Could a black hole interior contain an entire universe?

[krishnamurthy]

TL;DR

A thought experiment: Could our universe exist inside a black hole in a parent universe, with black holes in each universe containing further universes? Has anything similar been explored in physics?

I would like to share a thought experiment that occurred to me while thinking about black holes and the geometry of spacetime. I am not a physicist, and I am not proposing a formal theory. I am curious whether this possibility has been explored in theoretical physics.

The thought experiment

What if our entire universe exists inside a black hole in a larger parent universe?

In turn, black holes within our universe could contain other universes, whose black holes could contain further universes, potentially forming an infinite hierarchy.

The intuition that led me to this idea concerns the geometry inside a black hole.

We usually describe a black hole from the perspective of an external observer, using properties such as its mass and event-horizon radius. But inside the event horizon, spacetime geometry can be very different from our everyday intuition. Could the interior spatial geometry and volume be vastly different from what we might expect based on the exterior dimensions?

I wonder whether the interior of a black hole could contain a cosmological region with its own spatial extent and evolution, potentially much larger or more complex than its exterior appearance suggests.

I understand that this does not establish that another universe exists inside a black hole. The idea is purely speculative, and I may be overlooking fundamental constraints.

What I would like to understand

1. Are there solutions of general relativity in which a black-hole interior can develop into a separate expanding cosmological region?
2. How independent can the interior spatial geometry and volume be from the properties measured outside the event horizon?
3. Has anyone investigated the possibility of recursively nested universes of this kind?
4. What are the main mathematical or physical objections to such a scenario?

I would be grateful for references to relevant research or explanations of why the idea might or might not be physically possible.

I am sharing this simply as a thought experiment and an opportunity to learn.

 

[PeroK]

This was discussed recently:

T

High School Thread 'Questions about Schwarzschild Cosmology and White Hole Big Bangs'

Sep 3, 2026

Hi everyone,
I'm a student who is really interested in cosmology. I was thinking about how both the Big Bang and black holes start with a singularity, and I ended up finding out about Schwarzschild Cosmology and Nikodem Popławski’s model (where our universe expands like a white hole inside a parent black hole).
Since I'm just a beginner, I wanted to ask the experts here a couple of questions to help me understand:

1. How does a collapsing black hole turn into an expanding white hole? What is the physics behind that "bounce"?
2. If this theory is right, does it mean the black holes we...

• Tanya Li
• Replies: 7
• Forum: Astronomy, Astrophysics, Cosmology

 

[PeterDonis]

Are there solutions of general relativity in which a black-hole interior can develop into a separate expanding cosmological region?

No.

How independent can the interior spatial geometry and volume be from the properties measured outside the event horizon?

Not at all. A black hole spacetime has only three parameters, mass, charge, and angular momentum, and those are the same inside and outside the horizon.

Has anyone investigated the possibility of recursively nested universes of this kind?

There have been speculations about black holes spawning "baby universes" (Hawking wrote a popular book with a similar title), but they are only speculations and are outside the framework of standard GR (they basically involve some sort of speculative hypothesis about quantum gravity).

 

[krishnamurthy]

Thank you both for the references and explanations. I'll look into the existing models, particularly Schwarzschild cosmology and Popławski's work.

My intention wasn't to claim that a black-hole interior necessarily develops into a new universe, but to understand whether the geometry permits such a possibility and whether it could be extended recursively.

I'll read more before drawing any conclusions.

 

[pervect]

Not in general relativity, but in Einstein Cartan theory there are proposals that are not exactly the same as what you propose, but similar enough to be noteworthy.

A somewhat gee-whiz popularization can be found at https://www.aip.org/inside-science/every-black-hole-contains-a-new-universe.

A serious non-popularized treatment would probably require a graduate level thread, unfortunately.l

Einstein Cartan theory is very similar to General Relativity in the realm we can test experimentally, but it makes different predictions for black hole collapse. To oversimplify, Einstein-Cartan theory allows black holes to "bounce" and re-expand after they collapse, but they do not and cannot re-expand in the same universe they collapsed in. One nice feature of Einstein-Cartan theory is it doesn't have the same problem with singularities that GR does. The main difference between the theories is that in Einstein-Cartan theory, the property of matter called spin creates a property of space-time called torsion, which can act a bit like repulsive gravity, avoiding the singularity and allowing black holes to "bounce" rather than collapse to a singularity. General relativity assumes that there is no torsion. There's no experimental eveidence that the torsion predicted by Einstein-Cartan theory exists, because normally its effects are too small to measure.

