Building and Refining — Verticality and the Urban Core

The first outing proved that my fieldwork idea could work. The next two made me earn it.

This time the goal was twenty parks split across two sweeps. One would take me through Glen Park, Sunnyside, Ingleside, and the Excelsior—the steep interior south of Twin Peaks. The other would begin with BART, continue through the dense downtown neighborhoods, and eventually work its way toward the Marina.

Together they replaced the pleasant abstraction of “collecting park photographs” with the more accurate version: climbing a great many stairs while carrying everything I needed for the day.

The city refuses to lie flat

A digital map has a way of making San Francisco look deceptively manageable. Two parks can appear to sit a few blocks apart, their markers separated by a tidy line. The map does not breathe heavily. It does not stop halfway up a concrete staircase and reconsider its choices.

South of Twin Peaks, the vertical dimension quickly took over. Streets bent around hillsides. Entrances appeared above or below the route that looked obvious. Pocket parks occupied terraces, and stairways functioned as streets. Reaching the upper edge of a site could require a slow climb before I had even started photographing it.

I am not an athlete. Unfortunately, the best view of a hilltop park tends to be near the top of the hill.

That became the rule: if a path continued upward, I followed it. A photograph from the entrance might confirm that a park existed, but the higher ground often explained why it mattered. The climb revealed how the space fit into the neighborhood, what the benches faced, where the paths led, and which parts of the city appeared between the houses.

Sutro Tower, again

Sutro Tower kept reappearing like a very large, very dependable trail marker. I used it to work out which slope I was on and how one park related to the next. Depending on the angle, downtown towers or a bridge might complete the picture. Fog moving over the ridge could erase it all again.

That constant triangulation made the limits of the municipal records more obvious. An address and coordinate can place a park on a map. They do not show the steepness of its entry steps, the retaining wall beside a path, a terrace cut into the hillside, or the view a bench was positioned to catch.

Those details changed how I photographed. I still wanted the satisfying wide view, but I also needed the unglamorous evidence: approaches, stairs, edges, paving, fencing, and the point where a park met the sidewalk. The goal was not to make every location look spectacular. It was to make each page more honest about the place.

A different rhythm underground

For the second sweep, I left the car behind and took BART into the urban core. The train carried me under the bay; from there, the day became a long chain of walking routes through Chinatown, the Financial District, North Beach, and toward the Marina.

The pace was completely different. In the dense center of the city, small parks and plazas arrived quickly. One might be a hardscaped gathering place tucked between buildings; the next, a playground behind a fence in a residential pocket. Chinatown supplied its own version of verticality, with alleyways and stair-stepped streets replacing the more open slopes from the previous trip.

Then the city widened toward the northern waterfront. Tight architectural frames gave way to lawns, shoreline edges, and long views across the bay. My photographic rhythm had to widen with it.

Traveling by transit also sharpened one of the application's practical questions: what does “near BART” really mean? A station can be the beginning of a trip without being the end of it. A park may still require a Muni connection or a substantial walk, and the conditions of that trip can change. Baker Parks could offer scheduled proximity and possible connections, but live arrivals, disruptions, and detailed accessible routing still belonged with the transit agencies.

That boundary eventually became part of the interface. The app helps someone understand the likely shape of a trip, then hands off to the official source for the part that changes in real time.

The fieldwork starts shaping the tool

Twenty more sites meant far more than twenty new rows in an image list. Every outing created new maintenance questions. Where should hundreds of large photographs live? How could I keep their order and attribution straight? What happened when a City coordinate put a park in the wrong place? How should temporary closures sit beside information intended to remain useful for years?

The application grew in response. I added a media workflow that keeps large files outside the code repository, a local editor for images and park notices, reviewed coordinate corrections, and shared page components that did not need to be reinvented for every destination. Area and neighborhood pages began to reflect the same geography I was learning on foot.

This was the point when fieldwork and development stopped feeling like separate phases. A difficult entrance suggested better visual documentation. A misplaced marker suggested a correction workflow. A long transit connection clarified what the page should—and should not—promise. Then the next version of the app changed what I looked for on the next walk.

By the end of the two sweeps, the routine no longer felt experimental. Drive or ride in. Aim for ten. Climb what needs climbing. Photograph the sweeping view and the useful detail. Come home, process everything, and improve the system before doing it again.

The project was getting stronger. My legs were being consulted less politely.