Get AI summaries of any video or article — Sign up free
What are the Odds of Dying an Unfortunate Death? thumbnail

What are the Odds of Dying an Unfortunate Death?

Second Thought·
4 min read

Based on Second Thought's video on YouTube. If you like this content, support the original creators by watching, liking and subscribing to their content.

TL;DR

Surviving aircraft data is conditional on not being lost, so damage patterns among returnees can’t be used to infer which damage is most lethal.

Briefing

The central takeaway is that “where the bullet holes are” can mislead: most planes returning with damage show hits in wings and fewer in engines, but that pattern doesn’t mean engine strikes are rare—it reflects survivorship bias and conditional probability. Planes that take severe engine damage and catch fire are far less likely to make it back to base, so the “survivor” sample overrepresents aircraft with survivable damage locations. That means the most effective reinforcement isn’t where the visible damage on returning planes appears most common; it’s where damage most often proves fatal.

From there, the discussion pivots into a tour of unlikely death risks, using long-run odds to recalibrate everyday fears. Being struck and killed by an asteroid is estimated at about 1 in 74,817,414 per day, and the annual “91 people” figure is clarified as an actuarial projection rather than a literal count from impacts. Freak fireworks accidents land at roughly 1 in 50,729,141 per year, with 2016 reporting 11,000 fireworks-related injuries and four deaths. Fear-driven examples follow: dying from a wasp or bee sting is about 1 in 25,364,571, with nearly 100 deaths attributed to bee and wasp stings in the U.S. each year; plane crashes are around 1 in 11 million; and lightning is about 10 million to 1—still rare, but not zero. The transcript also highlights demographic differences, noting men are killed by lightning about four times as often as women.

Several risks are framed as counterintuitive because they don’t “feel” dangerous. Hot-water scalding is estimated at 5 million to 1, and the risk is elevated because it often involves children and everyday household conditions. Another surprising statistic: about 2,500 left-handed people worldwide die each year from using equipment designed for right-handed users, with right-handed power saws identified as a common culprit. Transportation comparisons sharpen the contrast between perceived and actual risk: train crashes are about 500,000 to 1, while car driving is far more lethal at about 8,000 to 1. The global baseline is stark—around 1.3 million road deaths per year, roughly 3,300 per day—making lightning and plane crashes look even less likely by comparison.

The concluding logic ties the entire odds list back to the opening question. When only the planes that return are counted, rare but catastrophic outcomes are systematically excluded. That’s survivorship bias: treating the frequency of damage among survivors as if it represented the frequency of damage among all aircraft. The practical implication is to reinforce based on what prevents loss, not what survives to be observed—an approach that depends on conditional probability rather than surface-level patterns.

Cornell Notes

The transcript argues that “damage location” on surviving aircraft can’t be used directly to estimate which threats are most dangerous. If planes with engine hits rarely return, then the low number of engine bullet holes among returning planes reflects survivorship bias, not safety. The discussion then lists odds for various causes of death—from asteroids and fireworks to bee stings, plane crashes, and lightning—showing how rare many feared events are. It also includes counterintuitive risks like scalding from hot water and injuries from right-handed tools affecting left-handed users. Finally, it compares transportation risks, emphasizing that everyday car travel is vastly more dangerous than most people assume.

Why does “most bullet holes are in the wings” not automatically mean wing hits are the main danger?

Because the dataset is conditional on survival. Only planes that make it back to base are observed, so aircraft that suffer fatal outcomes—especially severe engine damage and fires—are missing from the sample. That’s survivorship bias: the observed distribution among survivors is not the same as the distribution among all planes that took hits.

How do the odds for asteroid impacts illustrate the difference between daily probability and annual death counts?

The daily odds of being struck and killed by an asteroid are given as about 1 in 74,817,414. The transcript notes that a seemingly large annual number (e.g., “91 people”) is an actuarial projection based on impact likelihood and expected casualties per event, not a literal count from a known number of impacts.

What makes lightning a useful comparison point for fear calibration?

Lightning is described as about 10 million to 1, which is still extremely unlikely, yet it kills people each year. The transcript cites U.S. deaths of 39 in 2016 and 14 so far in 2017, and adds a gender disparity: men are killed by lightning about four times as often as women.

Which risks are highlighted as counterintuitive because they don’t look inherently dangerous?

Hot-water scalding is one example: it’s estimated at 5 million to 1, but it’s common enough to cause hundreds of deaths annually, with 150 deaths per year in Japan alone. Another is left-handed people dying from right-handed equipment—about 2,500 deaths per year—often involving right-handed power saws that jump forward.

How does the transcript quantify the gap between perceived and actual transportation risk?

Train crashes are estimated at about 500,000 to 1, while car driving is far more lethal at about 8,000 to 1. The global road-death baseline is given as roughly 1.3 million deaths per year, around 3,300 per day, making rare events like plane crashes or lightning look comparatively less likely.

Review Questions

  1. In the aircraft example, what specific statistical condition creates survivorship bias?
  2. Pick two “feared” causes of death from the list (e.g., lightning, plane crash, bee sting) and compare their stated odds; what does that comparison imply about fear versus risk?
  3. Why does the transcript treat “returning planes” as a misleading sample for estimating the danger of engine hits?

Key Points

  1. 1

    Surviving aircraft data is conditional on not being lost, so damage patterns among returnees can’t be used to infer which damage is most lethal.

  2. 2

    Engine hits may appear rare among returning planes because many engine-damage cases never make it back, a classic survivorship bias problem.

  3. 3

    Odds for rare events (asteroids, plane crashes, lightning) are presented to recalibrate fear using long-run probability estimates.

  4. 4

    Some high-impact risks are counterintuitive—hot-water scalding and right-handed tools harming left-handed users are both highlighted with specific odds and death counts.

  5. 5

    Transportation risk varies dramatically: car travel is far more dangerous than train travel, and both are far more common sources of death than many feared events.

  6. 6

    Demographic differences matter for risk estimates; men are cited as about four times more likely than women to be killed by lightning.

  7. 7

    Actuarial projections can differ from intuitive “how many people died” expectations, as shown in the asteroid discussion.

Highlights

The low number of engine bullet holes among returning planes doesn’t mean engine hits are unlikely—it signals that engine damage is often fatal and therefore underrepresented among survivors.
Lightning is framed as “gold standard” unlikely (about 10 million to 1) yet still produces measurable annual deaths, with men affected about four times as often as women.
Hot-water scalding and left-handed tool accidents are presented as surprisingly lethal risks precisely because they’re everyday and easy to underestimate.
Car deaths dwarf most other risks: roughly 1.3 million road deaths per year worldwide, about 3,300 per day.

Topics

Mentioned