 

[sbrothy]

There's also Penrose's "Conformal Cyclic Cosmology" and Smolin's "Cosmological Natural Selection". The last one, I think, has been refuted. Also, I'm not sure they're taken too seriously in the community. They're fun reads though.

 

[GR_fan]

Previous responses have been negative. However, within General Relativity, it is, indeed, possible to construct a model of a universe inside a black hole. That's allowed by the mathematics -- in a sense I'll unpack below -- but whether or not the model is realistic might be a different story. Certainly, there's no evidence that we do live inside a black hole.

Robert Oppenheimer and Hartland Snyder developed a very early model of the formation of a black hole. Their model involves gluing together two regions, an external region from Schwarzschild spacetime and an internal region from an FLRW spacetime. The exterior Schwarzchild region represents the spacetime outside of basically any localized non-rotating, uncharged mass, such as outside of a star. The interior FLRW region represents the material that's collapsing to form the black hole. FLRW spacetimes are also used to represent the universe as a whole in cosmology, so, while Oppenheimer and Snyder did not intend it, you could interpret the interior as a region of a collapsing universe.

As it turns out, the Einstein Field Equations are time reversal invariant. So, just as the Oppenheimer-Snyder model represents the collapse of a cloud of matter to form a black hole, so, too, the time reversed Oppenheimer-Snyder model can be used to represent the expansion of a cloud of matter out of a white hole. The geometry inside the expanding cloud is the same geometry as that of a region from an expanding universe. Moreover, in the Schwarzschild region, the time reversed version is identical to the ordinary version, so the exterior is still the exterior of a black hole. So, there you have it: an expanding universe with the exterior of a black hole.

Some caveats. The expanding region does not stay confined inside the horizon; instead, the expanding region eventually overtakes the horizon and grows to be arbitrarily large. Observers within the expanding region may be able to receive signals from outside the expanding region, so that they could, potentially, gather data indicating that they live inside an expanding cloud in a larger universe. Even though what I've described is allowed by the mathematics of General Relativity, there is no established astrophysical process that could lead to the scenario I described. The model violates the Cosmological Principle, though, personally, that doesn't bother me too much, since the CP has only been empirically established within our cosmological horizon.

There are other constructions that can be built from gluing together regions from different spacetimes. For example, you can glue an FLRW spacetime to the "throat" of a Schwarzschild black hole. Constructions like that are sometimes considered in the literature on inflationary cosmology. Unlike the time reversed Oppenheimer-Snyder model, this construction involves gluing the FLRW region to a spacelike surface within the Schwarzschild black hole.

Generally speaking, when you glue regions from different spacetimes together, you need to satisfy a specific set of mathematical conditions -- called junction conditions -- at the boundary. For technical details, see Poisson's Relativist's Toolkit. For gluing on spacelike or timelike surfaces, you can use the Darmois or Israel Junction Conditions. Those conditions ensure that the Einstein Field Equation is satisfied across the boundary. If you're interested in the mathematics, you need to ensure that the induced metric and extrinsic curvature match from both sides of the boundary. If the extrinsic curvature does not match, then there will be a "spike" in the matter-energy on the boundary, and, depending upon the situation, the spike might violate one of the standard energy conditions. Both features -- the spikiness and the violation of an energy condition -- are frequently considered objectionable, though at least some energy conditions are known to be violated in physically realistic scenarios.

1. Are there solutions of general relativity in which a black-hole interior can develop into a separate expanding cosmological region?

Yes, as I said above, there are various ways of building mathematical constructions of this kind within General Relativity.

Here's an important point. There's a theorem -- called Birkhoff's Theorem -- from which it follows that the spacetime outside some local overdensity is going to look like Schwarzschild spacetime. This is somewhat analogous to Gauss's law in electrodynamics. So, if the observable universe is a local overdensity in some larger, empty space, then the observable universe is going to look a bit like a black hole from the perspective of that larger, empty space.

1. How independent can the interior spatial geometry and volume be from the properties measured outside the event horizon?

As I said, in the Oppenheimer-Snyder model, an observer inside the expanding cloud could receive signals from outside the cloud. When the cloud expands beyond the event horizon, they can receive signals from outside the event horizon. However, if we're only considering the geometry of the time reversed Oppenheimer-Snyder model, the geometry inside the cloud is independent of the geometry outside the cloud, with the proviso that the appropriate junction conditions are satisfied on the hypersurface joining the regions together.

