Great insight! I always look back at my high-school internship at a real software company back in the 90's. I spent at least half my day sitting in the cube of the old unix greybeard (literally) just watching him code and once in a while asking what he was doing. It seemed like a waste of time, but 1) I became a really good c programmer from multiple summers of watching an expert, and 2) since I really wasn't in charge of typing, I had a second stack going in my head and when he would ask "what was the query we just sent to the database" or something like that, I'd remember and keep him going. So I think I helped him as much as he helped me.
I often try to recreate that with interns and juniors at my current job, but everyone is so anxious about not typing or not submitting commits they don't stay and watch me work. Plus I think I'm less comfortable with silence than my old mentor was. He didn't care if I was sitting beside him for a hour while he was quietly hacking away. I tend to feel the need to explain a bit too much about my thought processes.
I had a similar experience in my last position. I was helping to lead a very green second shift team and had very little experience on the system we were testing. I ended up shifting my hours to split them between first and second shift and would spend the first part of the day just watching a very experienced operator. All I really I did was just help him plot and interpret data in excel or matlab. When we transitioned to second shift I would be already be familiar with the issues of the day and got to practice the things I observed earlier in the day. Having the time to sit back and observe allowed me to ramp up extremely quickly on the program.
I feel the one-on-one relationship can work well in both situations where the individuals are peers and when it is more of a mentor-mentee relationship. In grad school I had a close friend that was on more of a VLSI track while I was on an RF track. We had different expertise, but a shared a common background of electrical engineering. This allowed us to bounce ideas off each other where there was enough competence to provide meaningful feedback and just enough diversity to provide a different perspectives/approaches.
I recently gave a listen to the BBC podcast The Bomb and some of the work done in the era seemed to also follow pairing of minds. Maybe it was just the way the material was presented, but there seemed to primarily be a team of two tackling each of the major components needed for the various atomic programs to succeed.
There was an article linked here a while back about tacit knowledge that I feel applies in a way to the mentor-mentee relationship. Many comments about skills learned by observation and imitation from one-on-one work. https://news.ycombinator.com/item?id=23465862
I often try to recreate that with interns and juniors at my current job, but everyone is so anxious about not typing or not submitting commits they don't stay and watch me work. Plus I think I'm less comfortable with silence than my old mentor was. He didn't care if I was sitting beside him for a hour while he was quietly hacking away. I tend to feel the need to explain a bit too much about my thought processes.