Exact Parallel Futures

A demo of an event-driven world model applied to global shipping

What this demo shows

Pick one cause, such as the Panama Canal closing, and watch the model run global shipping twice, with and without it. Every port and strait that differs is marked, and each one can be traced back to the cause.

Model trained on

IMF PortWatch daily container port calls and chokepoint transits for 401 ports and 28 straits and canals, January 2019 to December 2024.

What Exact Parallel Futures proves about the approach

Supports exact, unlimited what-if analyses

Rewind to any moment, change one thing and run it again. Everything the change doesn’t reach stays identical byte for byte, so every difference comes from your change.

In demo

Close the Panama Canal for a month and 254 ports and straits differ, while 170 stay identical in all 16 runs. Close Hormuz and 73 differ, while 351 stay identical.

See it in Change one cause

Delivers precise causal tracing of events

Point at any single event, trace what caused it, and test whether it would still have happened without that cause, within the model.

In demo

Click any affected port and the demo follows one ship step by step, with the cause and without it, to where the two futures part.

See it in Why?

Builds the macro world from micro events

System-wide behavior emerges from individual learned events, one entity at a time, rather than being modeled from the top down.

In demo

Each ship learns its own route, port and berth choices from real port traffic. The global effect, such as ships rerouting around a closed canal, emerges from those choices and is not set by rule.

See it in the globe, Play

These are the model’s answers, not forecasts. Its test against real disruptions it never saw is in progress.

World models · cause and effect

Exact Parallel Futures

Today’s world models predict what happens next. This one answers what would have happened if: change one cause, hold everything else fixed, and the model works out the consequence, from single ships up to the global network.

Change one cause

   
loading…

Why this is new

  1. Learned from the ground up

    Every ship’s choice of route, port and berth is learned from real port traffic. The global effect is not predicted; it emerges from those choices.

  2. Everything else held fixed

    Both futures share all their chance, so what differs is the cause’s effect alone. A model that samples each future afresh mixes the effect with noise.

  3. Every effect has a reason

    Click any port or strait to follow the ships, step by step, back to the cause.

Why?

Example use cases for global shipping

Canal and strait closures

See which ports feel a closure of Suez, Panama or Hormuz, and when, before it happens.

Shippers, port authorities

Rerouting plans

Test where diverted ships queue up and which ports take the extra calls.

Carriers, logistics planners

Supply-chain risk

Find which suppliers and lanes depend on one chokepoint, and which do not.

Supply-chain risk teams

Marine insurance

Trace a disruption to the exposed ports and price the exposure with a reason for each.

Insurers, underwriters

Port capacity planning

Estimate how much extra traffic a port would absorb if a neighbor closed.

Port operators, governments