Cities Can't Cherry-Pick Transit Stop Plans Anymore — New Audit Method

When a city decides where to put train, metro or bus stations, it usually compares several competing plans. The problem: each plan can quietly define its own measuring stick — which trips count, which neighborhoods get grouped together, which possible station sites are considered. That makes the final scores apples-to-oranges, and means a plan can "win" because of how it was measured rather than how good it actually is.

Julian Teusch and Oliver Keszöcze propose a fix called Freeze-Validate-Report. First you freeze the rules: the trips that matter, the neighborhood groups, the reporting areas, the list of possible station sites, and how ties get broken. Everyone agrees to this before seeing results. Then every plan is scored on the same common evidence: endpoint distance, meaning the walk from your front door to the nearest station plus the walk from the station to your destination. A five-part rule picks one plan for a single test report. In Porto, Portugal, it selected a fairness-oriented method called IFkCO, where the typical trip and the worst-off group's bad days averaged 805 and 872 meters of walking. In Chicago, it selected a simple grid pattern, at 1404 and 1656 meters.

Here's the catch, and it's a big one. The winner is fragile. Expanding the list of candidate station sites from 300 to 600 flipped the winning plan in both cities. Looking at a different hour on the same Chicago day picked a third plan. The authors are honest about this: they proved plans can be audited within agreed rules, not that a chosen plan will perform well once built. And the choice of which groups count, and which goals matter most, still belongs to the transport authority — a transparent audit of a biased goal is still a biased result.

To help outsiders check the work, the team released code, tamper-proof file fingerprints (hashes), decision logs and the proposal lists they used. It's a tool for scrutiny, not a magic answer.