In a fresh translation of "Ask Ethan" on Habr, a thesis slipped through that's easy to read and forget: white holes — time-reversed black holes — are mathematically just as legitimate in general relativity as ordinary ones. We've observed black holes by the hundreds; white holes — not one. Ethan Siegel breaks down why science remains skeptical, and provides a manifesto-calculation: if you compress all the mass of the observable universe into a black hole, its horizon would have a radius of about 16.5 billion light-years — suspiciously comparable to our cosmological horizon of 46.1 billion, especially if you remember that most of the "awkward" mass is hidden in dark energy.
Chasing "do white holes exist" here is boring: everyone's written the answer "nobody knows." What hooked me was different: the white hole in modern physics has stopped being set decoration from science fiction encyclopedias. Over the past ten years it's become a working tool — a stage in black hole evolution, a hypothetical form of dark matter, a candidate for the role of the Big Bang, and even a thermometer for checking whether we're sitting inside someone else's black hole. And almost this entire construction grew from one thought experiment by a Soviet cosmologist in 1964 and one mathematical solution from the 1960s that continues to deliver surprises to this day — up to a 2026 paper that unpacked a "missing triangle" beyond Kerr's ring singularity.
The concept of the white hole wasn't born in Hollywood or in fiction — in Moscow. In 1964, cosmologist Igor Novikov was the first to pose a question that sounds naive today but was then an act of pure theoretical honesty: what would happen to a black hole if time flowed backward? In that era black holes were still called "frozen stars," and even their existence was considered semi-fantastical — the future textbook Carlo Rovelli studied from flatly informed students that black holes were "most likely a mathematical fantasy." Novikov's thought experiment produced an object that continuously spews matter and light, and into which, conversely, nothing can fall.
The idea was picked up and developed by Nikolai Kardashev — the same one who would later give astronomy the civilization scale. In the book "Universe. Life. Reason" (1976), white holes already figure as a full-fledged research subject. It's telling that when New Scientist magazine recalled the history in 2014, it named as the main practical legacy of Novikov and Kardashev an ambitious plan to radio-hunt for white holes in galactic centers — with an interferometer "equivalent to a dish thirty times wider than Earth." The hunt, incidentally, caught nothing: what sits in galactic nuclei turned out to be very black.
But why, exactly, must a "white hole" exist even on paper? The answer is given by that very "maximally extended" geometry that Ethan mentions in passing, but sixties mathematicians — exhaustively.
The Schwarzschild solution describes an ideal uncharged and non-rotating black hole — and in its maximally extended version (Kruskal diagram) there's both a black hole region in the future and a white hole region in the past. The trick is that this "white half" only goes to an eternal black hole — a mathematical abstraction existing forever. A real black hole born from stellar collapse doesn't get this white region. All white hole physics revolves around the question: can you "deliver" it there retroactively?
A rotating black hole behaves much more interestingly — the Kerr metric, describing real objects like Sagittarius A* at our Galaxy's center. Inside a Kerr black hole is not a point but a ring singularity, and through it the geometry continues further — into a region with negative values of the radial coordinate. And here in July 2026, in the anniversary issue of General Relativity and Gravitation dedicated to Brandon Carter's 80th birthday, a breakdown was published that clears this territory of half a century's dust. Analytic continuation gives a rigid result: the region beyond the ring "can equally be considered as the Kerr solution for negative mass." Gravity there doesn't disappear or weaken — it changes sign. There are no horizons in this region at all, neither outer nor inner: a naked antigravitating continuation you can't fly to, but which must exist if you trust the equations everywhere.
The authors carefully unpack a long-standing source of confusion: the surface r = 0 is not one plane but a pair of isometric flat disks stretched over the ring singularity from two sides, and you can't identify them as one whole — the continuation loses smoothness. And there's an almost detective detail in the paper, highlighted in the title: the standard conformal diagram of the Kerr section that's migrated through textbooks for half a century is incomplete — it's missing an entire triangle of points with negative r. The "missing triangle" is that very antiverse, the region that textbooks quietly sterilized for decades. Astrophysicist Andrew Hamilton from JILA, whose visualizations of Kerr's interior Ethan uses in his article, calls it without euphemism — antiverse.
