In 2012 New York City started a program for 16- to 18-year-olds at the Rikers Island jail. It was paid for with a social impact bond. Goldman Sachs invested $7.2 million and would be repaid only if reoffending fell by 8.5 percent.
In 2015 the evaluation found that the program had no effect on reoffending. The program ended. Goldman Sachs lost $1.2 million, after a guarantee from Bloomberg Philanthropies, and the city repaid nothing.
“In God we trust; all others must bring data” is often credited to the quality expert W. Edwards Deming. Quote Investigator found “no substantive support” for that. Quote Investigator found it in 1978 testimony to a House subcommittee by the pathologist Edwin R. Fisher, who already called it a cliché: “In God we trust, others must provide data.”
Ballpark figure. A number that is “approximately correct.” As a verb, to ballpark something is “to estimate (something) roughly or casually.”
Back-of-the-envelope. A calculation “done quickly to provide a rough estimate.”
Long shot. In betting, “a bet in which the chances of winning are slight but the possible winnings great.”
Safe bet. “A thing very likely to happen.”
Bet the farm on it. This is said “to show confidence in an anticipated outcome.”
During the Cuban Missile Crisis of October 1962, President John F. Kennedy put a number on the risk of nuclear war. As historian Graham Allison reports it, Kennedy judged the odds to be “between 1 in 3 and even.”
In his April 2016 letter to Amazon shareholders, Jeff Bezos explained why the company takes big risks. “Given a ten percent chance of a 100 times payoff, you should take that bet every time. But you’re still going to be wrong nine times out of ten.”
He compared it to baseball. In baseball, the best possible swing scores four runs. “In business, every once in a while, when you step up to the plate, you can score 1,000 runs.”
A buyer who borrows to buy a house in a federal flood zone receives a written notice. One federal sample notice reads: “This area has a one percent (1%) chance of a flood equal to or exceeding the base flood elevation (a 100-year flood) in any given year. During the life of a 30-year mortgage loan, the risk of a 100-year flood in a special flood hazard area is 26 percent (26%).”
The notice also says: “Federal law will not allow us to make you the loan that you have applied for if you do not purchase flood insurance.”
U.S. intelligence reports may not use words like “likely” loosely. Intelligence Community Directive 203, signed by Director of National Intelligence James Clapper on January 2, 2015, says: “For expressions of likelihood or probability, an analytic product must use one of the following sets of terms.”
Each term has a range. “Almost no chance” means 1 to 5 percent. “Unlikely” means 20 to 45 percent. “Roughly even chance” means 45 to 55 percent. “Likely” means 55 to 80 percent. “Almost certain” means 95 to 99 percent.
In 2010 the CIA told President Obama that Osama bin Laden might be living in a compound in Abbottabad, Pakistan. In the spring of 2011, officials gave him numbers. “Estimates ranged from a low of 30 or 40 per cent to a high of 95 per cent.”
Obama later said on 60 Minutes: “At the end of the day, this was still a 55-45 situation. I mean, we could not say definitively that bin Laden was there.” He ordered the raid.
In September 1973 a Ford memo set $200,000 per death, $67,000 per injury and $700 per vehicle against the cost of an $11 safety fix. Total benefit: $49.5 million. Total cost: $137 million.
In 1977 the magazine Mother Jones made the memo famous in a story about the Pinto. But the memo was not about the Pinto. It concerned a proposed federal rule for all 12.5 million cars and light trucks sold each year, and the Pinto is not mentioned. The $200,000 was the federal traffic safety agency’s own estimate of the cost to society of one traffic death.
In Grimshaw v. Ford, the best-known Pinto lawsuit, the judge did not admit the memo as evidence.
After the Challenger accident in 1986, Richard Feynman found that people at NASA disagreed about the odds of losing a Shuttle. “The estimates range from roughly 1 in 100 to 1 in 100,000. The higher figures come from the working engineers, and the very low figures from management.” At 1 in 100,000, he noted, “one could put a Shuttle up each day for 300 years expecting to lose only one.”
His appendix to the presidential commission’s report ends: “For a successful technology, reality must take precedence over public relations, for nature cannot be fooled.”