Building on yesterday’s post about micromorts and metric design, I want to actually give a more concrete example of a possible “better metric” for mortality risk.
As a reminder, my main criticism of the “micromort” is that a million (1,000,000) is not a number most people can grasp intuitively. I suggest instead indexing the risk of death relative to some easily understand baseline risk.
For this example, let’s use the risk of dying on given day due to “unnatural causes” (e.g. an accident, murder, suicide, etc.) in the United States (US). I think most people would understand that this risk is quite low on any given day, but over the course of a lifetime is not insignificant.
Let’s call our metric Relative Mortality Risk (RMR). It is calculated as follows:
A real example of this formula applied:
\[\operatorname{RMR}_{\operatorname{skydiving}} = \frac{8^*}{1.6} = 5 = 500\%\]This result can be interpreted as: If you go skydiving you are 5 times (500%) more likely to die that day [due to unnatural causes] than on an average day. Given the RMR is over 1 (100%), skydiving would be considered a relatively risky activity. By comparison, the RMR for a day spent skiing is ~44%. As such, a day spent skiing would be relatively low risk.
In case you’re interested, here are the RMR’s for some of the other activities** listed on the micromort Wikipedia page:
Note, while I do believe our RMR metric is better, it’s still not perfect. It doesn’t take into account the relative frequency or rarity at which an “event” occurs or an “activity” is undertaken. For example, the murder / manslaughter risk above applies every single day while climbing Mt. Everest might be something you only do once in your entire life. In other words, it’s still hard to think about and compare chronic / frequent small risks vs. rare / one off high risk events.
* Micromorts per jump (assumes 1 jump per day spent skydiving)
** Unless otherwise specified these are approximately 1 days worth of the activity (e.g. 1 day spent skiing, 1 skydiving jump, running 1 marathon, etc.)
(Note, all data here is from the Wikipedia article on micromorts. I can’t attest to the quality / accuracy of this data, but this is meant to be just an example so it’s not particularly important that it’s super accurate.)