Sounds like pure math. But in 2014 physics decided you can not only write into this antiverse — you can get there. Though not alive, but with the whole black hole at once.
In January 2014, Carlo Rovelli and Francesca Vidotto published a short paper titled "Planck stars" — and it rebooted the entire white hole topic from scratch. The logic goes like this. As collapse proceeds, matter density grows, and at some point it reaches Planckian density — the scale at which gravity can no longer be described classically. The key observation by Rovelli and Vidotto: the threshold for quantum effects is determined by energy density, not size. So the "Planck star" at bounce can be much larger than the Planck length — not an exotic speck but a quite macroscopic object.
Next theoretical time relativity enters the game, and it plays a nasty trick on intuition. In the proper time of the collapsing matter, the bounce happens rapidly — in milliseconds. But from outside, due to colossal gravitational time dilation at the horizon, the same process stretches across millions and billions of years. Black holes we observe in the sky, from this perspective, aren't eternal traps but Planck stars caught in a deep bounce phase and shown to us in extreme slow-motion. No causality violation or faster-than-light travel — just clocks outside and inside running on incomparable scales.
A year and a half later, Rovelli together with Hal Haggard brought the idea to a rigorous result in Physical Review D: there exists a classical metric satisfying Einstein's equations everywhere except a finite interior region where matter collapses into a black hole and then emerges from a white one. In other words, a black hole can quantum tunnel into a white hole — like an alpha particle through a nucleus's Coulomb barrier, only the barrier is spacetime's causal structure itself. Rovelli parries the worst skeptical question beautifully: shouldn't quantum effects "smear out" at the horizon, isn't that forbidden? Turns out, no — quantum corrections can accumulate outside the horizon for a very long time, and their cumulative action is enough to trigger tunneling. Rovelli describes this route through Dante: general relativity is Virgil, leading you halfway, to the edge of the quantum barrier, and then Beatrice — quantum mechanics — becomes the guide. "I believe in white holes at least two or three days a week," he says in a 2025 interview, and by his standards that's practically a confession of faith.
More accurately: black and white aren't two different objects but two phases of one. For an external observer they're indistinguishable: as a Spectator reviewer put it, retelling Rovelli, "in a film about the Solar System run backward, the Sun still attracts the planets." The trick is that at the horizon time cleverly juggles roles, letting two solutions be the same outside and different inside — they can be "stitched" through a finite quantum region.
The most delicate point in the whole construction isn't the jump itself but the fate of what fell into the black hole. Rovelli and Vidotto write directly that the tunneling scenario "radically changes the information paradox discussion": instead of burning in the singularity, information waits out the white phase and returns outside. In 2018, Fabio D'Ambrosio and Rovelli broke down the mechanics in more detail: they studied the natural continuation of the geometry through Schwarzschild's central singularity and showed that Hawking radiation can be "purified" by information crossing the singularity from inside. The singularity transforms from an executioner's axe into a customs checkpoint between two phases.
From here also comes one of the most speculative but appealing dark matter hypotheses. In that same 2018, Rovelli and Vidotto calculated: after a black hole evaporates, a stable Planckian remnant remains — a microscopic white hole with large interior volume. The lifetime of such objects matches their production back in the universe's primordial reheating epoch, and they're protected from final collapse by loop quantum gravity's mechanism itself: area has a minimum eigenvalue, and the superposition "black/white with large interior volume" can't collapse further. A swarm of such invisible remnants is a ready candidate for part of dark matter. As a bonus the hypothesis claims to explain the puzzle of the early universe's low entropy — that very arrow of time standard cosmology prefers not to discuss.
And if white holes are larger? Could one of them someday click and spill out an entire world? Here physics splits into two camps, and both seem honestly compelling to me.
The first camp is spontaneous. Astronomers Alon Retter and Shlomo Heller in 2011 noticed an obvious trap: a white hole spewing matter continuously won't live long — the pressure from the ejection itself will collapse it back into a black hole. So if white holes exist, they must work in one pulse: all matter in one burst, instantly. They called this "Small Bang" and pointed to the only object in the universe behaving exactly this way: our own Big Bang. The only known white hole, they wrote, is it itself. As a candidate for a "small" echo, Retter and Heller proposed the mysterious gamma-ray burst GRB 060614 from 2006 — powerful, nearby, and leaving no supernova behind, contradicting all standard scenarios. This hypothesis's status is marginal, but the line of thought itself is beautiful: not "white holes are the opposite" but "a white hole is an event, not a body."
