AI summary by TwinMind
Particle Physics Discoveries that Disappeared
Sabine Hossenfelder
10m 50s
DataCamp Sponsorship
- Online learning platform for data science skills
- Interactive courses, no prior knowledge required
- Browser-based, no special software needed
- Featured courses: "Data Science for Everyone" (no coding), "Machine Learning for Everyone"
- Advanced options: Python, R, SQL skills
- Pricing: $25/month for unlimited access
- Free first chapter available via description link
Historical Context of Particle Physics Anomalies
- Higgs boson "discovered" in 1984 at Crystal Ball experiment (DESY, Germany)
- Made New York Times: "Physicists Report Mystery Particle"
- Turned out to be data fluctuation
- Actually discovered 2012 at LHC/CERN
- Supersymmetry "observed" 1984, then disappeared
- Pattern of false discoveries common in particle physics
How Particle Physics Experiments Work
- Physicists calculate expectations using best available theory
- Currently: Standard Model of particle physics
- 1984: Standard Model minus undiscovered particles
- Theory alone insufficient for measurement predictions
- Must account for experimental setup:
- Beam type and luminosity
- Detector sensitivity and configuration
- Combined theory + setup + detector = measurement expectation
- Look for deviations from expectation as evidence of new physics
Statistical Nature of Particle Physics
- Expectations always probabilistic, not deterministic
- Due to quantum indeterminism + classical uncertainty
- Possible to see signal when none exists
- Analogy: 100 random points in square divided into 4 sections
- Expect ~25 points per section
- Occasionally random distributions appear non-random
- No new physics, just noise mimicking signal
Confidence Levels and Discovery Thresholds
- Scientists assign confidence levels to observations
- Measures probability observation wasn't statistical fluctuation
- Examples:
- 1 in 20 chance = 95% confidence level
- 1 in 100 chance = 99% confidence level
- Discovery threshold (since mid-1990s): 99.99994% confidence
- ~1 in million chance of random fluctuation
- Called "five sigma" (5 standard deviations)
- Deviations attract attention below discovery threshold
Historical Examples of Disappeared Discoveries
Quark Substructure (1996)
- Quarks speculated to contain smaller particles ("prions")
- New York Times: "Tiniest Nuclear Building Block May Not Be the Quark"
- ~3 sigma significance (1 in 1000 chance coincidence)
- Same significance as current B meson anomaly
- Turned out to be statistical fluctuation
Higgs at LEP (2000)
- Large Electron Positron Collider at CERN
- Excess of Higgs-like events
- Nearly 4 sigma (1 in 16,000 chance random)
- Signal vanished
Supersymmetry Claims (2003)
- Supposed "sbottom quark" (supersymmetric partner of bottom quark)
- ~3 sigma significance
- Disappeared
Diphoton Anomaly (2015)
- Above 4 sigma significance
- Disappeared
Six Sigma Signals That Disappeared
- 1998: Tevatron "super jets" at 6 sigma
- 2004: HERA pentaquarks (5-quark particles) at 6 sigma
- Both had no known physics interpretation
Muon g-2 Anomaly
- Recently increased from 3.7 to 4.2 sigma
- Hasn't crossed 5 sigma discovery threshold
Non-Fluctuation Causes of Disappeared Discoveries
Top Quark False Discovery (1984)
- UA1 experiment at CERN saw 11 decays, expected 3.5
- Signature fit top quark expectations
- Made New York Times
- Problem: misestimated expected background events
- Top quark actually discovered 1995
CDF Collaboration (2011)
- ~4 sigma excess at Tevatron
- Not fluctuations but required better background understanding
Reasons Why Discoveries Disappear
- Fluctuations (most common)
- Miscalculations (background estimates)
- Analysis errors (creative but flawed methods)
- Systematics (detector triggers, filters, etc.)
Why Particle Physics Has Many Fluctuations
- Large datasets increase probability of finding signal-like fluctuations
- More data = higher chance of false signals
- Reason for introducing 5 sigma standard
- Prevents constant false discoveries
Current B Meson Anomaly at LHC
- Around since 2015, recently back in news
- Currently at 3.1 sigma
- Unusual that it persists rather than disappearing
- Suggests either systematic problem or real signal
