Your attached image doesn't work
You need to put in the embed code. It should have a .png (or the like) at the end.

But as for your argument, it doesn't work like that.
You would need Earth's surface to be flat, but sloped downwards to have that relationship.
The relationship which does work like that is for the angle.
If you want to use degrees it is 360 degrees/2*pi*r = 0.00899 degrees per km.
That means after each km, you will have curved down by 0.00899 degrees.
Quite difficult to measure.
And using radians will make the math easier as it is simply 1/r.
This also explains why the drop is not linear.
At the start, the slope is 0, so there is no drop.
But as you go further along, the angle of the ground changes meaning the rate of descent increases.
At the 1/4 way point, the slope is now 90 degrees straight down, meaning it is a 1:1 relationship.
The other way to easily see why it can't be linear is to keep on going.
Why stop at the south pole? Complete a full 360 degree trip.
If your statement was true, you have travelled the entire circumference, a distance of ~40 000 km, and so you should finish off 25484 km below where you started (4 Earth radii), yet clearly you are back where you started so it needs to be 0.
There are a few different ways to calculate it depending on exactly how you have defined it.
For the way you appear to have defined it, with the drop being the vertical distance measured from the starting point to a line parallel to a line passing level through you which passes through the final point, and the distance measured along Earth.
i.e. if we use this image here showing the different options:

You are using d1 and h3.
The angle at the centre relates to the distance by (using radians for simpler math):
d = r*a
a = d/r
The height relates to the angle by:
cos(a) = (r-h)/r
h=r*(1-cos(a))
Combining them you get h=r*(1-cos(d/r)).
This can be approximated as h=d^2/(2*r)
This comes from the small x approximation for cos, which in turn comes from the Taylor series expansion (starting from n=0):
cos(x)=((-1)^(n)x^(2n)/(2n)!) = 1-x^2/2+x^4/4!+...
The first 2 terms are the small x approximation, which gives:
cos(x)~=(1-x^2/2)
Putting in x=d/r we get:
cos(d/r)=1-d^2/(2*r^2)
And sticking that in the above we get:
h=r*(1-(1-d^2/(2*r^2)))
=r*(1-1+d^2/(2*r^2))
=r*d^2/(2*r^2)
=d^2/(2*r).
This is also where the often quoted 8 inches per mile squared comes from.
That means over a distance of 1.6 km (approximately 1 mile), where the 2 formulae cannot be distinguished by any sane means as the first error term is of the order 10^-16 km; we end up with a drop of 0.0002 km = 20cm =~8 inches.
As for how far you see, that depends a lot on altitude and how tall/high the object you are looking at is.
And it is important to distinguish between the drop due to curvature and how much should be hidden:

Consider this simple example of a person (shown in purple) looking towards the orange object in the distance.
The drop is more than the height of the object. So if they had their eyes level with Earth, their line of sight would be limited by the light grey line and they couldn't see the orange object.
But if they are standing above, they can have a line of sight like the red line and still see the object, or at least the top of the object.
Also note that the amount of the orange object hidden plus the height of the observer is not equal to the drop, the drop is much larger.
This is easiest to understand by considering the horizon as the 0 point. You have a drop to the left of the horizon towards the viewer, and you have a drop to the right of the horizon towards the object.

It is also complicated by refraction which results in light curving downwards allowing us to see more.
If you are standing on a beach, close to sea level, the horizon is typically around 5 km away, and has a drop of 2 m. Again, this is best done the other way, from the horizon there is a drop of 2 m which brings your eyes level to the horizon.
Then from the horizon, the curvature blocks the view, with the amount hidden being roughly d^2/2r, where that d is the distance from the horizon to the object.
As you gain altitude, you can see much further, as the horizon is much further away, as the curvature needs to drop more from the horizon back to you to obstruct your view.
And this matches what is observed quite well.
We see standing on the beach looking out that the horizon is relatively close, but we can see tall objects beyond it with the bottom missing.
We can see that as we gain altitude the horizon gets further away.
We also note that a telescope doesn't allow us to see beyond the horizon to see objects obstructed by the curve.
Instead, it just allows us to resolve objects that are otherwise too small to resolve, or see objects otherwise too faint to see.
Even looking at an object beyond the horizon will not bring the bottom back into view. Instead, we see the top, including portions much smaller than the part of the bottom that is hidden from view, clearly resolved, with the bottom still missing.
So what is observed matches curvature, not a flat Earth.