New Theory Suggests We’ve Been Wrong About Black Holes for 60 Years

How confusing inevitability with reality built decades of paradox.

What if general relativity never actually tells us that black holes already exist, but only that their formation is inevitable in an infinite future we can never observe? In a new theory, Daryl Janzen, a physicist at the University of Saskatchewan in Saskatoon, Canada, questions whether we’ve mistaken mathematical inevitability for physical reality, and shows how much of our black hole story rests on that quiet leap.

Black holes are among the most captivating and scientifically intriguing phenomena in modern physics, inspiring both scientists and the public alike.

But do they really exist? What if they are only ever forming, never formed?

Just imagine — what if the whole edifice of black hole physics is built on an invalid logical inference that’s gone unnoticed (or unacknowledged?) for the better part of a century?

Inevitability is not actuality — that’s obvious enough. Yet for sixty years physicists have ignored relativity’s most basic rule, and we’ve taken for granted that the latter is implied by the former. Like fools walking around imagining we’re all dead because someday we’ll die, they look at the evidence that nothing can stop black holes from collapsing toward their horizons and imagine that a process which remains forever incomplete has already come to its end.

Consider the following. We build a spaceship with three items onboard: a hot cup of coffee, a thermometer to measure the coffee’s temperature, and a clock that measures the arctangent of elapsed time since launch. The ship, which has perfect insulating walls, is launched — and through some future technological innovation it is capable of constant proper acceleration away from Earth for all time.

A continuous signal is transmitted back to Earth, sending two pieces of information: the coffee’s temperature and the arctangent time.

According to Newton’s Law of Cooling, the coffee’s temperature will approach the ambient cabin temperature exponentially and asymptotically — meaning it will very quickly approach the cabin temperature, but it actually takes infinite time to reach equilibrium. Therefore, the temperature value sent back to Earth will exponentially approach a finite value, but it will only reach that value exactly in the infinite future.

Since the elapsed time is sent back as its arctangent, that value will also asymptotically approach a finite value of π/2 in the infinite future.

And since the ship will be forever accelerating, the signal that transmits this information will quickly become practically invisible due to redshift. However, in principle the signal will forever be received back on Earth, as the coffee gradually cools and elapsed time increases — all while the ship’s velocity asymptotically approaches the speed of light.

A lost signal, rediscovered

Now imagine that the signal is lost, humanity forgets the experiment, and all records of the spaceship’s existence are lost. Centuries later, an advanced civilization picks up the very weak signal and observes that it’s still continually transmitting the two values.

After some observation, they determine that both values are gradually changing — that they’re asymptotically approaching finite limits. Thus, they know that no matter how long they wait they’ll never observe either of the transmitted values grow or decay beyond those limits.

These scientists are aware that because of the finite speed of light, the values they observe are not the ones presently associated with the physical object. They occurred some time in its past, when the signal was sent.

Due to the experiment’s basic physical setup, it is perfectly clear to us that the two values can’t ever reach their asymptotic limits. But the scientists observing the signal don’t know this. To them, the values very well could be approached asymptotically — or it may be that “now, out there” they’ve already evolved beyond the asymptotic limits, even though they’ll never be seen to reach that state.

But as scientists, they also know this ontological question is meaningless. From an empirical standpoint, from Earth all they can ever know is: when the signal they now observe was emitted, the values were still approaching their asymptotic limits.

And those values will never appear to have been reached “already” — no matter how long they wait.

But still, they wonder: what happens locally after the one value reaches π/2 and the other drops to its asymptotic limit? Surely, the arctangent-like value won’t simply end at a finite number, nor should the exponential decay flatten completely. In reality, has the event where the two values are reached already happened, or is it still to come?

Gravitational collapse to a black hole

The scenario just described mirrors, in all its basic features, the general relativistic description of spherically symmetric collapse to a black hole’s event horizon.

In that scenario, when the radius of a star collapses to the finite value of its event horizon, only a finite amount of proper time will have passed on a clock carried by a particle at its surface. These two values — the star’s radius and the elapsed time since it started collapsing — are approached in essentially the same manner as the coffee’s temperature approaches its asymptotic limit while the arctangent time approaches π/2.

And from the standpoint of external observation, the light emitted outwards from a collapsing star exponentially fades and quickly becomes invisible — though in principle, no matter how long an external observer waits, they will forever “observe” that when the light they are now receiving was emitted, the star still had not collapsed below its event horizon — just as the future civilization above will forever see that the coffee hadn’t yet reached equilibrium when the signal now being observed was sent.

