Each supplier i has a hidden reliability ri (on-time, in-full probability) and a trust score Ti that the buyer actually observes and acts on. Every round the supplier attempts to fulfil its allocated order; the outcome updates trust by exponential smoothing of realised performance against expectation:
performance_i = 1 if delivery succeeds else 0
T_i(t+1) = T_i(t) + k · (performance_i − T_i(t))
order_i = capacity_i · allocation_share_i(T)
k is the trust responsiveness — how fast a buyer's confidence updates from one delivery. A disruption event randomly knocks a supplier's effective reliability down for a few rounds (a real-world shock: a port closure, a factory fire, a labour strike), which shows up as a string of missed deliveries and a trust drop even though the supplier's long-run capability hasn't changed.
- Diversified allocation splits order volume across all suppliers in proportion to Ti2 (softmax-like), so a shock to one supplier only dents a slice of total volume — this is supply-chain resilience through relationship breadth.
- Single-source allocation routes nearly all volume to the single highest-trust supplier each round, which is cheaper to coordinate but means a disruption there stalls most of the chain's service level at once.
- The 3D scene places each supplier as an orbiting node around the central buyer hub; node size tracks its current order share, node color tracks trust (red → low, green → high), and pulses along the connecting beam represent orders actually shipping.
Real-world relevance: this is the qualitative mechanism behind supplier scorecards and dual/multi-sourcing policy — procurement teams trade the efficiency of concentrating volume with a trusted partner against the resilience of spreading risk, exactly the trade-off this model animates.