Richmond data projectWhich Richmond greenways are open end to end? Tests before the AI draft
Metro Vancouver publishes its Regional Greenway Network 2050 as open data. The file splits each route into segments and gives every segment a status. Our learners ask one question of the Richmond rows, write the tests first, and then watch an AI-style draft fail them. The code and every count below are ours, run on the file downloaded on 23 October 2026.
Richmond segments in the Regional Greenway Network 2050 file, by status (our count of the 51 rows tagged Richmond; the status meanings are Metro Vancouver's own wording, shortened)| Status in the file | What Metro Vancouver says it means | Richmond segments |
|---|
| Operational - Class 1 | “complete routes that are available to the public for walking and cycling” | 23 |
| Operational - Class 2 | “interim routes” that “do not meet the 3m wide width goal and/or are not separated from traffic” | 13 |
| Planned | Future routes already in providers' plans, still needing planning, land, design, funding or construction | 12 |
| Proposed New | Conceptual routes proposed to fill gaps in the network | 3 |
The question: which named Richmond routes are open along their whole length? The 51 segments carry only 13 route names, and one of them, Steveston Greenways / South Dike / East Richmond Trails+, is split into 24 segments: 11 in Class 1, 6 in Class 2, 5 Planned and 2 Proposed New. That single route is where a hasty answer goes wrong.
Here is the whole exercise as one Python file. It uses only the standard library and runs beside the downloaded JSON:# Tests first, then the AI-style draft, then the fix (Modern Age Coders, written for this page)
import json
DATA = json.load(open("mv_regional_greenway_network_2050_all.json", encoding="utf-8"))
SEGMENTS = [f["attributes"] for f in DATA["features"] if f["attributes"]["Municipality"] == "Richmond"]
# Step 1: the learner writes the tests before asking for any code. A tiny route of our own invention:
FIXTURE = [{"Name": "Test Route", "Status": "Operational - Class 1"},
{"Name": "Test Route", "Status": "Planned"}]
def run_tests(open_routes):
results = []
results.append(("half-built test route is not open", "Test Route" not in open_routes(FIXTURE)))
every = open_routes(SEGMENTS)
results.append(("open routes are a subset of route names", set(every) <= {s["Name"] for s in SEGMENTS}))
results.append(("no open route has a Planned segment",
all(s["Status"].startswith("Operational") for s in SEGMENTS if s["Name"] in every)))
return results
# Step 2: the draft, the kind an assistant returns first. It reads well and is wrong.
def draft_open_routes(segs):
return sorted({s["Name"] for s in segs if s["Status"].startswith("Operational")})
# Step 3: the version that passes, written after a test failed.
def open_routes(segs):
names = {s["Name"] for s in segs}
return sorted(n for n in names if all(s["Status"].startswith("Operational") for s in segs if s["Name"] == n))
for label, fn in (("draft", draft_open_routes), ("fixed", open_routes)):
print(label, [("PASS" if ok else "FAIL") + ": " + name for name, ok in run_tests(fn)])
What the run printed (our output, 23 October 2026)| Version | Test 1: half-built route | Test 2: real names only | Test 3: no Planned segment |
|---|
| AI-style draft | FAIL | PASS | FAIL |
| Fixed version | PASS | PASS | PASS |
Why the draft fails
It keeps a route if any one segment is operational. In Python that is the difference between any() and all(), and it is the most common shape of mistake in AI-written data code: correct syntax, wrong meaning. The draft names 10 routes; only 6 are open end to end.
What the fix returns
Six routes have every Richmond segment operational: Cambie Skytrain Bridge, Delta South-Surrey Regional Greenway, Knight St. Bridge, Oak St Bridge, Railway Ave. Greenway and West Dike Trail. The four the draft added wrongly include the 24-segment Steveston route and the Sea Island Recreational Trail.
Why test 1 uses made-up data
The two-row Test Route is ours, invented on purpose. A small case whose right answer you know by heart catches a wrong idea faster than a 51-row file whose answer you are still trying to find.
Three things the file cannot tell you
Three segments, OBJECTID 618, 812 and 1005, were Operational in the PreviousStatus column and are Planned now; the file does not say why, so we do not guess. Six rows carry the placeholder NA as their route description. And Metro Vancouver says the dataset reflects progress between 2020 and 2023, so a route opened since may still read Planned here.
The file is Metro Vancouver's, not the City of Richmond's: we could not find a City of Richmond dataset published under an open licence, so we used the regional district's, filtered to the rows tagged Richmond. Some of those rows are bridges at the city's edge, which is why we count routes and never print a total length. Used under the Open Government Licence - Metro Vancouver, which allows use including for commercial purposes (licence). Contains information licensed under the Open Government Licence - Metro Vancouver. Metro Vancouver does not endorse this page or our classes.
