Living Worlds

Research · Crowds · Oct 10, 2026

Crowds in four Jülich experiments

Not met (1 of 4)109 runs: counterflow, crossing, merging, a bottleneck

Improved A later registered study met the bar: Crowds in four Jülich experiments nobody had opened.

The test at a glance

What it tests
Accuracy against reality on the same four crowd measures.
Data
Jülich Pedestrian Dynamics Data Archive: 109 runs of counterflow, a right-angle crossing, a T-junction and a bottleneck with groups.
Held out
All 109 runs, unopened before registration.
Compared against
Social force, and straight walkers (ORCA reported).
Bar
Lower rank sum than both in more than half the runs of each experiment.
Result
1 of 4 experiments: not met.

With groups added, the crowd beat social force and the straight walkers in every run at the T-junction (11 of 11), but not in more than half of counterflow (3 of 15), the right-angle crossing (10 of 43) or the bottleneck with groups (17 of 40). Registered before any of the 109 runs was opened. The first pass stopped on a bug before anything was scored; the protocol was amended to fix only that, and every run was re-run once. Correction (Oct 11, 2026): 19 of the 43 crossing files state metres in their own header but were read as centimetres; the count above includes them and is left as registered.

Protocol, result and scorer

Protocol, with its amendment Data checksums First pass notes Second pass notes Result

Next

A new crowd, registered on four Jülich experiments nobody had opened: met (the result).

Crowds

More in this area.

Oct 10, 2026 Crowds Crowds against real pedestrians, six recorded scenes Not met (3 of 6) Oct 11, 2026 Crowds Crowds in four Jülich experiments nobody had opened Met Oct 11, 2026 Crowds Crowds among shoppers, a day nobody had opened Met (6 of 6) Oct 14, 2026 Crowds A learned walking model, on experiments and days nobody had opened Met (3 of 3 sets, on both seeds)
All research