What the model misses
Our scores are an attempt to read 3,273 Toronto parks through a consistent structural lens — canopy, connectivity, enclosure, comfort, activity. Consistency is the point: it lets us compare parks fairly. But consistency comes at a cost. Many of the things that make a park actually feel like a parkcan't be derived from satellite imagery, OSM ways, or Google review counts. This page exists so we never let the score pretend otherwise.
Weather
An ideal April afternoon and a windy November Tuesday produce different parks. We don't sample weather; you do, the moment you walk in.
Seasonality
A ravine that's a brilliant dappled forest in August is a wet, leaf-strewn track in late October. Our canopy estimates don't capture phenology — they tell you what's potentially there, not what's there today.
Programming nuance
Two parks may host the same number of events per year, but one runs them at noon for tourists and the other at 7pm for the people who live across the street. Our event counters can't tell those apart.
Memory and emotion
The park where you and your kid used to feed ducks isn't measurably different from the park next door. It's different to you, and that matters.
Cultural history
Some parks carry deep meaning for specific communities — Grange Park for the AGO crowd, the Toronto Islands for Toronto's queer community, Sherbourne Common for newer arrivals to St Lawrence. We don't read history.
Time of day
Allan Gardens at 2pm is not Allan Gardens at 11pm. Our scores are not time-resolved. Volunteer observations can be — record the hour you saw it.
Invisible communities
Quiet uses that don't show up in Google reviews — meditation, anglers, weekday-morning seniors, recovery-group meet-ups — are present, but our signals undercount them. They are not less real for being quiet.
Informal uses
Birthday parties on the basketball court. Late-night skating on a closed pool. Communal garden plots. Pickup soccer leagues with no permits. None of this lives in any database we can read.
Park politics
Conflicts over off-leash zones, tent encampments, festival noise, BIA programming, condo shadow lines. These shape parks more than any structural metric. We can name disagreement; we can't adjudicate it.
Accessibility lived experience
We can sometimes measure ramps and surfaces, but we don't yet measure whether the actual experience of using a wheelchair, an ASL conversation, or a sensory-sensitive child works in this park. Volunteer observations begin to.
The view from inside
Almost all of our signals look at parks from outside — from the city's network, from photographs, from review aggregators. The view from inside the park, looking out at the people on the benches, is the one that matters most. That's what observation campaigns try to recover.
If a park you love scores low
That's a finding, not a verdict.
The model sees what it sees. If it scores a park you cherish at 38, that means our structural reading is missing something the park gives you. We want to know about it.
Three places to leave that signal:
- Provide feedback — if you study, lead walks, or design parks.
- The validation form on any park page — rate it on the dimensions you care about; we publish aggregates and divergence on /insights/disagreement.
- Volunteer observations — if you want to capture an actual hour at the park.