Set the search criteria before you see the map

So she made it difficult. She didn't search for events near our sites. She searched for the actual signature — depth anomalies, dual-rupture initiations, doublets separated by seconds rather than the hours or days that usually separated aftershocks, the specific mechanical strangeness that made a seismologist's own catalog flag an event as worth a second look rather than filing it under routine — and let the geography fall out wherever it fell, refusing herself the comfort of choosing the criteria after she already knew the answer they'd produce.

She sat with three points for two days, running the same discipline against them she'd run against Heath's basalt — checking each one individually, refusing to let the shape of the whole persuade her that any single piece was stronger than it actually was — and could not disprove any of them, and could not, either, honestly call three points a ring. Three points were a triangle. A triangle was what apophenia handed you for free, the minimum furniture required to start seeing a room.

And her own statement of the bound, at the table:

"Four points don't uniquely fix a new dipole axis," Amaleto said, her voice quiet, measuring the words so they couldn't be mistaken for speculation. "I know that. You need converging remanence directions or an explicit geometric construction before the claim becomes tight. But these aren't random. The paleopole data is already there, sitting in the public record. If these loci are anchoring a fragmented south pole now, the remanence shows they were positioned to do it before. The rock memory is already there. We just weren't looking at it as a single system."