What happens inside a black hole?

In 2015 LIGO detected gravitational waves from two black holes merging, matching general relativity’s predictions in detail. In 2019 the Event Horizon Telescope published an image of the shadow of the supermassive black hole in M87, and in 2022 of the one at the centre of our own galaxy. Decades of tracking stellar orbits around the galactic centre had already given its mass, about four million times the Sun’s, long before the picture existed.

Roger Penrose showed in 1965 that once collapse has produced a closed trapped surface, general relativity gives paths that simply stop: light rays that come to an end after a finite stretch and cannot be continued. The theorem does not assume spherical symmetry, which is what the earlier collapse models had relied on. It does assume other things, and what it proves is that those paths are incomplete, not that curvature anywhere becomes infinite.

It is widely read as evidence that general relativity is incomplete rather than as evidence that infinities are physically real.

Hawking showed in 1975 that black holes radiate, with a temperature set by their surface gravity, and therefore lose mass and can eventually disappear. For a black hole of the Sun’s mass that temperature is far below the cosmic microwave background’s, so for now it takes in more than it gives off. The result combines quantum field theory with curved spacetime, and it has become the best-known test that candidate theories of quantum gravity are held to, although the radiation itself has never been observed.

Quantum mechanics requires that information about a system be preserved. Hawking radiation appears to depend only on a black hole’s mass, charge and spin, so when a black hole finishes evaporating, everything that fell in would seem to be gone. Hawking drew that conclusion himself in 1976 and proposed that predictability fails.

It is now widely believed that information escapes somehow, though not by everyone. Much of the case comes from string theory, which counted the microscopic states of certain black holes and got exactly the Bekenstein-Hawking entropy, and whose duality between gravity and ordinary quantum systems points the same way. More recently, calculations of how the radiation’s entanglement entropy evolves have reproduced a curve consistent with nothing being lost. There is no agreed mechanism.

General relativity says nothing locally special happens at the horizon of a large black hole: an infalling observer would notice nothing. In 2012 Almheiri, Marolf, Polchinski and Sully argued that this cannot be held together with a pure final state and with the information leaving from near the horizon, and that the most conservative way out is that the infalling observer burns. They proposed it as the least bad option, not as a finding. Both pictures cannot be right. A 2021 review by Almheiri and colleagues argues that the horizon stays smooth and the paradox is resolved another way; the disagreement is not settled.

A theory of quantum gravity would be required to say. There is no experimentally confirmed theory of quantum gravity. Candidates exist, and none has been tested: the scale at which quantum gravity is expected to matter is far beyond any present experiment.

Black hole complementarity, proposed by Susskind, Thorlacius and Uglum in 1993, holds that the inside and the outside are two descriptions of one thing and that no observer ever sees both, so asking what is in there independently of the outside is not a question the theory answers. Whether that is deep insight or a change of subject is itself disputed.

The holographic principle, proposed by ’t Hooft and developed by Susskind in 1995, holds that everything inside a region can be described by information on its boundary. A conjecture by Maldacena gave it a concrete form for one particular kind of universe, with a negative cosmological constant. Ours does not appear to be one of those, and no version of the principle has been established for it.

In several candidate theories the singularity is smoothed into something finite, such as a bounce or a region where spacetime stops being classical. Ashtekar and Bojowald found one such resolution of the Schwarzschild singularity in a simplified model built on the quantum geometry of loop quantum gravity. These are calculations within unverified frameworks, not descriptions of observed objects.

The original conclusion (that information really is lost) has not been refuted, and it still has defenders. Its critics argue that it would violate fundamental principles of physics. Unruh and Wald, who defend it, argued in 2017 that it violates none.