Black hole singularity is a surface not a point

Ethan Siegal (a former theoretical cosmologist who has lots of practice in his second career doing science outreach) did it well enough at a pop-sci level that I'll just point to his https://bigthink.com/starts-with-a-bang/what-universe-expand...

(I don't think I could do better [*]).

Here's a sketch for a crash syllabus that would take you closer to an answer I'd write, not being a practiced science communicator:

My approach would be to teach you some differential geometry on a differentiable Euclidean plane (mostly relating the classic Euclidean distance to the integration of a line element), then what a Riemann manifold is, then how a 3+1-d pseudo-Riemannian one differs from a 4-d Riemannian manifold (and understanding the Ricci curvature in an Einstein manifold), and take you to understanding the simplest of metrics on the Lorentzian manifold, and the concept of geodesics and how they separate into spacelike, timelike, and null. I'd also teach you early about affine distance so that you don't stumble into problems understanding that a pulse of light from the ground to a mirror on the moon and back to the ground takes about two seconds, and how a pulse of matter -- including a pulse of light -- loses energy in an expanding spacetime. (That's another where does it go question, and a good one to think about.) Then I'd introduce Raychaudri-equation-style thinking, with a spray of timelike geodesics separating, as a way of understanding the metric expansion of space and the FLRW metric (where each Friedmann-equation dust represents an enormous number of timelike and lightlike geodesics).

I'd also teach you about the Lagrangian and Eulerian specifications of the flow field, and how they relate to one another. We can have a idealized (freely-falling, feels-no-forces) Lagrangian observer follow one line in a spray of geodesics which are initially extremely close to each other, and which separate with the metric expansion of space. Some of the initially-close geodesics causally disconnect from our chosen Lagrangian observer, with close-but-less-close ones disconnecting quickly, and very-close ones staying practically parallel for a very very long time. This is basically the Raychaudri equation, as applied to cosmology. We'd want to explore radar distances between our Lagrangian observer and ideal reflective objects attached to other geodesics on the spray.

We then can relate all that to a spacetime-slicing approach where we track what's on 3-d spacelike hypersurfaces, in a Eulerian style, going from our Rachaudhrian spray to a collection of space-filling dusts or fluids that dilute away differently over time. This is the usual picture cosmology students operate with.

Understanding that, especially how expansion generates several cosmological horizons, is half of the key to answering your question. The other half is understanding that one can run the relevant equations under a time-reversal, with initially enormously distant objects freely falling towards each other and ending up practically on top of each other in the early history of expansion.

Along the way we'd also be talking about the thermodynamics, as expansion is adiabatic.

Our causal physics are all related to an extremely hot, extremely dense, extremely low-entropy volume in our billions-of-years-ago past, which we retrodict by studying fractions of later volumes (fractions as small as careful laboratory experiments and as big as large scale galaxy surveys). Anything close to that patch causally disconnected from us very early, and we'll never be able to hear from those parts of a big spray, and they'll never hear from us.

Just outside our very early universe, things probably look very similar to things just outside it. The logic here is that as our galaxy crosses out of a cosmic horizon of somone far away, our galaxy doesn't do anything weird, and likewise there are many galaxies currently crossing out of our cosmic horizons, and they probably aren't doing anything weird either.

Studies of the expansion history, still-viable cosmic inflation scenarios, and global spatial curvature have led to estimates (e.g. Guth's work) that some our early hot dense patch is at most 10^-23 of basically the same early hot dense stuff. That's fairly comparable to the number of atoms of water in the North Atlantic ocean, all of which are interchangeable, although they all have different histories of where they've been in Earth's oceans, the pressures and densities they've experienced on their travels, and so on. The pre-inflationary patch's tiny elements are pretty interchangeable although they'll have slightly different histories of expansion, galaxy formation, and so on, given tiny differences in their very early histories ("initial conditions"). Some may be overdense and quickly collapse. Some may be underdense and thus produce few if any stars.

Now, is that primordial hot dense patch embedded into something bigger? Good question! Does it even matter, given that it causally decoupled from us so early? Good question! How do we even begin to investigate that? Good question! That's all live postgrad and postdoc research, with a lot of focus on trying to make the low entropy part of our hot dense early universe seem un-special.

Siegel again: https://bigthink.com/starts-with-a-bang/cosmic-inflation-pas...

Once you have that under your belt you can join the manifold (pardon the pun) papers exploring the physical implications of various guesses about what's outside the everything-everywhere-everywhen fully determined ("block universe") picture painted by a notional exact solution of the Einstein Field Equations of General Relativity, which we can only successively approximate by sampling signals from our past.

But at least you'd then understand what it means to say that mean energy-densities fall over cosmological time, and that the centres of mass of galaxy clusters are separating over cosmological time, and that our distant distant descendants won't see any galaxies not presently in our local group.

For extra credit you could play around with embeddings of de Sitter space in higher-dimensional manifolds and run into the usual frustrations of it being quite hard to recover known physics -- one can even largely justify a statement like embedding a 3+1d spacetime into a higher dimensional spacetime is generally not possible. Of course, many people still attempt to make that work not so much to answer your question, but to find ways of more easily calculating the way our visible universe behaves.

[*] I'd have maybe said "its own future" and otherwise present a wordier version of what Siegel wrote (explicitly raising time-orientability), but really I'd want to explain why I'm mostly a blockworlder in spite of how us small temporary knots of atomic nuclei feel about that https://en.wikipedia.org/wiki/Eternalism_(philosophy_of_time...> and that maybe the real question is why our brains encode the concept of expansion at all. Anyway our puny brains can't hold all knowledge, we can't just pour in mathematical physicslike kung fu, helicopter piloting, or motorcycle-hotwiring skills in The Matrix movies, and the behaviour of the universe at scales of billions of lightyears didn't change once humans started printing cosmology textbooks. And it's OK if you haven't worked through any of those; just be careful of memorizing factoids from people who haven't worked through any of them either.

Yea, okay.

You are quite correct that, since the singularity is a line in spacetime, different incoming light rays (or free-falling observers, for that matter) can hit it at different points. Depending on how you choose your coordinates, you can set it up so that those points have different "time" coordinates. But that doesn't make the singularity a point in space. It means you're running up against relativity of simultaneity--whether or not different events on a spacelike line (or more generally a spacelike surface) happen at the same time depends on your choice of coordinates. You can, in fact, choose coordinates in which all events on the singularity happen at the same time (for a "time" coordinate that is genuinely timelike--see below). The standard Penrose chart does that, for example.

You are also correct that a good choice of coordinates can make it easier to see certain properties of a spacetime geometry. But it can also make it harder to see other properties. In this case, your choice of Eddington-Finkelstein coordinates is making it harder for you to see why your claim that the singularity is a point in space is wrong, and why the things I said above are true.

For example, inside the horizon, the Eddington-Finkelstein "time" coordinate that you are using is not timelike. It's spacelike. In other words, it's not actually a "time" coordinate (even though it's labeled as such). It is actually a "space" coordinate! You should be able to see this by observing that the singularity is a spacelike line, and in E-F coordinates it's a vertical line--i.e., the only coordinate that changes along it is the "time" coordinate. That means the "time" coordinate must actually be spacelike there.

And, for extra confusion, the r coordinate in Eddington-Finkelstein coordinates is also spacelike, even inside the horizon (unlike in Schwarzschild coordinates, where it becomes timelike). So in this chart there is no coordinate that is timelike inside the horizon! That means any timelike curve inside the horizon must have more than one coordinate in this chart that changes along it (in the simplest case, a radial timelike curve, both the "time" and r coordinates must change along the curve).