I worked a lot on this for the testing when DICE dev Julian Manolov first implemented DarkEthereal's ideas during BF4 CTE.
To cut a long story short, Having a coefficient of 0% (i.e viewspeed under "the centre of crosshair" match) makes "effective" sensitivity feel slower as zoom increases / FOV decreases with the different scopes in the game. This is a pretty objective phenomenon and every person who was testing it reported the same results.
The idea of USA in the first place was to create a system which could remove the need for random turn-rate multipliers, which were never going to work for everyone because they didn't/don't take into account user base FOV selection, even if the different ADS FOV multipliers were reasonably and accurately modelled in the first place (which they weren't).
A setting of matching 4:3 distance (i.e 133%) made the best compromise across ALL the zoom levels in BF4. which is why it is the default. It was not simply to "match CS:GO". There has been way too much emphasis put on that throw away remark from Dark.
If you're not interested in matching ALL zoom levels as close as possible (for example, if you only ever rock iron sights) then having a closer to screen center coefficient might work out better, and of course, the less you change FOV with a low-zoom scope the less the coeffecient matters anyway. But as been already mentioned, it is literally impossible to match a sensitivity across ALL the screen space of a 3D game, due to the 3D/2D distortion factor.
BF1 doesn't have as high zoom levels as BF4 with sniper scopes, and critically, there is no "blanking of the edges" outside the scopes. That is to say, your brain still gets viewspeed information from your peripheral view in BF1 with snipers (albeit blurry), whereas it doesn't in BF4. What you find when your peripheral view is blanked, off, and you can only see the "centre portion" of the FOV (which looks "flatter" and is the slower moving portion of the screen as you turn) is this creates a sensation of lowered "effective sensitivity". Which is one of the reasons why BF4 needed a slightly faster overall setting of 133% to be the default, cause we had to account for 20x zoom scopes with blanked out edges.
When I play BF1, since I only use zoom levels up to 4x zoom, and no scope blanks peripheral, I play at a coefficient of 88.8888%. In other words, exactly the halfway point between the centre of my screen, and the horizontal edge on a 16:9 display. This is also exactly the centre of any "FOV curve" also, since the 3D/2D conversion bends the image of course. Aiming to anything outside that point is slightly slower, aiming to anything inside that point is slightly faster compared to hipfire FOV.
One of the problems with this whole thing is the industry standard terms for defining what is "sensitivity" as far as muscle memory / aiming ability goes. There is a huge fixation in FPS gaming of matching 360 distances, but that is actually just "turn-rate" - an objective value, whereas "sensitivity" has to be subjective - the key is in the word - your SENSation of movement. Effective sensitivity - which is what matters when we're talking about aiming and muscle memory, is a combination of a turn rate across a monitor distance / FOV. I hear people all the time using this website or their tape measure to match 360 distances between games that don't even have the option to replicate the same FOV they use, and that is completely pointless, since a different FOV at the same turn-rate, changes effective sensitivity (i.e aiming muscle memory) which was the whole reason USA or aiming multipliers are needed in the first place.