Re:Canada grounds Boeing 737 Max 8
by mikeash on Wednesday March 13, 2019 @03:54PM
(#19382996)
Definitely! If you take the fatalities for the last ten years and convert that to a risk estimate, the result will have gigantic error bars. We can be confident that the real risk is greater than zero, and it's pretty clear that it's small, but the exact number is pretty much impossible to know.
The farther back you look, the less noise you get in your data, but you also incorporate data that's less and less relevant to the current environment. Ten years seems like a decent cutoff, although it is a cutoff that happens to have had a wild swing recently.
My original point still stands if you want to take the data back a bit further: even if you include Colgan, the resulting number is way below the 117 per billion listed on that Wikipedia page.
The farther back you look, the less noise you get in your data, but you also incorporate data that's less and less relevant to the current environment. Ten years seems like a decent cutoff, although it is a cutoff that happens to have had a wild swing recently.
My original point still stands if you want to take the data back a bit further: even if you include Colgan, the resulting number is way below the 117 per billion listed on that Wikipedia page.