The 3D version shows one scripted breach: the packet either stops dead at the DMZ or (with segmentation off) sails straight through. This 2D companion turns that single anecdote into a real statistical experiment. Each "attack attempt" is an independent Bernoulli trial: with segmentation ON, the DMZ gateway blocks the attempt with the probability p you set on the slider (representing real IDS/firewall rule strictness — no defence is 100% effective); with it OFF there is no checkpoint at all, so every attempt reaches the Level 1 PLC.
Run enough trials — either by clicking repeatedly or by starting auto-fire — and the observed block rate (blocked ÷ total attempts) should converge toward the configured p by the law of large numbers, exactly the way a fair coin's observed heads-rate converges to 50% over many flips. The convergence chart plots that running estimate against the configured line so you can watch the gap shrink as the sample size grows, and the |Δ| readout gives the numeric gap.
- Level stack — the same six Purdue levels (L0 field devices through L4–5 corporate IT) as the 3D scene, drawn top to bottom with the DMZ as the sole chokepoint between corporate and industrial networks.
- Packet animation — each trial's outcome is played out visually: a green tick stopping at the DMZ, or a red tick punching through to the PLC ring.
- Outcome strip — a scrolling history of the last 80 trials (green = blocked, red = breached) so streaks and randomness are visible at a glance, not just the aggregate rate.