This question seems like the tip of an iceberg of complexity. The workers’ age, physical health and motivation probably matter. The contents of their non-work lives probably matter. In the case of programming, slightly degraded performance might mean enough bugs to be net negative, or it might just mean doing the same thing slightly slower. Caffeine-use patterns probably matter; use of other stimulants probably matters, too. In my own life, I’ve seen my personal productivity range from 80 hours/week to 0 hours/week over multi-month spans.
This question seems like the tip of an iceberg of complexity. The workers’ age, physical health and motivation probably matter. The contents of their non-work lives probably matter. In the case of programming, slightly degraded performance might mean enough bugs to be net negative, or it might just mean doing the same thing slightly slower. Caffeine-use patterns probably matter; use of other stimulants probably matters, too. In my own life, I’ve seen my personal productivity range from 80 hours/week to 0 hours/week over multi-month spans.
Agreed. I’d love to see data on all of those as well.