But regardless of which past events are observed now, we’d also like to know whether collapsing stars should be thought to have already crossed this observational threshold, or if they are more accurately thought of as still approaching their event horizons.

In essence, while the image of the star freezes and fades, we’d like to know: Do gravitationally collapsing stars really pass through their event horizons and form massive singularities in our universe? Or, do they instead remain forever collapsing toward their event horizons, approaching asymptotic limits in both radius and proper time just as, in reality, the coffee’s temperature approaches equilibrium with the ship’s cabin while the arctangent time approaches π/2?

Spherical gravitational collapse geometry

General relativistic space-time diagrams are analogous to map projections of the globe, such as the Mercator projection. The key difference is that, instead of projecting Earth’s curved surface onto a flat plane, a space-time diagram projects two dimensions of space-time: one spatial dimension, and time. General relativity treats space-time as a curved geometry, much like the curved surface of Earth.

The figure below shows the same gravitational collapse scenario in three diagrams of the same underlying geometry. This is the Schwarzschild geometry, which describes a spherically symmetric gravitational field. The projection shown is the ingoing Eddington–Finkelstein diagram, which is commonly used to illustrate spherical gravitational collapse.

The red and blue lines in panel (a) (which are both gray in panels (b) and (c)) show the light cone structure — namely, the paths that photons take when moving radially in this geometry. “Infalling” photons, shown in blue, follow straight, tilted paths and reach r = 0 in finite intervals. “Outgoing” photons, shown in red, only actually move outwards when they are outside the “event horizon” (where the lines bunch together and become vertical). Near the event horizon, the “outgoing” photon lines tilt toward vertical, so an outward-directed photon remains forever at the same radius there. The broad band of nearly-vertical red lines shows how photons close to, but not precisely at, the horizon take increasingly long to escape. Inside the horizon the photon lines actually tilt inward, so even “outward”-directed photons must evolve toward r = 0 in this region.

Three world lines are also shown: the surface of a star, which begins collapsing from a radius outside the horizon, crosses the horizon and reaches r = 0 after a finite interval; an infalling astronaut, who follows a similar path starting from a larger radius; and an observer who remains forever at fixed r outside the collapsing star.

Different interpretations

General relativity is agnostic about which distant events are coincident with one another. This is by design: whether or not any two particular events occur “simultaneously” is not supposed to have any objective meaning in standard general relativistic descriptions. Panel (a) is therefore a faithful representation of the collapse scenario that makes no interpretive leaps. After the star begins collapsing, it inevitably reaches an event horizon after a finite proper time has elapsed, for instance — just as the coffee’s temperature reaches equilibrium in finite arctangent time.

But when does that event occur, for instance, from the perspective of the observer who remains outside?

Panels (b) and (c) present two fundamentally different interpretations of collapse that may be given from an external observer’s perspective, based on two different descriptions of “now” that are available to them — namely, the solid black lines. Panel (b) presents the canonical interpretation: at some finite “time” the star’s surface plunges beneath its event horizon, then “later” it forms a singularity at r = 0; shortly “after” that, the infalling astronaut also crosses the horizon, and “later,” at the top of the panel, also reaches the singularity.

“All the while,” the outside observer remains at fixed r. Additionally, because of the tilting of the photon lines, no matter how long they wait, they will always in principle be able to see the surface of the star and the infalling astronaut as they “were, before” reaching the horizon. While any light from the collapsing star or signal from the infalling astronaut must soon be gravitationally redshifted beyond detectability, the outside observer will in principle always see photons that were emitted “before” the star reached the horizon.

But this is a Cartesian reading of a flat projection of a curved manifold — a move that any geographer would immediately recognize as naïve.

Alternatively, the gravitational collapse scenario can be interpreted with respect to the solid black “now” lines in panel (c) (in fact, these are the lines relevant in the derivation of the geometry as a spherically symmetric space-time). In that scenario, the star begins collapsing “first,” then “later on” the infalling astronaut dives in as well. “Subsequently,” they both reach the event horizon “simultaneously,” each after finite proper time has elapsed — but only after infinite time has passed from the perspective of the outside observer. Since “now” is meaningless “after” infinite time has passed, these spacelike “now” lines do not extend beyond the horizon, which is in the infinite “future” in this scenario.

General relativity doesn’t specify

General relativity does not tell us which of these two alternative interpretations is true. General relativity is, by its very design, agnostic about which simultaneity structure, and therefore which of the two distinct ontological readings of this space-time, accurately represents “external reality” at any moment from the outside observer’s perspective. Therefore, on a faithful general relativistic reading, nothing at all can ever be claimed from the perspective of any external observer about whether the collapsing star has “already” reached its horizon or is “still” collapsing toward it.