1. Has anyone investigated the possibility of recursively nested universes of this kind?

It hasn't been very popular, but various authors have investigated such scenarios. The closest to the scenario that you describe -- where we have universes nested inside universes -- is the Russian doll cosmology explored in some relatively unknown works, such as those by Robert Oldershaw. There's also an alternative to dark matter, known as Friedmons, which are supposed to be microscopic black holes with FLRW regions on the inside. These proposals are not mainstream. But as I also pointed out, there are more mainstream contexts -- such as in the inflationary multiverse -- where something qualitatively similar happens, at least from a mathematical perspective.

1. What are the main mathematical or physical objections to such a scenario?

We don't have empirical evidence of such scenarios and at least some of the scenarios -- like the time reversed Oppenheimer-Snyder model -- do not involve established astrophysical processes. More generally, the sort of gluing construction that I described is often seen as fairly artificial. Astrophysicists typically see it as allowing for an approximation of some more realistic model, but, unless that more realistic model is established, one could still object that the model is artificial.

My own personal take is that we don't know what spacetime, as a whole, is like, so it's worthwhile to explore the various possibilities offered by General Relativity. The trouble is that General Relativity offers an infinite range of possibilities. Using the junction conditions I described, you can combine solutions to the Einstein Field Equation, almost like playing with Lego. What we'd like to have is some way of identifying what sort of features are typical among the solutions to the Einstein Field Equation, but we don't (yet!) have an agreed upon method for doing so.

 

[PeterDonis]

in the Schwarzschild region, the time reversed version is identical to the ordinary version, so the exterior is still the exterior of a black hole.

No, this is not correct. There is no black hole in the spacetime you describe, the time reverse of the 1939 Oppenheimer-Snyder model. There is a white hole in the past, out of which matter comes that expands to form some kind of object that then persists infinitely into the future. The Schwarzschild exterior region is the exterior of that object--if you let yourself free-fall through the Schwarzschild region, you will hit the object, not a black hole.

It is possible to construct a different model, which is its own time reverse, and which consists of a white hole in the past, matter coming out of it that reaches a point of maximum expansion at a finite radius, and then the matter collapses back into a black hole in the future. In such a model, if you free-fall inward from the Schwarzschild exterior region, you might hit the collapsing matter before you reach the black hole event horizon, or you might not, depending on the specifics of when you start falling. But that model is not the time reverse of the 1939 Oppenheimer-Snyder model (as noted above, it is its own time reverse).

Neither of these models is anything like our best current model of the universe, of course.

 

[PeterDonis]

you can glue an FLRW spacetime to the "throat" of a Schwarzschild black hole. Constructions like that are sometimes considered in the literature on inflationary cosmology

Can you give a reference? I suspect that describing the surface of gluing as the "throat" of a black hole is misleading, at least in the context of this thread.

 

[PeterDonis]

if we're only considering the geometry of the time reversed Oppenheimer-Snyder model, the geometry inside the cloud is independent of the geometry outside the cloud, with the proviso that the appropriate junction conditions are satisfied on the hypersurface joining the regions together.

That proviso is not empty, however. It enforces a connection between the mass parameter of the Schwarzschild vacuum region and the stress-energy inside the cloud. So the two geometries are not independent.

 

[PeterDonis]

There's also an alternative to dark matter, known as Friedmons

Can you give a reference?

 

[PeterDonis]

the Russian doll cosmology explored in some relatively unknown works, such as those by Robert Oldershaw

Can you give a reference?

 

[GR_fan]

No, this is not correct. There is no black hole in the spacetime you describe, the time reverse of the 1939 Oppenheimer-Snyder model. There is a white hole in the past, out of which matter comes that expands to form some kind of object that then persists infinitely into the future. The Schwarzschild exterior region is the exterior of that object--if you let yourself free-fall through the Schwarzschild region, you will hit the object, not a black hole.

Well, I agree with all of this. When I said that the exterior is the exterior of a black hole, I meant that the exterior region is a portion of Schwarzschild spacetime. I did not mean that the spacetime includes a black hole. It does not.

Neither of these models is anything like our best current model of the universe, of course.

Agreed, with the proviso that, if you make the FLRW region sufficiently large, the FLRW region approximates the geometry of the actual universe. In that sense, a time reversed Oppenheimer-Snyder model, with a sufficiently large cloud of matter, is not directly ruled out by observational data.

 

[GR_fan]

Can you give a reference? I suspect that describing the surface of gluing as the "throat" of a black hole is misleading, at least in the context of this thread.

