I was going to say,
that I guess in the good old days, but then I really don't know just how 'good' those days were, before all of the modernization that we have today.
Never the less, back when I was flying, I'll use the 727 as an example, the only time we every flew a coupled (autopilot) approach was in recurrency. For some reason the FAA felt that we had to show that we were able to let the autopilot shoot an approach to minimums, while we sat on our hands. I usually took advantage of letting the autopilot shoot the approach in the simulator, by taking a short nap. Admittedly this could cause a slight problem, if the guy in the right seat also took the same advantage to take a nap as well.
Back then most of us used 10,000 feet as our 'transition' altitude. That meant after takeoff, that the autopilot was turned on, the shoulder harnessed came off and the seat back was slightly reclined. On the descent it was just the opposite. Seat and seat back to normal, shoulder harness on and autopilot off.
Then I'd hand fly all the way to landing, no matter the weather. I don't think I can remember anybody ever using the autopilot on a visual approach. Back then we landed at quite a few airports whose runways did not always have any type of glide slope or a visual glide slope. Just used the old 3-1 method for judging a glide path. Worked very well.
It looks like for a lot of people, that has changed in today's world. Is it better? To be honest I don't know, as it seems that there are still way too many landing accidents, even with all this automation.
However, an accident such as the one at SFO, completely baffles me that with four pilots in the cockpit, two very experienced, that an accident like this could happen, modernization or no modernization.