Scale
A shared knowledge environment can be read at several grains: one retrieval, one working context, a project, a population, or the wider substrate. What looks like a stable region at one grain may resolve into several moving parts at another.
There is a recurring compositional pattern across these levels. Smaller units participate in larger contexts; larger contexts can sometimes be treated as units when the question moves up a scale. This recurrence is useful because it lets us compare dynamics without assuming a single privileged level.
The levels are not identical objects. A frame is a unit of context, a membrane is a computed boundary around co-activity, and an actor participates through its interactions. Treating them as interchangeable would hide the distinctions the observatory is meant to preserve.
Every reading therefore needs a declared grain and time window. Concentration across an entire substrate can mean something quite different from concentration inside one task context. Scale is part of the interpretation, not a cosmetic choice made after measurement.