Contested parks
Parks where public sentiment most disagrees with the model. We collect simple thumbs-up / thumbs-down feedback on every park detail page; here we surface the parks where the up/down split is closest to 50/50 (i.e. opinion is genuinely split). The goal is to help us see where the model is missing something the metrics can’t pick up: programming, perceived safety, time-of-day character.
No contested parks yet
Each park needs at least three votes before it can land here. As people vote on the detail pages, this list will populate with the parks whose up/down split is genuinely close.
Why this page exists
Algorithmic urbanism has a real failure mode: it can score a park well on every measurable dimension and still be wrong about how that park actually feels. Programming, time-of-day character, social composition, perceived safety, the texture of the people who use the park: these don’t show up in OSM tags or building footprints. Public votes are a thin signal but a real one.
We treat this list as a backlog of cases the model should learn from, not as a verdict.