When I did the original math I felt a little pang that I didn’t check the source of the 1o minutes of exercise = 1 microlife value, but the result was so overwhelming it didn’t matter. When I added in the time to travel and actually exercise it became a much closer thing, so the error bars on that estimate began to matter. Thanks to reader Steve B. I was able to access the appendix of the microlife article that provided sources for the value of exercise, which turns out to be these two papers.
Both papers are strictly correlational (average exercise per week vs. lifespan), making no attempt to correct for the fact that healthier people will find it easier to exercise and are more likely to do other life-extending things. Exercise was determined by self report, and it looks like those microlifes come from a reduced chance of dying during the study, rather than being tacked on to the end of your life.
There’s a lot of evidence that exercise reduces various markers and effects change associated with a longer life, so I’m pretty sure it’s still good, but I no longer have any faith in this particular number.