fully resolve where they came from or how they got there.

This one sits in a different category from the others. The Hubble tension is an anomaly in a number. The hard problem of consciousness is an anomaly in the very structure of scientific explanation.

Neuroscience has made genuine progress on what David Chalmers called the “easy problems” of consciousness — explaining attention, memory formation, behavioral control, the neural correlates of various conscious states. Functional explanations are possible. What remains completely opaque is why any of this is accompanied by subjective experience at all. Why does processing information feel like something? Why isn’t it all just information processing in the dark, with no inner light?

This isn’t a question that better brain scanning will resolve. The challenge is explanatory, not empirical: no account of neural firing patterns or computational processes seems, even in principle, to explain why there is “something it is like” to be a conscious entity. The explanatory gap between objective, third-person descriptions of physical processes and first-person subjective experience has resisted every attempt to close it since Chalmers formalized the problem in 1995.

Proposed solutions range from illusionism (consciousness is a sophisticated cognitive illusion — but then who’s being fooled?) to panpsychism (consciousness is a fundamental feature of the universe, like mass or charge) to integrated information theory and global workspace theory, neither of which fully closes the gap. As of 2026, the debate among philosophers and neuroscientists is not converging.