In World War II, the US Air Force collected data on damage patterns in aircraft after their missions. They were trying to figure out where to put additional armor on planes so more pilots would make it back home on subsequent missions.
So where should the armor go?
A) Where the red dots are. B) Where there are no red dots.
The red dots show where planes can take damage and still make it home. The blank areas? That's where damage was catastrophic. Those planes never returned to base and couldn't be part of this data collection study.
The data doesn't show where planes were damaged. It only shows where survivors were hit.
That's (quite literally) survivorship bias.