Build a world that can push back.

Edit roads. Compute the equilibrium. Search for counterintuitive interventions.

NETWORK OBSERVATORY

Where the traffic settles

Converged

Swipe the map to inspect and select roads.

Saved baseline60.2 min

Current network60.2 min

Change0.0 min

Baseline: Nine-junction commuter network, 3,500 travellers. The baseline is a snapshot; it stays fixed while you edit.

Relative equilibrium gap 9.22e-8 · 27 solver iterations. No meaningful unilateral travel-time improvement remains at this tolerance.

AUTOMATED INTERVENTION SEARCH

Which road would you change?

Try removing each road, then try increasing its congestion capacity by 50%. We recompute the whole equilibrium for every intervention. All comparisons use the current network at the current demand.

Design or import a network recipe

Define 2–30 junctions and up to 80 directed roads. Junction positions use an 800 × 460 canvas. Road time is free + slope × flow. Paste a network object, or a previous export containing a network.

{
  "name": "Nine-junction commuter network",
  "demand": 3500,
  "origin": "A",
  "destination": "I",
  "nodes": [
    {
      "id": "A",
      "x": 100,
      "y": 80
    },
    {
      "id": "B",
      "x": 400,
      "y": 80
    },
    {
      "id": "C",
      "x": 700,
      "y": 80
    },
    {
      "id": "D",
      "x": 100,
      "y": 230
    },
    {
      "id": "E",
      "x": 400,
      "y": 230
    },
    {
      "id": "F",
      "x": 700,
      "y": 230
    },
    {
      "id": "G",
      "x": 100,
      "y": 380
    },
    {
      "id": "H",
      "x": 400,
      "y": 380
    },
    {
      "id": "I",
      "x": 700,
      "y": 380
    }
  ],
  "edges": [
    {
      "id": "AB",
      "from": "A",
      "to": "B",
      "free": 8,
      "slope": 0.006
    },
    {
      "id": "BC",
      "from": "B",
      "to": "C",
      "free": 6,
      "slope": 0.008
    },
    {
      "id": "AD",
      "from": "A",
      "to": "D",
      "free": 12,
      "slope": 0.002
    },
    {
      "id": "DG",
      "from": "D",
      "to": "G",
      "free": 8,
      "slope": 0.004
    },
    {
      "id": "BE",
      "from": "B",
      "to": "E",
      "free": 4,
      "slope": 0.01
    },
    {
      "id": "DE",
      "from": "D",
      "to": "E",
      "free": 7,
      "slope": 0.005
    },
    {
      "id": "EF",
      "from": "E",
      "to": "F",
      "free": 7,
      "slope": 0.005
    },
    {
      "id": "EH",
      "from": "E",
      "to": "H",
      "free": 9,
      "slope": 0.004
    },
    {
      "id": "GH",
      "from": "G",
      "to": "H",
      "free": 5,
      "slope": 0.006
    },
    {
      "id": "HI",
      "from": "H",
      "to": "I",
      "free": 8,
      "slope": 0.007
    },
    {
      "id": "FI",
      "from": "F",
      "to": "I",
      "free": 7,
      "slope": 0.008
    },
    {
      "id": "CF",
      "from": "C",
      "to": "F",
      "free": 8,
      "slope": 0.003
    }
  ]
}

MODEL BOUNDARIES

This is a fictional static, single-origin traffic-assignment model with nonnegative affine road costs. It uses a pairwise path-flow solver, reports convergence, and preserves flow at every junction. It does not simulate queues, traffic lights, collisions, induced demand, public transport or real-world construction costs. Zero free-flow time is permitted as a mathematical abstraction. Road upgrades affect the congestion term only. Search is a deterministic enumeration, not a live AI result.