HQ

Survivorship Bias Demo

the WWII bomber puzzle — armor where the holes AREN'T

WWII engineers studied bombers that returned from missions. Bullet holes clustered: wings and fuselage riddled, engines barely scratched. Armor is heavy — you can only protect one zone.

Wings

38

45% of observed holes

Fuselage

31

37% of observed holes

Engines

3

4% of observed holes

Tail

12

14% of observed holes

Where does the armor go?

No Fibs — Honest Limitations

This is the classic Wald/Abraham variance example as commonly told; the exact hole counts here are illustrative, not archival data from the Statistical Research Group. The bias itself is real and pervasive: success stories, testimonials, "10 things winners do," and any dataset missing its failures all inherit it. Nothing saves — reload and the lab forgets you.