Ultra-high-energy gamma rays — photons above 100 TeV — come from somewhere, but from what process? Two mechanisms produce them: inverse Compton scattering (leptonic — electrons upscatter ambient photons) and pion decay (hadronic — protons collide with matter or radiation). The distinction matters because hadronic sources also produce neutrinos, while leptonic sources don't.
Reader note (2026-11-17): Checked against the full text of arXiv:2603.22555 (HAWC and IceCube public-alert search; v1 of March 2026), which I had earlier held as matching. The 24 coincidences match, with details the essay's population framing omits: 12 of the 24 spatial coincidences are close to MGRO J1908+06, and six come from the Cygnus region, where the paper says many are likely associated with the extended Cygnus Cocoon; one HAWC alert can be associated with more than one IceCube alert and vice versa, so the 24 are coincidences, not 24 separate events. The paper's chance estimate is an average of 0.002 random spatial coincidences with a standard deviation of 0.04, from drawing randomly from the alert position distributions, and it notes this does not account for any declination dependence of the IceCube containment radius. Its conclusion is that the coincidences are above chance but that the angular resolution and present statistics prevent firm conclusions about which sources are involved.
The authors (arXiv:2603.22555) cross-reference archival data from HAWC (a gamma-ray observatory) with public neutrino alerts from IceCube. They develop a quality metric for ultra-high-energy gamma-ray events and search for spatial coincidences with neutrinos. They find 24 coincidences — more than expected from random chance.
The excess suggests probable astrophysical associations: sources producing both gamma rays and neutrinos, which means hadronic acceleration. But the angular resolution of both instruments prevents pinpointing the specific sources. The signal is statistical, not individual.
The through-claim: multi-messenger astronomy works by counting coincidences, not by identifying sources. The 24 excess events don't point to 24 objects — they point to a population of hadronic accelerators whose individual members remain unresolved. The method detects a class before it detects an instance.