Survivorship Bias / 幸存者偏差

Author: Dexin Kong
ORCID: https://orcid.org/0009-0008-3831-5725
Structured and refined with assistance from ChatGPT
AI Automatic Translation (Unreviewed)

Background

Around 1943, during World War II, the U.S. Air Force found that a large number of bombers failed to return from missions over Europe.

To reduce losses, the military began analyzing bullet hole distributions on aircraft that successfully returned, hoping to determine which areas needed additional armor.

Mathematician Abraham Wald from Columbia University’s Statistical Research Group (SRG) reached a very different conclusion.

He argued that the areas requiring reinforcement were precisely the ones with “almost no bullet holes.”

Because the aircraft that managed to return were already “survivor samples.”

Planes hit in critical areas often never made it back at all, and therefore never appeared in the statistics.

Discussion

In the real world, similar phenomena can be found almost everywhere.

The results a system can see become increasingly alike, while the voices it can hear grow more and more uniform.

Gradually, the system can only rely on these repetitive results and unanimous voices as the basis for understanding reality.

In Unavoidable Wrong Answer, the overloaded department was dealt with.

In Replacing the Person, the difficult accountant was replaced.

In Workplace Thinking, the people who did not understand the rules were eliminated.

In The White Rabbit, the “harmful white rabbits” were gradually filtered out.

In The Unlikable One, those who were different were pushed out.

And so, “the world finally became peaceful.”

People admired each other more and more.

Reports looked better and better.

PowerPoint slides became more and more polished.

Everything appeared prosperous and thriving.

That night, Jack and Rose were drinking champagne and talking softly to each other.

The band on the deck was still playing.

The dance had not yet ended.

And ahead of them, the outline of the iceberg was slowly beginning to emerge.

(Lost at Sea Series — End)

• 1912 — RMS Titanic disaster
• 1986 — Space Shuttle Challenger disaster
• 2001 — Enron scandal
• 2008 — Lehman Brothers collapse
• 2011 — Fukushima Daiichi nuclear disaster
• 2023 — Silicon Valley Bank collapse

Note:
This project is an ongoing independent research effort developed in spare time.
Some ideas and terminology may continue to evolve over time.
The English version may contain translation imperfections or semantic deviations.
The Chinese version remains the primary semantic reference.