Exactly, perception is what's key here. The scale is logarithmic in order to quasi-linearize it to what we hear....but everyone's perception is different (and ever changing). This is one reason why a CFM/dB ratio is not applicable. (neither is a non-logarithmic scale)

That said, since ambient in a very quiet room is ~20dB, saying 50CFM at 30dB is a better ratio than 25CFM at below-ambient noise levels is also preposterous. This is where high speed fans 'win'

The few things that matter, IMO....are:
1) maxing out fan performance for how much noise do you desire/require
2) minimizing noise for an amount of performance you desire/require
3) have flexibility in fan performance (undervolting for 24/7 and 12V for benching)

Based on these points is how I organized my data. Each datapoint is at 1V increments so it's easy to count down from 12V and compare. Also for this reason, I have not declared a 'winner.' The data, IMO, speaks largely for itself.

There are a few cases where I'd personally use a lower performance fan, however. If I needed a lot of fans for not a lot of money....Yate Loons. Some situations need easy manipulation of voltage.....say if at 7V (very easy to wire up w/o a rheobus) I want to hear nothing, a Silverstone FN121 would be necessary (and not a bad choice anyway).