General relativity does not say whether the cat is alive or dead.

From the perspective of every point in the external universe, even those infinitesimally close to the collapsing star, it forever remains undecided and unknowable whether the collapsing star actually — i.e. ontologically — is still approaching its event horizon — just as the temperature of the coffee forever approaches equilibrium with the spaceship’s cabin — or has already plunged beneath it.

Yet for half a century physics has focused only on the dead cat. The interpretation of gravitational collapse that’s been standard physics for decades holds that (b) really happens — ignoring both the equally valid picture in (c) and the undecided reality of collapse in general.

The only thing external reality can ever know for sure is that when the information now being received left the star, it had not yet reached its event horizon. Since no external event is ever causally connected to the horizon formation event, this is the only empirically valid claim that can ever be made.

Naming the fallacy

The primary problem with the canonical picture of black holes emerging through gravitational collapse is one of metaphysical overreach: we infer one of two essentially different potentialities to be true while ignoring the other, even though both must remain forever unobservable. The inference is therefore scientifically illegitimate, unjustified metaphysics.

This does not bear on the Penrose-Hawking singularity theorems themselves. Rather, the issue is with ontological overreach in drawing specific physical implications on the basis of these mathematical theorems. The theorems tell us that if certain energy and causality conditions hold, and if space-time is extended in a particular way, then geodesic incompleteness is inevitable. The upshot in the case of black hole singularities is that these must be a global feature of the space-time manifold.

But inevitability and actuality are not the same, and conflating them is a modal fallacy. We might name this the tense-import fallacy — or, more specifically, the present-tense import fallacy — the slide from atemporal mathematical features (event horizons, singularities) to present-tense claims about what has “already” occurred in our universe.

The canonical interpretation of black holes as real, already actualized objects within our universe is tied to a deeply problematic view by which space-time manifolds, along with the individual events such as those in the regions “inside” event horizons — including singularities at r = 0 — physically exist.

But the points in space-time should not be confused with physical reality. Rather, space-time should be understood as a set of events that happen in our existing reality.

From this perspective, the manifold is a descriptive tool, not the fabric of reality itself. It does not have to be maximally extended in any ontological sense; it only has to describe the physical events that occur in our existing universe in its domain of applicability. And there is no reason to expect that this domain should extend beyond an event horizon — namely, to suppose that collapsing stars are not still collapsing, just as the coffee’s temperature is still dropping. No existing theorem disproves this possibility.

The singularity theorems apply to the map, but that does not mean the map’s global structure should be reified and interpreted as physical reality.

From fallacy, to paradox

It is instructive to consider a live instantiation of the present-tense import fallacy. Penrose, for example, in Gravitational Collapse: The Role of General Relativity, began his description of a collapse diagram equivalent to our panel (a) above with the following accurate and precise statement (note that “2m” is used for the event horizon radius here):

The light cones tip over more and more as we approach the center. In a sense, we can say that the gravitational field has become so strong, within r = 2m, that even light cannot escape and is dragged inward toward the center. The observer on the rocket ship, whom we considered above, crosses freely from the r > 2m region into the 0 < r < 2m region. He encounters r = 2m at a perfectly finite time, according to his own local clock, and he experiences nothing special at that point. The space-time there is locally Minkowskian, just as it is everywhere else (r > 0).

But then, through a sleight of hand, the next paragraph employs the present-tense import fallacy, giving exactly the Cartesian reading (b) of the manifold:

Let us consider another observer, however, who is situated far from the star. As we trace the light rays from his eye, back into the past toward the star, we find that they cannot cross into the r < 2m region after the star has collapsed through. They can only intersect the star at a time before the star’s surface crosses r = 2m. No matter how long the external observer waits, he can always (in principle) still see the surface of the star as it was just before it plunged through the Schwarzschild radius. In practice, however, he would soon see nothing of the star’s surface—only a “black hole”—since the observed intensity would die off exponentially, owing to an infinite red shift.

Here, the words “after,” “before,” “always… still,” “as it was” and “just before” are all doing heavy ontological work, tying events along the star’s worldline back to the external observer, assigning them to the external observer’s past. But on a faithful general relativistic reading, this move is unjustified.

And it’s this sleight of hand, subtly triggering the tense-import fallacy, actualizing what’s only inevitable, that sets off a cascade of downstream paradoxes — paradoxes that have occupied a central place in physics ever since.

