Pawel Mazur and I proposed that black holes are not holes. They are equally compact objects with a physical surface and a non-singular dark energy interior.
The end point of complete gravitational collapse, in classical general relativity, is a singularity hidden behind an event horizon. That idealization produces the information paradox Hawking emphasized, and a number of related difficulties with quantum mechanics. We argued that those difficulties are resolved if the collapsed object is a gravitational vacuum condensate star, or gravastar: a thin quantum surface in place of the would-be horizon, and a regular de Sitter interior. The object is as compact as the Schwarzschild radiues, no singularity needed.
The theoretical basis has since been developed from the low energy effective theory of gravity, using the conformal anomaly and a dynamical treatment of vacuum energy. That same framework is relevant to cosmological dark energy. I am working on the observational signatures that could distinguish gravastars from black holes in gravitational-wave and multi-messenger astronomy.
The idea has been taken up outside the journals. Charles Q. Choi wrote about gravastars in Scientific American in 2018. In 2024 Kurzgesagt made a video that explains some of the core features through analogy with their characteristic visual style. The papers below are the technical development, and the students page is an invitation to work on what is still open.
Kurzgesagt, 2024.
A gravastar can look like a black hole in a LIGO merger or an Event Horizon Telescope image, at the precision we have now. Those data have not confirmed the idea. Chirenti and Rezzolla (2016) showed that the ringdown of GW150914 cannot be modeled as a rotating gravastar. Present searches for gravitational-wave echoes in LVK data are consistent with noise. Early claims of a signal were about 1.5σ; the collaboration did not accept them.
The 2023 review already says the ringdown is not yet precise enough to test a discrete spectrum. If later detectors see a clean Kerr ringdown, with no echoes, at a signal-to-noise ratio that should have shown a nearby surface, the thin-shell picture will be in trouble. That is the present state of the comparison.
A short reading list and some open problems are on the students page.