The black hole that was still there

Here is a wonderfully unreliable argument: three-dimensional gravity has no local gravitational waves; therefore it probably has no interesting black holes.

The first half is true for vacuum Einstein gravity. The second half does not follow. A black hole is defined by causal access to the outside world, not by the presence of a propagating graviton. The distinction becomes impossible to ignore in the solution introduced by Bañados, Teitelboim, and Zanelli in 1992.^[1]

The BTZ black hole lives in a theory with negative cosmological constant. For the uncharged solution we discuss here, every smooth local patch is still AdS₃. The black hole is in how the geometry is assembled globally.^[2]

Your neighborhood is innocent

Suppose you carried a tiny curvature laboratory across the horizon. In the idealized classical solution, it would not suddenly announce that the local curvature had become a new species of curvature. That is not how this horizon works.

The analogy with a cylinder returns, but now with causal structure at stake. Identifying points in a spacetime can change which paths are globally possible. Local measurements and global escape questions ask different things. A horizon belongs to the second category.

For a nonrotating BTZ black hole, one useful form of the metric is

The horizon sits at \(r=r_+\). The vanishing coefficient of \(dt^2\) and divergent radial coefficient there are features of these coordinates, not a divergent curvature invariant. The periodic identification of \(\phi\), meanwhile, is not decorative notation. It is part of the global specification.

A small black hole with a large demand

A black hole brings thermodynamics into the room. In units with \(\hbar=k_B=1\), the nonrotating solution has

The “area” in the entropy law is now the circumference of a circle. The dimension changed; the demand for an explanation did not. What microscopic states are being counted?

This is where “no local gravitons” becomes a productive provocation rather than a shortcut. If the entropy is large, then a quantum description must account for many states, even though the classical bulk lacks the familiar local waves. The challenge is to explain what the theory actually counts.

There is a useful restraint here. Finding a classical family of black holes does not tell us the exact quantum spectrum. Classical mass parameters can vary continuously; a particular quantum boundary theory may have a discrete spectrum. Moving between those descriptions requires an argument, not just a change of vocabulary.

The two clues finally meet

At this point in our reading route, we have two pieces that seem made for one another. Brown–Henneaux gives a boundary symmetry with a central charge. BTZ gives a macroscopic entropy. Two-dimensional CFT has a way to connect symmetry data with high-energy state counting.

Strominger would use that connection to reproduce the black-hole entropy in the appropriate asymptotic regime.^[3] The next article explains why that agreement is so striking—and why it still does not hand us the name and address of a complete boundary theory.

The surprise of BTZ is therefore more than “a black hole exists.” It turns a seemingly sparse theory into a sharp accounting problem. We removed local waves and somehow kept a very large bill.