Penrose introduced his cosmic censorship conjecture in the quoted paper to protect the world from singular pathologies that could come to exist outside event horizons, for example, due to physically realistic angular momenta. But this conjecture is empirically unnecessary from an external standpoint when one recognizes that the external universe does not need to be “censored” by an event horizon at all — because the collapsing star that precedes it forever remains causally linked to the outside.

Neither the formation of the horizon nor the singularity that follows can ever be causally connected to the external universe. For all the external universe can ever know, the star may be still approaching the event horizon.

Similarly, Hawking’s proof that black holes should emit radiation rests on an assumption that there is already a completed event horizon. But for any point in the external universe, the only portion of the collapsing star’s worldline lying in its past light cone is the portion still collapsing toward the horizon. The state required for Hawking’s derivation to have any bearing on empirical reality — the completed event horizon — never enters the past light cone of any external event. A horizon that does not yet exist can’t already radiate into the observable universe.

Finally, on the illicit premise that black holes do emit this radiation, which would carry only information about their overall mass, electric charge and angular momentum, Hawking argued that information must be lost in the process of a black hole’s complete evaporation. This loss of information underpins yet another paradox that has puzzled physicists for decades. But the paradox itself ultimately rests on the same tense-import fallacy.

In the end, there are no paradoxes here. Each rests on a false premise: the assumption that an inevitable, globally defined structure has already been actualized in the external world. Strip away that assumption, and the cascade of “problems” collapses entirely.

So do black holes exist now? Perhaps — though it could equally well be that they never do come to exist — that they forever fall toward an event horizon, just as the coffee’s temperature never reaches its asymptotic limit. And even if they do, their existence has no bearing on anything we can ever observe or measure. In any case, it turns out that the decades of paradoxes built upon their supposed presence were never paradoxes at all, but consequences of a single, subtle interpretive misstep.

Scientific implications

The central outcome of the preceding analysis is twofold.

First, with respect to what black holes are right now in the external universe, general relativity is formally agnostic. The theory offers no preferred simultaneity structure for distant regions, and therefore does not adjudicate between two equally valid possibilities:

1. collapsing matter approaches an event horizon only in the infinite future;
2. the collapse has already completed, forming an event horizon and singularity.

Nothing in general relativity breaks the symmetry between A and B. The theory simply does not answer the question of which of these represents “what exists out there now.”

Second, general relativity is not agnostic about what remains empirically accessible.

Because photons and gravitational waves propagate at finite speed, every observation we ever make samples the past light cone of the observing event. And by the causal structure of Schwarzschild and Kerr geometries, no point on or inside an event horizon ever lies in any external observer’s past light cone. Therefore:

Every detectable signal from a gravitationally collapsing object — whether electromagnetic or gravitational — must have been emitted at a stage of its evolution when the collapsing surface was still outside its horizon.

This is a strict consequence of causal structure, independent of how one resolves the ambiguity in A–B above. General relativity is ambiguous about “what is out there now,” but unambiguous about “what was true when the signal we observe was emitted.” The proper physical state at the emission event is always pre-horizon.

Much of the literature on black holes tacitly collapses these two domains — treating an ontological assumption about the unobservable present (scenario B) as though it determined the empirical past. This is the present-tense import fallacy: inferring that an inevitable, globally defined structure must already have been actualized, and then projecting that assumption back into the observational domain. The familiar cascade of paradoxes — Hawking radiation, information loss, cosmic censorship — requires this illicit step before any of them can arise.

Once the causal structure is respected, the downstream problems evaporate.

Nothing on or inside a completed horizon is ever causally connected to the external universe, and therefore none of the inferences that depend on treating that region as empirically relevant can be justified.

The scientific implications become clear when we consider gravitational-wave astronomy. When facilities such as LIGO, Virgo and KAGRA detect the inspiral and merger of two “black holes,” the waves they measure were emitted at the moment of collision. And because no signal from a completed horizon can ever reach us:

The colliding objects responsible for every observed merger must have been ultra-compact, still-collapsing bodies that had not yet formed horizons at the moment of impact.

This conclusion is not optional; it follows directly from general relativity’s causal structure.

And by the same reasoning, every merger detectable from any point in the external universe — now or in the distant future — must likewise involve objects still collapsing toward their horizons at the moment their signals were emitted.

In a universe that continues to expand and undergo hierarchical structure formation, ultra-compact Kerr-like bodies will continue to merge and grow. But from any external vantage point, each collision will always be observed in its pre-horizon phase. The observational future of the universe is thus not one of mergers between completed black holes, but of ever-larger, ever-more-compact collapsing objects whose horizons are approached only asymptotically from the outside universe’s perspective.

Never miss a breakthrough: Join the SciTechDaily newsletter.
Follow us on Google and Google News.