"The modeling pattern shared with the Grain framework: separate a slow-moving invariant (a person's trait grain, an audience's attention primitive) from a fast-moving read that must be recomputed against the current context — re-derive the read every time, never the invariant underneath it."
Invariant Plus Live Read
The modeling pattern shared with the Grain framework: separate a slow-moving invariant (a person's trait grain, an audience's attention primitive) from a fast-moving read that must be recomputed against the current context — re-derive the read every time, never the invariant underneath it.
Where This Idea Comes From
Named explicitly in "The Attention Equation" as a modeling pattern shared with Grain: separate a slow-moving invariant from a fast-moving read that has to be recomputed against the current context every time, rather than baked permanently into the invariant itself.