patterns.

The cluster that wasn't

When you tilt the data the wrong way, things that look like coincidence start to look like design.