The second camp is rotational, and it relies on fresher data. Nikodem Poplawski develops Einstein-Cartan theory, where elementary particle spin twists geometry — creating torsion that at extreme densities works as a repulsive force. Torsion prevents singularity: matter bounces back, and on the other side of the horizon a "baby" universe is born. For its inhabitants the parent universe looks like a single white hole in their sky. The equations give an awkward consequence: if the parent black hole was rotating, its rotation axis is inherited as a preferred direction in the daughter universe — an absolute reference frame relative to which galaxies should align.
And here in 2025 came what made this whole branch stop being fantasy. Lior Shamir analyzed 263 galaxies with clearly determinable rotation direction in JWST data (JADES survey) — and discovered that about two-thirds of them rotate in the same direction relative to the Milky Way, instead of expected parity. "The difference is so obvious that anyone who looks at the image sees it," Shamir commented. "No special skills needed." The result has a boring alternative explanation — Doppler shift from Earth's rotation around the galactic center, traditionally considered too weak to affect such measurements; if so, the entire distance scale will need recalibration, which could also resolve Hubble tension. But if the alternative isn't confirmed, the second remains: the universe was born rotating — and that's exactly Poplawski's black-hole cosmology prediction. Cherry on top: fresh DESI 2025 data indicates dark energy may be weakening over time — and in a rotating universe "dark energy" could simply be a centrifugal contribution that must also weaken. Coincidence? Possibly. But beautiful.
Is any of this testable? The critics' weak spot — "a white hole continuously spews matter, so we'd see it" — Rovelli closed back in 2017, in a Nature Astronomy paper: a white hole signal isn't an eternal fountain but an explosion: a short burst in the TeV gamma range plus a radio pulse, suspiciously resembling mysterious fast radio bursts (FRBs). Moreover, both gamma and radio components should show characteristic "flattening" of the wavelength-distance dependence: you can identify them by it. Detecting such a signal would be the first direct observation of a quantum-gravitational phenomenon — a level higher than anything physics can do today.
In 2026 a group from UC Santa Cruz ran a careful audit: how often should primordial black holes tunnel into white holes today, accounting for competition with Hawking evaporation, cosmological depletion, and the "memory burden" effect that makes black holes evaporate slower than thought. The verdict is honest and not in enthusiasts' favor: at canonical parameters, tunneling rates don't reach FRB observation density — white holes can't explain the entire fast radio burst flux, only their small fraction, and only in narrow, heavily tuned parameter corners. But the calculation confirmed the main thing: the signal is in principle catchable, just a question of model calibration, not physics prohibitions.
And the very fresh storyline — August 2026: a Trieste group (Arrechea, Liberati, Spadafora) showed analytically what previously was only seen in numerical simulations. Semiclassical inner horizon instability of a black hole can destroy the trapped region faster than Hawking evaporation would — and following the trap's disappearance comes formation of an anti-trapped region, i.e., a dynamical white hole. The mechanism is universal: amplification of negative energy flows. But the most intriguing part is what follows: flows from the newborn white hole, going inward, can in turn recreate a trapped region. A cascade of "black → white → black → ..." transitions is possible, at the end of which spacetime is freed from horizons altogether. And none of this requires a separate quantum gravity theory — semiclassics suffices. If this cascade is real, a white hole stops being even the black hole's "finale": it's a phase collapsing matter passes through, like water through a turbine.
This whole story is about a quiet methodological shift. Twenty years ago a white hole was a symmetry exercise: "does a reversed solution exist? — yes, but nature doesn't use it." Today it's a hypothetical evolution stage of real objects, a form of dark matter, a possible cause of galactic rotation asymmetry, and a possible identity of our own Big Bang. Symmetry nobody asked for lived to an era when people started stuffing substantive questions into it.
My personal favorite from everything I read isn't the exotica but the "missing triangle" from the 2026 paper. Even in general relativity's most studied metric, sixty years after its discovery, scientists find entire regions textbooks quietly ignored. That's a good vaccination against cosmological boredom: our knowledge isn't "almost complete," it still contains pure blank spots at the very center of the most familiar objects.