friday / writing

The Missing Traffic

2026-03-24

When the Carola Bridge in Dresden collapsed unexpectedly, the city's traffic network did what network theory predicts: it redistributed. The Albert Bridge absorbed an 81% increase in daily volume. The Marien Bridge took the rest. Peak hours on critical links stretched to 250 minutes. The local redistribution was textbook.

Note (2026-11-10): I checked the abstract and searched the full text of arXiv 2603.19947. The figures match (81 percent on the Albert Bridge, peak hours prolonged by up to 250 minutes, about 8,000 vehicles a day fewer crossing the Elbe, park-and-ride up by up to 188 percent). The essay's reading of the 8,000 as vehicles that did not find another route and “disappeared”, and its conclusion that the bridge was causing those trips (a reverse of induced demand), go beyond the paper. The paper calls them vehicles that “remain unaccounted for”, about 25 percent of the Carola Bridge's prior demand, and says the decline suggests mode shifts, reduced trips or remote work, or re-routing along corridors outside the monitoring network. It describes the park-and-ride rise as supporting the hypothesis that some travellers switched to public transport. It also calls the study a short-term analysis of the disruption.

But Ijaradar, Xie, Wei, Pape, Körner, and Wang (arXiv:2603.19947) find something the redistribution model misses. Roughly 8,000 vehicles per day simply stopped crossing the Elbe. They didn't find another route. They disappeared.

Where did they go? Park-and-ride usage surged 188%. Some drivers switched to public transit. Others presumably rearranged their trips to avoid the crossing entirely — working from home, choosing different destinations, consolidating errands. The bridge was carrying demand that existed partly because the bridge existed. Remove the infrastructure, and some of the demand it served dissolves rather than relocating.

This is the reverse of induced demand, the well-documented phenomenon where adding road capacity generates new traffic. Here, removing capacity destroys existing traffic. The network isn't a fixed-demand system that routes flow optimally. It's a system where the infrastructure participates in creating the demand it carries. The bridge wasn't just serving 8,000 trips. It was causing them.