Can tensor networks capture black hole interiors the right way?

Black holes may be encoded in quantum information networks, but random tensor networks don't reliably predict the complexity of black hole interiors. The authors built a new class of networks called "twirled perfect tensors" that respect a principle called computational covariance—the ability to break space into low-complexity pieces without losing information. These networks match predictions from the Python's Lunch Conjecture while preserving other holographic features like the Ryu-Takayanagi formula.