← What Went Wrong 1962-2025

Compare

The same sky holds three different safety records. Pick any two of the three FAR operating-rule worlds and read them side by side - cadence, per-year counts, cause distribution, phase of flight, and who walks away. Every distribution is drawn on a shared scale, so the bar lengths themselves are the comparison across the gutter. No live picker means no JavaScript: the two panels switch on pure CSS.

What this compares - and how honest each figure is

Part 121's cause and phase charts are real, tallied from the 14 landmark cases this file carries - a notable-case corpus, not the full Part 121 record (each chart says so). Part 91's are the site's illustrative published-range profiles. Part 135 carries no records yet, so its panels name the exact CAROL swap-point instead of inventing a number. Cause categories differ in kind between the worlds - GA is dominated by pilot loss of control, the airlines by machine, systems, and rare events - and the shared axis shows that contrast honestly.

Choose an operating-rule world for each panel

Panel A

Scheduled airlines

14 CFR Part 121
A few percent of the U.S. accident file

The safest way humans have ever moved. Between Colgan Air 3407 (2009) and the Reagan National midair (2025), no U.S. airliner killed a scheduled passenger for roughly sixteen years.

Fatal-accident cadence Years apart ~16-yr gap, 2009-2025 real
Accidents / year ~30 overwhelmingly minor illustrative
Rate / 100k hrs far below 0.1 per 100k flight hours illustrative
Occupant survival Most survive many events have zero fatalities real
Probable cause real - Landmark cases in this file - n=14, 1,613 fatalities
Phase of flight real - Landmark cases in this file - n=14

General aviation

14 CFR Part 91
Carries almost the entire U.S. accident burden

Private flying carries almost the entire U.S. accident burden - a steady drumbeat of roughly a thousand crashes and a couple hundred fatal ones every year, largely unchanged for decades.

Fatal-accident cadence ~200 / year fatal accidents real
Accidents / year ~1,100 all severities illustrative
Rate / 100k hrs ~5 about 1 fatal illustrative
Occupant survival Often total loss fatal accidents rarely leave survivors real
Probable cause Approx. published NTSB / AOPA ranges - share of fatal GA
Phase of flight Approx. published NTSB profile - share of all accidents

Commuter & air taxi

14 CFR Part 135

Commuter and air-taxi flying sits between the two worlds - tighter oversight than private Part 91, lighter than the scheduled airlines. This file does not yet carry its records; the panels below name the exact CAROL fields that fill them.

Fatal-accident cadence n/a awaiting CAROL
Accidents / year n/a awaiting CAROL
Rate / 100k hrs n/a awaiting CAROL
Occupant survival n/a awaiting CAROL
Probable cause
Live-data slot No Part 135 records in this file yet swap: probable_cause narrative over far_description = Part 135 (needs NLP categorization, like the GA cause file)
Phase of flight
Live-data slot No Part 135 records in this file yet swap: broad_phase over far_description = Part 135 (field is blank in post-2008 CAROL records)

Panel B

Scheduled airlines

14 CFR Part 121
A few percent of the U.S. accident file

The safest way humans have ever moved. Between Colgan Air 3407 (2009) and the Reagan National midair (2025), no U.S. airliner killed a scheduled passenger for roughly sixteen years.

Fatal-accident cadence Years apart ~16-yr gap, 2009-2025 real
Accidents / year ~30 overwhelmingly minor illustrative
Rate / 100k hrs far below 0.1 per 100k flight hours illustrative
Occupant survival Most survive many events have zero fatalities real
Probable cause real - Landmark cases in this file - n=14, 1,613 fatalities
Phase of flight real - Landmark cases in this file - n=14

General aviation

14 CFR Part 91
Carries almost the entire U.S. accident burden

Private flying carries almost the entire U.S. accident burden - a steady drumbeat of roughly a thousand crashes and a couple hundred fatal ones every year, largely unchanged for decades.

Fatal-accident cadence ~200 / year fatal accidents real
Accidents / year ~1,100 all severities illustrative
Rate / 100k hrs ~5 about 1 fatal illustrative
Occupant survival Often total loss fatal accidents rarely leave survivors real
Probable cause Approx. published NTSB / AOPA ranges - share of fatal GA
Phase of flight Approx. published NTSB profile - share of all accidents

Commuter & air taxi

14 CFR Part 135

Commuter and air-taxi flying sits between the two worlds - tighter oversight than private Part 91, lighter than the scheduled airlines. This file does not yet carry its records; the panels below name the exact CAROL fields that fill them.

Fatal-accident cadence n/a awaiting CAROL
Accidents / year n/a awaiting CAROL
Rate / 100k hrs n/a awaiting CAROL
Occupant survival n/a awaiting CAROL
Probable cause
Live-data slot No Part 135 records in this file yet swap: probable_cause narrative over far_description = Part 135 (needs NLP categorization, like the GA cause file)
Phase of flight
Live-data slot No Part 135 records in this file yet swap: broad_phase over far_description = Part 135 (field is blank in post-2008 CAROL records)
Real - tallied from this file's records Illustrative - approx. published ranges Live-data slot - awaiting CAROL

The CAROL swap-point

Every slotted figure fills the moment the NTSB CAROL export drops into src/lib/source.ts. Each row below is the field that unlocks it:

  • far_description cohort membership - the Part 121 / 91 / 135 split
  • event_year x isFatal fatal-accident cadence and per-year counts
  • broad_phase phase-of-flight distribution (blank in post-2008 records)
  • probable_cause (narrative) cause-family distribution - needs NLP over report text
  • total_fatal / injury fields occupant survival and lethality
  • FAA flight-hours (external join) true rate per 100k flight hours - not carried in CAROL

Set: the three FAR operating-rule worlds that carry U.S. civil aviation. The airline record is the real notable-case corpus behind Two Industries; the GA distributions are the illustrative profiles behind Probable Cause and Phase of Flight. See the Methodology.