Ephemera

My father, a railroad man, spent most of his life investigating transportation accidents and potentially fraudulent claims. One of my cousins managed development of on-board software for several Apollo missions. These influences may account for some of the selection bias evident below.

Horror Stories

Dustin Self was 19 years old. He had read Into the Wild, and was eager to test himself against wilderness. He bought a knife, tent, and sleeping bag, and "read a lot about how to survive." After he had driven from his native Oklahoma to the high desert of eastern Oregon, his GPS navigation software told him the most direct route to the town of Lakeview followed a dirt road over Steens Mountain. Self trusted his software. One month later, a ranch foreman found Self's pickup truck, which had slid off the muddy jeep track. Eighteen months later, a hunter found what was left of Dustin Self. Instead of walking two and a half miles back down the road, he had walked seven miles towards wilderness. Why? Possibly because his GPS software had told him to go over Steens Mountain.
A GPS device told a 28-year-old woman to drive her vehicle into a remote part of Death Valley. She survived, but her 11-year-old son didn't.
The Therac-25 accidents of 1985-87 killed three people and injured at least three others. Two causal factors were "overconfidence in software" and "confusing reliability with safety".
The Mars Climate Orbiter was lost on 23 September 1999 because part of its software did not satisfy its specification. This error cost about $125 million. According to an article in IEEE Computer (no longer online), this problem could have been corrected during the flight had JPL managers been more willing to consider the possibility of errors in software.
A software problem that cost $1.2 billion dollars: the destruction of a Titan 4 rocket and its payload, 30 April 1999.
Ariane 5: the destruction of Flight 501, 4 June 1996.
A pseudoscientist demonstrates the Dunning-Kruger effect.
The airline pilot who thinks I'm a "paid Pentagon blogger". Update: That same pilot, Rob Balsamo, now claims to have mathematical proof Flight 77 could not have hit the Pentagon on 11 September 2001, but (oops!) his math is wrong.
The civil engineer and 9/11 conspiracy theorist who denied the validity of his own data because his data contradict his paper based on that data.
The music critic who sincerely believes that 112 equals 126.
Technical risks of shipboard systems.
Why you should check your bags when travelling through Des Moines or San Francisco.
The asteroid is coming.

Success Stories

Einstein's general theory of relativity implied the family of FLRW models, which led Georges Lemaître to predict an expanding universe and what is now known as big bang cosmology. After the expanding universe was confirmed, others predicted the cosmic microwave background (CMB) radiation. Precise measurements of the CMB show nearly perfect agreement with the predicted isotropic black-body spectrum.
How the on-board shuttle group develops software.
Not everyone views the on-board shuttle software as a success. For an opposing view, read "Mob Software: The Erotic Life of Code" by Richard P. Gabriel and Ron Goldman.
ACL2 is one of the great success stories of computer science. (In 1975, as an undergraduate at the University of Texas, I used an early version of the Boyer-Moore theorem prover in a graduate course taught by Woody Bledsoe.)
An Australian programmer named Warren Stutt recovered the missing four seconds of data recorded at the end of Flight 77. (All of Rob Balsamo's calculations had been based on Rob's insistence, in the face of all evidence, that no seconds were missing. Oops.)
No list of success stories would be complete without mentioning the 43rd President of the United States.