Throttled Cascade

v4 · Phases 1—4 · Field + Agents
Companion to "Dispersion and Stratification"
Reformulation v0.4
What this is. A discrete chain of N=12 ranks. Capability $C_n$ flows from the frontier (n=0) toward the periphery (n=11) through throttles $\kappa_{n,n+1}$. Each throttle is a strategic primitive: lowering it concentrates the capability gradient at one position — a moat. The strain bands below show where the gradient is currently piling up. The capability profile shows the kink directly. Drag the throttle sliders and watch the system restratify.
What this is. Three historical arcs encoded as scripted $\kappa(t)$ trajectories. At each event, one or more throttles change abruptly — modeling regime shifts of Section 3.4. Pick an arc from the sidebar; the manual controls become read-only. Watch the kink relocate as the chain's topology effectively changes (a "new tier activated" is a previously-closed throttle suddenly opening). Compare the moat's shape and persistence across arcs.
What this is. The in-group has a true throttle profile $\kappa$ — your sliders. The out-group only sees a distorted profile $C_n^{vis}$, the in-group's strategic narrative version of the true $C_n$. They run a steady-state inversion to estimate $\hat\kappa$ from what they see. Push the distortion strength $\lambda$ up and watch the inferred $\hat\kappa$ flatten — the out-group sees a smooth gradient where the moat actually is. The dashed cool-blue line on the profile is what they see; the bars are what's real. The inference panel shows the gap. Weaponized underdetermination is the gap.

Timeline — pick an arc —

start end

Capability profile $C_n$ vs rank

frontier · n = 0 periphery · n = 11

Capability flow particles drift at κ-rate, colored by origin rank

frontier · n = 0 periphery · n = 11

Agent population people sit at ranks; mobility tracks $\dot C_n$ — promotion ↑, displacement ↓

colored by origin rank — promoted/displaced agents show as out-of-place colors 60 agents

Strain bands gradient distribution across throttles

each band: throttle (n, n+1) width ∝ |C_n − C_{n+1}|

Moat metric $M(n^*, t)$ — gradient share at each throttle

0—1 5—6 10—11

Throttle inference true $\kappa$ vs inferred $\hat\kappa$

true κ inferred κ̂ per-throttle, scaled 0—1

Readouts

t
0.00
C₀ frontier
0.00
C₁₁ tail
0.00
max M
0.00
at throttle
—