The idea in this construction is that you glue a space-like surface from a Schwarzschild region (the "throat") by modifying a spacelike surface in the early history of an FLRW model. This is one standard way of representing the formation of a universe in an inflationary cosmology. Nonetheless, I need to correct what I wrote. The idea isn't exactly that the joining hypersurface is spacelike; instead, the idea is that spacelike surfaces must be modified in order to glue the black hole throat to the FLRW region. Hopefully, you forgive me for making this error. :) As far as citations, here's a paper by Garriga, Vilenkin, and Zhang, published in JCAP in 2016: https://iopscience.iop.org/article/10.1088/1475-7516/2016/02/064

Please note that the "black holes" referenced in the title are not exactly the black holes OP has in mind. Instead, I'm referring to the details of the construction provided in the paper, where a baby universe is connected to a parent universe via a wormhole. That's what I mean by "gluing" across a black hole throat.

Can you give a reference?

Here, you are asking for a reference for a paper on Friedmons. Please note that I did not say that this is a *plausible* or *mainstream* hypothesis. I merely mentioned that the idea has been explored. Here are some relevant citations:

Barashenkov, V. (1983). The Laws of the General Relativity and the Phenomena of the Microworld (S. Syrovatkin, Trans.). In, Einstein And The Philosophical ProblemsOf 20th Century Physics. Progress Publishers

Markov, M. (1974). On Black and White Holes. Symposium - International Astronomical Union, 64, 106–131.

Markov, M., & Frolov, V. P. (1971). Metric of a closed Friedman world perturbed by an electric charge: Theory of electromagnetic “Friedmons”. Theoretical and Mathematical Physics, 3, 301–311.

For Friedmons as dark matter candidates, see:

Dokuchaev, V. I., & Eroshenko, Y. N. (2014). Black Hole Atom as a Dark Matter Particle Candidate. Advances in High Energy Physics.

Polishchuk, R. F. (2012). Hypothesis of friedmons as dark matter particles. Bulletin of the Lebedev Physics Institute, 39, 226–228.

Polishchuk, R. F. (2017). Hypothesis of friedmons as dark matter particles and hypothesis of initial cosmological constant decay. Bulletin of the Lebedev Physics Institute, 44, 114–117.

Can you give a reference?

Here, you are asking for references to the Russian doll cosmology. If I'm not mistaken, the idea can be traced back to two letters by mathematician IJ Good, published in Physics Today, in 1972 and 1983 respectively (see his "Chinese universes" and "Progress on Chinese universes"). Oldershaw offered a similar idea in several papers from the 1970s, published at times in reputable journals, that he called the "self-similar universe", though the connection to black holes is less clear than it is in Good's letters. I don't want to give the impression that these ideas are currently mainstream; I mention them only as examples where physicists examined ideas similar to those that OP mentions.

 

[PeterDonis]

When I said that the exterior is the exterior of a black hole, I meant that the exterior region is a portion of Schwarzschild spacetime.

Since you agree that there is no black hole in the spacetime, it should be obvious that the description "the exterior of a black hole" is very misleading. "Exterior Schwarzschild vacuum region" would be a much better and more neutral description.

 

[GR_fan]

That proviso is not empty, however. It enforces a connection between the mass parameter of the Schwarzschild vacuum region and the stress-energy inside the cloud. So the two geometries are not independent.

Well, yes and no. You are right that the condition enforced by the condition is important. However, with the right thin shell on the boundary between the two regions, you have a great deal of latitude in the geometries chosen for the two regions. As I mentioned previously, however, the resulting thin shell may be unrealistic, in which case, even though the model may be compatible with the Einstein Field Equation, the model would also be physically unrealistic/objectionable.

 

[GR_fan]

Since you agree that there is no black hole in the spacetime, it should be obvious that the description "the exterior of a black hole" is very misleading. "Exterior Schwarzschild vacuum region" would be a much better and more neutral description.

Fair enough. I agree.

 

[PeterDonis]

if you make the FLRW region sufficiently large, the FLRW region approximates the geometry of the actual universe

Yes, agreed. It is logically possible that our observable universe is actually within an expanding FLRW region that is of finite extent, and that there is a white hole horizon somewhere beyond it. We can't rule that out observationally; we can only say that on theoretical grounds such a model seems highly implausible.

 

[PeterDonis]

with the right thin shell on the boundary between the two regions, you have a great deal of latitude in the geometries chosen for the two regions.

