Florian List was staring at a million simulated galaxies, each one flickering with artificial gamma rays, when the truth began to shift. For over a decade, physicists have debated the origin of a mysterious glow at the heart of the Milky Way—a diffuse halo of high-energy light known as the Galactic Center Excess (GCE). Some believed it was the signature of pulsars, those dense, rapidly spinning remnants of dead stars. Others held out hope that it might be the first indirect glimpse of dark matter, the invisible substance that makes up 85% of the universe’s mass. Now, thanks to a new machine-learning analysis led by List and his colleague Nick Rodd of Lawrence Berkeley National Laboratory, the scales have tipped back toward mystery.
The GCE spans thousands of light-years around the galactic center, a region so bright and cluttered with gamma-ray sources that disentangling individual signals is like trying to hear a whisper in a thunderstorm. Previous statistical models leaned toward pulsars as the source, but they ignored a critical detail: the energy of each individual gamma-ray photon. By training their algorithm on more than 1 million simulations that included both spatial and spectral data, the team achieved a more nuanced picture than ever before.
What they found was startling. Earlier studies suggested the glow came from relatively bright, unresolved point sources—consistent with known pulsar populations. But the new analysis reveals those sources would have to be far fainter than previously thought, so dim that their collective emission would mimic the smooth profile expected from annihilating dark matter. To account for the signal with pulsars alone, there would need to be at least 35,000 of them clustered in the galactic core—far more than the few hundred to few thousand earlier estimates suggested, and more than current astrophysical models predict should exist.
This doesn’t prove dark matter is responsible, the researchers emphasize. But it does mean the door is still open. “Our new analysis shows that the sources would have to be so faint that they would be almost indistinguishable from the emission expected from annihilating dark matter,” says Rodd. That single sentence reshapes a 15-year debate, pulling dark matter back from the brink of dismissal.
Published in Physical Review Letters, this work underscores how new tools can breathe fresh life into old questions. In an era where machine learning is often seen as a shortcut to answers, here it serves a deeper purpose: reminding us of how much we still don’t know. The Milky Way’s central glow remains unclaimed, a riddle written in gamma rays. And for now, one of the universe’s greatest mysteries—dark matter—still has a seat at the table.