AI summary by TwinMind
The Biggest Physics News of 2024
Sabine Hossenfelder
10m 16s
Modified Gravity Debate (#5)
- Wide binary systems data analysis produced contradictory results on Modified Newtonian Dynamics (MOND)
- Two stars orbiting at large distances used to test main Dark Matter competitor
- MOND predicts gravity law changes from 1/R² to 1/R at certain distances
- Should affect orbital velocity in wide binaries
- Different research groups analyzed same data, reached opposite conclusions
- One author claimed MOND confirmed, another ruled it out
- First author made additional checks but results unchanged
- Two contradictory results still unresolved, debate continues into 2025
Jonathan Oppenheim's Post-Quantum Gravity Theory (#4)
- First genuinely new approach in stagnant field after 40 years
- Addresses quantum physics + gravity compatibility problem
- Revolutionary approach: gravity doesn't need to become quantum theory
- Instead, gravity needs random element - unavoidable SpaceTime shaking
- Makes gravity compatible with quantum uncertainty
- Proposes testable predictions through precise mass measurements
- Develops new mathematics combining two types of randomness
Google Quantum AI Error Correction Breakthrough (#3)
- First demonstration that error correction scales as predicted
- Key finding: adding more qubits for error correction doesn't just increase total error
- Works if qubit quality sufficient
- Tested up to ~50 qubits (long way from million needed)
- Essential for building commercially relevant quantum computers
- New chip with 105 qubits performed calculation faster than conventional computers
- Result not practically useful - just random distribution
DeepMind's Alpha Proof Mathematics AI (#2)
- Alpha Proof + Alpha Geometry reached International Mathematical Olympiad silver medalist level
- Represents neurosymbolic AI: combines deep neural networks with symbolic reasoning
- Like large language models but with enforced logic
- Timothy Gowers (Fields Medal winner) described as "definitely impressed"
- DeepMind called it "massive breakthrough" and "phase transition" in AI mathematical ability
- Steady progress over years, arbitrary to mark July 2024 as breakthrough point
- Implications: once mathematics solved, physics applications next
Dark Energy Weakening Evidence (#1)
- Evidence building that dark energy not constant but getting weaker
- Background: dark energy causes universe expansion
- Simplest type: cosmological constant (Lambda) - should be constant
- Data doesn't fit constant model well
- Supporting evidence from multiple tensions:
- Hubble tension: discrepancy in expansion rate measurements
- Sigma 8 tension: discrepancy in universe clumpiness measurements
- New 2024 data analyses:
- Dark Energy Survey: examined supernova light vs distance
- DESI experiment: studied baryon acoustic oscillations in galaxies/clusters
- Patterns from early universe sound waves in plasma
- Both found tentative evidence dark energy weakening
- Combined evidence just below 4 sigma significance
- Implications if confirmed: universe expansion might not eternally accelerate
- Could slow down or reverse instead
Channel Updates
- 2025 visual changes: less pink, more purple colors
- Otherwise content remains same
- Thanks to community, comparison to "good old blogging days"
