technology 5 min read

The Leaky Black Hole That Could Rewrite Simulation Physics

A Japanese physicist built a working analog black hole on silicon — then discovered it leaks information in ways real black holes shouldn't. The finding could reshape how we simulate extreme physics, and what we can trust about those simulations.

  • Quantum Computing
  • General Relativity
  • Physics Experiments
  • Simulation

A Black Hole That Shouldn’t Exist — Until It Did

Antonino Flachi sat down at his desk at Keio University and built a black hole. Not a metaphor. Not a mathematical abstraction. A functioning analog of an event horizon, carved into a grid of 604 million bits running on a digital circuit.

He called it a digital acoustic black hole. What he found inside it may force physicists to rethink exactly how much they can trust their simulators of extreme gravity.

How You Build a Fake Singularity

Black holes are nearly impossible to study directly. They swallow light. They warp spacetime into geometries that no telescope can resolve past their own shadows. But the math that describes them — the equations of general relativity — looks strikingly similar to the math that describes sound moving through flowing fluid.

If water flows toward a drain faster than sound waves can travel upstream, those waves get trapped. An acoustic event horizon forms. That was the basic insight behind acoustic black holes, first proposed decades ago as tabletop experiments with liquid helium and Bose-Einstein condensates. Messy. Expensive. Hard to control.

Flachi replaced the fluid with logic gates.

His grid stores information as 32-bit words. At every clock tick, each cell updates based on rules derived from fluid dynamics. Where the simulated flow drags disturbances inward just as fast as they try to spread outward, an event horizon emerges — not from the wiring itself, but from the state of the system. A closed curve drawn across a lattice of numbers.

He then measured five properties against theoretical predictions: horizon stability, disturbance propagation, ringdown after perturbation, and superradiance — the process by which waves extract energy from a rotating black hole. All five came out consistent with the math. The horizon held stable for three hours. This was not a toy model flailing toward resemblance. This worked.

Then He Poked One Bit

Curiosity got the better of him. Inside the event horizon, Flachi flipped a single bit. One zero became a one among 604 million.

Most of the resulting disturbance stayed contained, as any event horizon should. But a small fraction leaked out. Short-wavelength waves, capable of traveling faster than the effective speed of sound built into the circuit, carried evidence of the interior violation across the boundary. The deeper the flipped bit sat inside the horizon, the less escaped — the leakage dropped off exponentially with distance from the boundary.

Real black holes do not leak. In general relativity, nothing crosses the event horizon outward. The fact that this digital analogue did means something important: the behavior of an event horizon depends on the underlying substrate in which it emerges.

What This Means for Simulation Trust

This is not a crisis for astrophysics. No one claiming that real black holes leak information should be worried by a digital circuit. But it is a crisis for anyone who assumes that analog simulation faithfully reproduces universal gravitational phenomena.

Flachi’s result draws a line between properties that are robust across physical substrates and properties that are artifacts of the specific rules governing the system. Superradiance survived. Ringdown survived. Horizon stability survived. But impenetrability — the defining feature of an event horizon — did not hold universally. It required long wavelengths, smooth disturbances, and the continuum assumptions baked into the acoustic analogy. Break those conditions and the horizon develops pores.

For researchers using digital or acoustic black holes to probe questions about Hawking radiation, information paradoxes, or quantum gravity effects, the implications are direct. If your simulator has a leaky horizon, you cannot use it to answer questions that require an impermeable one. The device still works. Just not for everything you might have assumed.

Who Wins, Who Loses, and What Comes Next

Computational physicists win here. Flachi demonstrated that a reconfigurable digital circuit can replace expensive experimental apparatus and produce results verifiable against exact mathematical predictions. The controllability — identical initial conditions, down to individual bits, repeated endlessly — is a luxury no fluid-based analog can offer. This architecture opens the door to parameter sweeps and targeted perturbations that would be impractical in laboratory fluids.

Theoretical physicists lose a convenience assumption. The idea that event horizon properties are universal across analog implementations gets qualified. Horizon transparency turns out to depend on wavelength-scale physics embedded in the substrate. Any future work using digital black holes to model information retention will need to account for this explicitly.

Quantum computing researchers should watch closely. A 604-million-bit logic gate array sustaining coherent black hole analog physics for three hours is a nontrivial computational feat. The same circuit architectures that host these simulations are the ones being scaled toward practical quantum advantage. The overlap is worth monitoring.

What comes next is straightforward in principle and hard in execution. Flachi’s team and others will need to map the leakage threshold systematically — at what bit-flip density, what disturbance spectrum, and under what clock rates does the horizon lose integrity? Does the effect scale predictably, or is there a critical regime change? And critically, can the architecture be modified to suppress short-wavelength transmission without destroying the other verified properties?

The preprint sits on arXiv, unpublished. Peer review has not yet weighed in. But the core finding is sharp enough to stand: you can build a black hole on silicon, and when you do, it doesn’t always behave like the real thing. That discrepancy is not a failure. It is data.

The Bigger Question

Flachi’s experiment forces a distinction that many in the field have treated implicitly rather than explicitly. Analog black holes are tools, not replicas. They reproduce certain behaviors of curved spacetime while discarding others. Knowing exactly which behaviors transfer and which dissolve into substrate artifacts is the difference between a useful instrument and a misleading mirror.

The leaky horizon is not a sign that the experiment failed. It is a sign that the experiment succeeded well enough to reveal something previously hidden — that even the most fundamental property of a black hole, its ability to keep things in, is contingent on the physics that hosts it.

That is the kind of result that shifts a field. Not by overthrowing established theory, but by drawing sharper boundaries around what our simulations can actually tell us.