{snip}
The singularity (at r=0) in Schwarzschild 's solution cannot be a true singularity. It must be an approximation. It must be hiding a true mass source [alternatively, flux-capacitor posted a solution by Andrei Sakharov (?) positing that it hides a wormhole].
The answer lays in
topology.
Actually, Sakharov's hypothesis about wormholes was, although clever and ahead of time, just an idea without much mathematical demonstration. There are more recent (2015)
mathematical demonstration clearly showing that the so-called "true singularity" at
r = 0 can be cancelled (in fact, more precisely, does not exist) "within" a black hole. Because a black hole has no interior.
First, Karl Schwarzschild himself posited
r > 0.
r cannot go negative according to Schwarzschild. It is clearly written in his two founding papers, attached to this post in both German and English, that almost nobody ever has read in their original 1916 form nor in their more recent English translation. That is to say, space coordinates are always
real. Doing analytic continuation for
r < 0 is nothing but allowing space coordinates to become imaginary. In other words, it is allowing someone to make distance measurements of a hypersurface, but outside of that hypersurface. It is meaningless.
Second, and this is related as this choice to operate using real quantities and not imaginary ones, to stay in the real, laboratory physics, Schwarzschild considered only the metric signature (+ – – –)
Third, the main error everyone did and still does is to posit
r is a radius, i.e. a radial distance from a center point, like in a sphere. But
r is no a
radius, it is a
space marker. For
r Schwarzschild used the words "polar coordinates" (
Polarkoordinaten):

But Schwarzschild introduces what he calls an "auxiliary quantity" (
Hilfsgröße)
R, and it is through it that he expresses his "exterior" solution in his first paper in January 1916:

So the variable
r chosen by Schwarzschild is strictly positive, and the intermediate quantity
R is not free but has a lower limit α (which will later be called the "Schwarzschild radius" mostly written
Rs):
r =
Rs was at first considered as a "true singularity" but is has been shown after Schwarzschild's death that is is only a "coordinate singularity" that can be eliminated through various changes of coordinates.
What about
r = 0, that is to say
R =
Rs ?
It has been shown (
reference) that a change of coordinates can eliminate this "true" singularity, because such singularity is not really "true" it is caused by a wrong choice of local topology.
The "magical" change of variable is to take:
r = Rs (1 + Log ch ρ)To explain quickly how this is done mathematically without resorting to the full maths of the peer-reviewed paper, I will decrease the number of dimensions so everyone can visualize. The Schwarzschild exterior metric describes a 4D object:

Removing the part of the metric related to time, keeping only the spatial part of the metric, we get the description of a 3D hypersurface:

We continue and remove one space dimension, to get a 2D surface:

Studying the surface described by this metric, it is easy to show it corresponds to a parabola:

Applying a rotational symmetry to the parabola around the z-axis, returning to a 2D surface embedded in a higher dimensional 3D space, one gets the shape of a
diabolo:

which connects two spaces together through a throat circle:

The diabolo is represented above with two circular boundaries on both sides to better show the general shape of the object, but this surface is actually borderless as it connects two flat spaces:

That is to say in higher dimensions, this bridge connects two Minkowski spaces together through a throat surface.
One fundamental question for the black hole model is: can we choose the topology of a hypersurface freely, or does the topology of a hypersurface is intimately ensueing from its metric?
In the correct interpretation of the topology description of the object, contained in its metric, there is no center in this hypersurface. When you chose, like Schwarzschild, to keep real quantities everywhere, i.e. to be able to measure every point of the hypersurface in this solution of the Einstein field equations, everything becomes clear.
r < 0 is not real. Space does not "become timelike" and
t does not "become spacelike" anymore beyond this boundary. There is no more "central singularity". When you approach the event horizon of the black hole, if you make periodic measurements of the space marker
r, you will notice that
r is decreasing. Reaching the boundary,
r has a minimal value. But going further ahead toward the (imaginary) "center" of the object, you will notice the space marker
r is now
growing again. The hypersurface is not contractible.
r never becomes negative. All coordinates stay real.
There is also some successive work showing such event horizon acts as a
one-way membrane with limited time transit for a test particle as well as for the point of view of a distant observer, contrary to the established "freeze frame" model of the black hole where time seems frozen according to a distant observer watching a test particle falling toward the event horizon (the fall of the test particle seems to take an infinite amount of time for the distant observer).

The key is to take
radial frame-dragging into account during the gravitational collapse of a destabilized neutron star (which is never done classically, even for the Kerr black hole where only azimuthal, i.e. rotational frame-dragging within the ergosphere is considered, due to the dense body spinning around its axis). Indeed in a Machian sense, space is also dragged along the radial implosion of the star. If some of you express interest in this area I can detail further about this crucial point of infinite time vs finite time transit.