Toronto’s ravine paradox

Toronto has more park area than most large North American cities, and most of it is in the ravines. The Don, the Humber, Highland Creek, the Rouge: they thread through the city like green capillaries, and the Toronto Park Catalogue counts every one of those parcels. About 800 of the city’s 3,273 mapped parks classify as Ravine / Naturalized; another 16 large parcels classify outright as Wilderness / Conservation. Their natural-comfort scores are routinely above 90. Their connectivity scores are routinely below 15.

On the model’s urban-vitality terms (connectivity, edge activation, enclosure) most of these parks are profoundly weak. They are surrounded by ravine slopes, not streets. They have one or two access points. Nothing commercial fronts them. The urban form Jacobs cared about is, by design, almost entirely absent.

That tension is so consistent across the Toronto Park Catalogue that it produces a visible structure on the two-axis chart:

Almost every ravine point sits in the upper-left quadrant: high comfort, low integration. This is the pattern we name the ravine paradox. The same conditions that make these parks valuable as ecological retreats make them inaccessible as everyday urban places. They’re what Hugh Brady called “parks for ducks, not people”, at least if “people” means residents using a park casually on a weekday.

What the data picks up

The metric model isn’t hostile to ravines; it just measures things that ravines aren’t. When we run the “What would improve this park?” logic against a typical ravine parcel, it suggests improving edge activation and connectivity. Both would, in practice, undermine the very thing the parcel does well. This is the model misapplying its lens.

The right reading is that the Jacobs framework and the conservation framework are answering different questions. Connectivity-driven ranking puts ravines near the bottom; comfort-driven ranking puts them at the top. Neither answer alone is useful. The point of the two-axis chart is to make that tradeoff visible rather than collapse it into a single ranking.

The ravine-edge disconnect

A subset of these parks does deserve a closer look. The model flags any ravine park with both very high natural comfort and very low edge activation, a condition we call the ravine-edge disconnect. These are ecological assets that don’t quite reach the people who live next to them. Examples from the current data:

The intervention here isn’t to urbanise a ravine. It’s to find the ravine’s edges, the places where the slope meets a street, and ask whether a single mid-block path, a sightline, a bench, or a small entrance plaza could let the ravine reach a few thousand more people without compromising what makes the ravine work in the first place.

The model can identify which parks have this condition. It cannot tell you whether any specific intervention there would be welcome. That second step is design and politics, not data.