Here’s West Hunter with an example of how following long inductive chains can cause you to arrive at wrong conclusions that perhaps end in Iraq getting buried in a ten-foot-deep layer of white phosphorus.
I would like to supply a similar case.
In 1989, a Missouri-class battleship called the USS Iowa was test-firing its 16-inch guns. Something went wrong. As explosive charges were loaded into the breech of gun turret number two, they suddenly detonated, sending the explosion back into the turret crew. Forty-seven servicemen died in a wash of fire.
How did it happen? The gun barrel in question was cold. No cold weapon had ever caused a spontaneous explosion in all of recorded maritime history. Navy investigators found traces of brake fluid, calcium hypochlorite, and steel wool inside the barrel. The remains of a sabotage device?
The story developed an interesting Brokeback Mountain-esque winkle when it was revealed that Clayton Hartwig, captain of the centre gun, had been in a covert relationship with a sailor in the turret crew. What’s more, he’d been in charge of the loading operation. Was this an act of revenge from a jilted lover? Both men had died in the explosion.
Elaborate theories of sabotage and murder-suicide looked right past the real reason for the explosion. Additional bags of explosives called “trim bags” are normally inserted into the main charge to correct for weight variations. Unlike the explosives in the main charge, the “trim bags” are not tightly packed, making them susceptible to the shock of the gun’s power-driven rammer. The Iowa, incidentally, had a rammer that forced the bagged explosives into the breech 0.6 meters further than regulation guidelines, and with greater force. The explosive charges remain stable under heavy pressure, but the loosely-packed trim bags were very unstable under those conditions.
A second technical inquiry established that the chemical remains in the gun barrel were most likely from a mixture of cleaning fluid, lubricants, and seawater. No reason to suspect conscious sabotage.
Finally, a test rig was built that simulated the Iowa’s 16-inch cannon, as well as the over-ramming. It was done five times, ten times, fifteen times, but nothing happened.
The testers persisted. As Nassim Taleb would point out fifteen years later, you have to make outliers part of your plan. Even if cold explosions almost never happen, you cannot escape the awful tyranny of that “almost”.
On the 18th test, the charge exploded inside the cold barrel, blowing the test rig apart.