The consumer playbook says simplify: fewer numbers, more whitespace, progressive disclosure. Applied to an expert surface, that playbook deletes the expert’s peripheral vision. An operator responsible for forty machines watches a texture and waits for part of it to change.
Density is what makes that texture possible. Strip the screen down to "what matters" and you have decided, in advance, what will matter. The whole point of a monitoring surface is that you don’t know.
01 · LOUDNESS, NOT VOLUME
The design problem is how loudness is allocated, not how much information is present. A steady channel should cost almost nothing to ignore: a hairline, a dim figure. A channel leaving its band should be the loudest thing on screen without moving anything else. Most dashboards get this backwards: everything is equally loud, so the anomaly must be hunted rather than noticed.
Whitespace is not a virtue on an expert surface. Unearned loudness is the vice.
02 · WHAT WE’RE STILL TESTING
The open question is motion. Expansion-in-place preserves spatial memory but adds movement to a surface people stare at for hours. The Kilnmoor reference build settles it: rows never move, weight changes in place, and a persistent NEW marker does the work animation cannot.