◈ counterfactualResearch preview

A CITY. TWO FUTURES.

What if the shortcut
makes everyone slower?

Change the network. Test the idea. Follow the evidence.

EXPERIMENT 001

The shortcut paradox

Fictional network · real mathematics
Inspired by Braess’s paradox

NETWORK OBSERVATORY

After the intervention

DIRECTED NETWORK / EQUILIBRIUM FLOW40.0 min4,000 travellers45.0 min0 travellers45.0 min0 travellers40.0 min4,000 travellers0.0 min4,000 travellersOORIGINNNORTHSSOUTHDDESTINATION

Swipe the map to inspect each road.

Outer roads Central shortcut
Without shortcut65.0 min
Your network80.0 min
Difference per traveller+15.0 min
!

The shortcut backfires.

Individually faster choices create a slower equilibrium. At this demand, the change adds 1,000 traveller-hours in total.

Does the conclusion survive more traffic?

37 demand levels
02550751001,0004,0007,00010,000MINUTES / TRAVELLERTRAVELLERS
Without shortcutYour intervention

FALSIFICATION CHECK

One result isn’t a universal rule.

We rerun the same network at five traffic levels. A change that helps now can hurt later.

2,000 travellers-15.0 min
3,000 travellers0.0 min
4,000 travellers+15.0 min
5,000 travellers+20.0 min
6,000 travellers+15.0 min

Differences are relative to the same network with its shortcut closed. Traffic levels are sensitivity tests, not probabilities.

ASTRA / EXPERIMENT PARTNER

Let the model make a prediction.
Then make it face the result.

Astra proposes changes within your upgrade budget. The engine tests them. Astra then reviews which predictions survived.

Checking connection…

The local simulation is fully usable. Live AI experiments require the owner’s API connection.