A thesis that I think is underappreciated is that there are superlinear returns to hours worked in a week up until very very high numbers, so it was fascinating to read this report (h/t Misha Yagudin):
A few highlights:
I usually sleep for 8.5 hours a night. I thought that for this week I could be tricky, starting at 04:00 and sleeping later and later so I’d only have to sleep six times for maybe eight hours a pop. Not only did it totally work, but my wake times didn’t advance as fast as my bedtimes, so I only lost 6.38 hours per day to sleep.
I thought this would make me tired and unable to concentrate on difficult programming, but energy and focus were actually really good except for one hour early Sunday night. I blame it on epic Viking metal and other super-energizing music, plus maybe the seven bars of 90% dark chocolate I ate. I had one or two cups of tea but no other caffeine, and I woke without an alarm every morning.
Normally, I work a focused-but-relaxed 60 hours in a week. I doubled that last week, but I feel like I was perhaps three times as productive. I could keep the problems in my head without cache eviction due to memory pressure. (I mean, there wasn’t anything else to think about.) With ever-deepening focus, I felt unstoppable. It was like getting 4.5 40-hour weeks’ worth of work done in one.
Most nights last week I programmed in my dreams, with vivid Tetris effect one night of doing CSS tweaks. (One night I had a nightmare of watching a YouTube video and then panicking upon realizing I wasn’t working.) Being that deep into my CoffeeScript, I found myself writing terser and terser code, since why do in five lines what you can obviously do in one?
After the laptop upgrade, I was surprised by how much more I wanted to work. The simple friction of slow builds and poor CodeCombat level simulation performance had been weighing down my enjoyment and efficiency this whole time. I’m never waiting three years to upgrade my gear again.
Man barely moves for a week, staring at patterns of light on a flat object and trying to make the patterns change. Every 2-4 hours, a stimulus is presented and he records how happy he is. He eats and sleeps as fast as he can so he can go back to looking at the lights.
It is the happiest week he has ever recorded by a wide margin.