But then the thin shell is itself a third region; you aren't gluing the other two regions together directly.

Nothing like that is being done in the Oppenheimer-Snyder model; it just glues an FLRW region and a Schwarzschild region together directly.

 

[GR_fan]

Yes, agreed. It is logically possible that our observable universe is actually within an expanding FLRW region that is of finite extent, and that there is a white hole horizon somewhere beyond it. We can't rule that out observationally; we can only say that on theoretical grounds such a model seems highly implausible.

Indeed! Agreed. (My only misgiving is that many of the arguments people give for thinking that various solutions of the Einstein Field Equation are "highly implausible" appear dubious to me, but I certainly don't want to give the impression that this is a mainstream idea or that, from a Bayesian perspective, our credence should be > 50%.)

 

[PeterDonis]

a space-like surface from a Schwarzschild region (the "throat") by modifying a spacelike surface in the early history of an FLRW model.

I see that you clarified what you meant, but I think it's worth noting that this kind of gluing is not possible. A spacelike surface from Schwarzschild spacetime that contains a "throat" has topology $S^2 \times R$. No FLRW spacetime has spacelike surfaces with that topology; the only possibilities there are $R^3$ and $S^3$.

The idea isn't exactly that the joining hypersurface is spacelike; instead, the idea is that spacelike surfaces must be modified in order to glue the black hole throat to the FLRW region.

Ah, ok. Yes, I've seen other papers talking about things like this. I think we had a thread some time ago on geometries that are something like the D&D "bag of holding", where a very small "throat" connects a large interior region to an exterior FLRW universe.

 

[GR_fan]

But then the thin shell is itself a third region; you aren't gluing the other two regions together directly.

Nothing like that is being done in the Oppenheimer-Snyder model; it just glues an FLRW region and a Schwarzschild region together directly.

In the original Oppenheimer-Snyder model, the two regions are made to match without a thin shell, because the extrinsic curvature from both sides of the joining hypersurfaces exactly matches.

But people often consider extensions of the Oppenheimer-Snyder model where the extrinsic curvature does not match from both sides. In this case, there is no third region. Instead, there's a delta function of matter-energy on the joining hypersurface that exactly compensates the discontinuity in the extrinsic curvature. So there's no additional regions, but there is a modification to the matter-energy distribution. Moreover, since we include a delta function in the stress tensor, we need to consider the distributional Einstein equation instead of the ordinary Einstein equation. All of this is discussed in Poisson's Relativist's Toolkit.

In cases where there is a third region, we don't have a "thin shell"; instead, we have a thick shell.

 

[GR_fan]

Ah, ok. Yes, I've seen other papers talking about things like this. I think we had a thread some time ago on geometries that are something like the D&D "bag of holding", where a very small "throat" connects a large interior region to an exterior FLRW universe.

Right. My apologies, again, for the confusion. In the future, I'll try to do a better job at double checking what I wrote before pressing "Post reply"!

 

[PeterDonis]

many of the arguments people give for thinking that various solutions of the Einstein Field Equation are "highly implausible" appear dubious to me

Such things are matters of opinion, yes. But the argument for a white hole not being in our past seems pretty strong to me: because not just the initial singularity, but the white hole horizon with an asymptotically flat region outside it, would have had to be "built in" to the universe "by hand" from the start. There's no way for such a configuration to evolve from something else that we know can exist, the way a black hole can evolve from gravitational collapse of a massive object.

 

[PeterDonis]

people often consider extensions of the Oppenheimer-Snyder model where the extrinsic curvature does not match from both sides

Can you give some references?

 

[PeterDonis]

a delta function of matter-energy on the joining hypersurface that exactly compensates the discontinuity in the extrinsic curvature

Strictly speaking, I suppose this isn't a "third region", but that's just quibbling over terminology. The point is that you've explicitly broken the direct connection between the "interior" and "exterior" spacetime geometries and their properties by putting in something between by hand, and since mathematically you can pretty much adjust that something between however you like, you can of course obliterate any significant connection between the interior and exterior by doing so.

In other words, you're allowing yourself to completely ignore any requirement to match the interior and exterior geometries, and then saying that the interior and exterior geometries don't match. Well, of course not. But that's just gerrymandering your model to get the conclusion you want.

 

[GR_fan]

Such things are matters of opinion, yes. But the argument for a white hole not being in our past seems pretty strong to me: because not just the initial singularity, but the white hole horizon with an asymptotically flat region outside it, would have had to be "built in" to the universe "by hand" from the start. There's no way for such a configuration to evolve from something else that we know can exist, the way a black hole can evolve from gravitational collapse of a massive object.

