Ordinary claims require ordinary evidence

A corol­lary to ex­traor­di­nary claims re­quire ex­traor­di­nary ev­i­dence is “or­di­nary claims re­quire only or­di­nary ev­i­dence”noteAt­tributed to Gw­ern Bran­wen, who hasn’t pop­u­larized this claim nearly as much as Carl Sa­gan pop­u­larized the ex­traor­di­nary ver­sion.

This corol­lary is prob­a­bly more im­por­tant in prac­tice, as, in our day-to-day lives, we en­counter more or­di­nary claims than ex­traor­di­nary claims, but may be more tempted to re­ject many of those or­di­nary claims by de­mand­ing ex­traor­di­nary ev­i­dence (a bias known as “mo­ti­vated skep­ti­cism”).

Ex­am­ple: A failing new employee

One con­tributed ex­am­ple:

A few years back, a se­nior per­son at my work­place told me that a new em­ployee wasn’t get­ting his work done on time, and that she’d had to micro­man­age him to get any work out of him at all. This was an un­pleas­ant fact for a num­ber of rea­sons; I’d liked the guy, and I’d ad­vo­cated for hiring him to our Board of Direc­tors just a few weeks ear­lier (which is why the se­nior man­ager was talk­ing to me). I could have de­manded more ev­i­dence, I could have de­manded that we give him more time to work out, I could have de­manded a video­tape and signed af­fi­davits… but a new em­ployee not work­ing out, just isn’t that im­prob­a­ble. Could I have named the ex­act prior odds of an em­ployee not work­ing out, could I have said how much more likely I was to hear that ex­act re­port of a long-term-bad em­ployee than a long-term-good em­ployee? No, but ‘some­body hires the wrong per­son’ hap­pens all the time, and I’d seen it hap­pen be­fore. It wasn’t an ex­traor­di­nary claim, and I wasn’t li­censed to ask for ex­traor­di­nary ev­i­dence. To put num­bers on it, I thought the pro­por­tion of bad to good em­ploy­ees was on the or­der of 1 : 4 at the time, and the like­li­hood ra­tio for the man­ager’s re­port seemed more like 10 : 1.

Or to put it an­other way: The rule is ‘ex­traor­di­nary claims re­quire ex­traor­di­nary ev­i­dence’, not ‘in­con­ve­nient but or­di­nary claims re­quire ex­traor­di­nary ev­i­dence’.

In ev­ery­day life, we con­sider many more or­di­nary claims than ex­traor­di­nary claims — in­clud­ing many or­di­nary claims whose in­con­ve­nience might tempt us to dis­miss their ad­e­quate but or­di­nary ev­i­dence. In those cases, it’s im­por­tant to re­mem­ber that or­di­nary claims only re­quire or­di­nary ev­i­dence.


  • Bayesian update

    Bayesian up­dat­ing: the ideal way to change prob­a­bil­is­tic be­liefs based on ev­i­dence.