I remember a trial done by the MOD's DERA (now QinetiQ) in the 90s that said "see and avoid" below 2000ft was about 99% effective if you look 70% outside and 30% on the instruments
I'd like to know how they arrived at that amazing 99% figure.
You can put four pilots into a 4-seater and fly around under a radar information service, on a gin clear day, and count how many of the reported targets that group of 4 people manage to spot.
It won't be 99%, it won't be even 50%.
Flying alone, on autopilot so hands and eyes free, I manage perhaps 10% - but then I fly higher than GA average, as high as CAS allows, so nearly everybody is below me, against the ground and very hard to spot.
And the majority of reported traffic is 'level unknown' i.e. nontransponding so you have no idea where to look for it vertically, but when you do spot it, it is way down below maybe 95% of the time.