Right. Agreed. Two points.

First, it's not clear what sort of features can be reasonably "baked in" to the universe. It's not clear -- particularly in the relativistic context -- that we should understand the laws as evolving from state of the universe to another. But that's a tangential debate in philosophy of physics, and I think reasonable people can disagree.

Second, we often take a simplistic GR model to be a "stand in" for some more complicated phenomena. That's true even in the original Oppenheimer-Snyder model. So, could there be another model, with some qualitatively similar features to the time reversed Oppenheimer-Snyder model? I have no idea, but I think it's a possibility we shouldn't discount.

Can you give some references?

Sure. Here's one paper, where they work out a thin shell for both the classical (GR) case and in what the authors argue is a reasonable extension to a quantum mechanical case: https://arxiv.org/pdf/2001.06148

 

[GR_fan]

In other words, you're allowing yourself to completely ignore any requirement to match the interior and exterior geometries, and then saying that the interior and exterior geometries don't match. Well, of course not. But that's just gerrymandering your model to get the conclusion you want.

Again, my point was never that we should think this is a realistic possibility. Instead, my point was that General Relativity includes solutions of the kind that I described.

It should be a familiar point that the Einstein Field Equation allows for almost any geometry. Take any metric you'd like, calculate the Einstein tensor, and then divide by $8 \pi G$. The result is the stress-energy tensor needed for the assumed geometry.

Given this latitude in the geometry, astrophysicists usually constrain the space of solutions that they consider by demanding various energy, causality, and regularity conditions are satisfied. If you drop those constraints, you can build basically whatever you'd like. Given this large degree of latitude in the geometry, my view is that we should constrain the geometry of the visible universe from observations and then be agnostic about what spacetime is like for regions arbitrarily far beyond our cosmological horizon. It's fine if you disagree; I'm merely offering my opinion. :)

 

[PeterDonis]

It should be a familiar point that the Einstein Field Equation allows for almost any geometry. Take any metric you'd like, calculate the Einstein tensor, and then divide by $8 \pi G$. The result is the stress-energy tensor needed for the assumed geometry.

Yes, of course. Which is why...

Given this latitude in the geometry, astrophysicists usually constrain the space of solutions that they consider by demanding various energy, causality, and regularity conditions are satisfied.

But where do these conditions come from? From our observations of how things work in the portion of the universe we can observe and measure and manipulate. We observe that all the stuff we can see obeys these conditions. That's why physicists consider them to be reasonable constraints.

There is one exception: we know that quantum fields, under certain conditions, can violate the energy conditions. That's why, for example, models like the "bounce" models in one of the papers you referenced, or models like the Bardeen black hole (which has been discussed in previous PF threads, the name is actually a misnomer since there's no event horizon anywhere), are considered--because they make use of this property of quantum fields in the deep interior of the models to produce something that turns a gravitational collapse into something like emission of Hawking radiation. This avoids a number of issues with models that do have event horizons.

we should constrain the geometry of the visible universe from observations and then be agnostic about what spacetime is like for regions arbitrarily far beyond our cosmological horizon.

But that means we're allowing things in regions we can't see to be totally unconstrained. Which is pointless. If we're going to impose constraints on our models from observations at all, we're basically adopting the cosmological principle, that the parts of the universe we can't see are basically similar to the parts we can see. Otherwise we have no basis for constructing any model at all.

 

[PeterDonis]

could there be another model, with some qualitatively similar features to the time reversed Oppenheimer-Snyder model? I have no idea, but I think it's a possibility we shouldn't discount.

To me this "possibility" is like the "possibility" of Russell's teapot or Sagan's invisible dragon in the garage. The kind of model you're describing is basically: "the part of the universe we can see looks just like a flat, spatially infinite FLRW universe, but instead of applying Occam's Razor and assuming that the rest of the universe just extends that model everywhere, I'm going to arbitrarily put in a white hole horizon and an asymptotically flat empty region beyond it, way out where we can't see it, so my model is observationally indistinguishable from the straightforward FLRW universe everywhere model." Yes, logically and mathematically you can do this. But what's the point? If we're going to allow the part of the universe beyond what we can see to be anything we like, why pick a white hole particularly? As you point out, we can, by introducing appropriate thin shells and other paraphernalia, make the "exterior" region outside what we can see be anything at all. At which point we're not doing physics, we're just indulging in aimless speculation.