So, in that sense, the earlier machines, where the tuner had to think about each note, ended with a better tuning than today's machine decisioned tunings. <snip>
I love tuning with tunelab for pitchraising as it's almost uncanny how accurate it is. But for a decent result to satisfy a proper pianist I think you still need the amazing computer that is the human brain....the computer that can listen to g1 and discern which harmonic is most prominent and compromise that as little as possible but enough....etc. surely that's better than an intellectual decision made before 10ths, octaves, fourths fifths etc are taken into account and adjusted for....and in no time at all as well.
In theory, I can perhaps agree with the above, but practice has shown a different comparison. It is rare to find a piano that is evenly out of tune, with all sections the same amount off. When tuning aurally, the tech has no way to change the correction of, say, every octave, by an amount that will compensate for that octaves divergence from pitch. To illuminate: it is common to find a piano with a bass section 1 or two cents off at the very bottom, but 4 cents or more at the top, then at the first note over the break, it may be that there is 6 or 8 cents off and an octave above that, near the middle of the piano, only 3 cents off. Continuing upwards, the divergence may gradually increase. My SAT will allow me to measure the divergence as I progress through the scale,( I begin at A0 and go up, and usually measure 5 or 6 notes above where I am), and make compensations on the .3-1.5 cents range, leaving my tuning within .2 cents of the pitch I am aiming for. I have never seen an aural approach that comes near this accuracy. There are two reasons for this. Beginning in the middle of the piano, which is what aural tuning requires, lets any change of pitch elsewhere alter the beginning section, throwing a margin of error into the whole tuning. Also, aural tuning requires using the previous pitches as guides, and the previous pitches are being changed after they were decided upon. Those decisions may have taken into account the expected deviation to come, but that guess is nowhere near as accurate as the measured deviation that the SAT uses to compute the correction factor.
When I want my triple octaves to be pure, (usually for concerto use), I have to make these sub-cent corrections for the whole piano to hang together. I can't afford to leave a cent here and there variation due to changing pitches after I have tuned the previous notes. And when I want to play 4 octaves all together and have the lowest note's 8th partial to line up 3 octaves above, the machine can be programmed to do exactly that, yet the work to make them within tolerance by ear(if possible), will leave far less time to polish the unisons.
If a piano is exactly at pitch and needs to be changed up or down say 4 cents and be completely in tune, it is possible for an aural tuning to do that, maybe. I haven't seen it happen, though. I was trained by one of the best, and sold aural tunings in the professional world for years. When Al Sanderson finally developed the programmable SAT, I got one, and it made me a better tuner, in that I could tell the machine what I wanted, and it would provide the info to do just that. The consistency allowed the continual refinement of the program until the machine's guidance provided a template that I couldn't find anything that I needed to change. As an aural tuner I had to reinvent the tuning every time I sat down a the piano and got no benefit from the cumulative refinement of my efforts stored in an electronic chip.