gap: #186 closed NDW on replacing the ambient cars, never on adding to them #254

New issue

Closed

opened 2026-09-01 01:40:08 +00:00 by viberfox-agent · 0 comments

viberfox-agent commented

2026-09-01 01:40:08 +00:00

Collaborator

Found while QA-ing what #186 shipped (see #251). To be clear up front: I re-measured #186's whole coverage argument and it reproduces exactly — this is not a challenge to the finding, it is about the question the record leaves open.

What I did

Verified the note's case, then asked the next question a maintainer revisiting the frontier row would ask.

The note rejects NDW under value 4, "add, don't replace", reasoning from the row's own wording — "real vehicle counts in place of invented ambient traffic" — that substituting measurement for the generated cars empties every street with no loop on it. That is correct and I confirmed it: 0 reporting loops within the 260 m placement radius at the Grote Markt, the Dam and Domplein.

But value 4 does not say "reject". It says add, don't replace. So: what about keeping all 180 invented cars everywhere and letting the real measurement inform them where a loop does exist?

What happened

The note never says. It closes on the replacement reading and the row moves to Checked and not taken, which reads as "NDW is closed".

The note's own measurements are the case for the additive version:

The A2 point south of Utrecht has 30 reporting loops within 1 km, 2 within 390 m.
Rotterdam Coolsingel has 2 within the 260 m placement radius — the note keeps this as "the honest counterexample".
The loops are "dense on motorway mainlines", and cartopolis_geo::traffic::drive_speed already gives motorway the top speed (26.0 m/s) — so the invented cars are fastest and densest exactly where the coverage is.
The note itself writes: "where a loop exists the measurement is better than the invention, and there is no loop where anybody is standing." The first half of that sentence is an additive feature and is left hanging.

The coverage objection does not reach this shape at all, because nothing is emptied — the generated cars stay everywhere, and measurement only modulates them where it exists.

What a user would expect instead

One of two things, neither of which is a build:

A sentence in "Which value settled it" closing the additive option explicitly — e.g. that a national 52 MB re-slice every minute is too much to pay for motorway density alone, which is a fair answer the note has the material for. Or
A candidates row for the additive version, so it is asked once more with a narrower scope.

As written, the next person who asks "could the motorway traffic be real?" finds a note that answers a different question and has to redo the whole check — which is the exact cost the "Checked and not taken" table exists to prevent. The note is strong enough that this is worth one sentence rather than a re-run.

Where the seam is

docs/notes/ndw-traffic.md, the "Which value settled it" section, and the NDW row in docs/direction.md.

What I verified while here

The layer this decision preserves does work: a headless capture over the Grote Markt (--at 53.2194,6.5665,200 --look=-30,0 --dump-state) reports traffic_cars=180, traffic_walkers=220 — the full pool placed and visible.

Found while QA-ing what #186 shipped (see #251). To be clear up front: I re-measured #186's whole coverage argument and **it reproduces exactly** — this is not a challenge to the finding, it is about the question the record leaves open. ## What I did Verified the note's case, then asked the next question a maintainer revisiting the frontier row would ask. The note rejects NDW under **value 4, "add, don't replace"**, reasoning from the row's own wording — "real vehicle counts *in place of* invented ambient traffic" — that substituting measurement for the generated cars empties every street with no loop on it. That is correct and I confirmed it: 0 reporting loops within the 260 m placement radius at the Grote Markt, the Dam and Domplein. But value 4 does not say "reject". It says *add, don't replace*. So: what about keeping all 180 invented cars everywhere and letting the real measurement inform them where a loop does exist? ## What happened The note never says. It closes on the replacement reading and the row moves to **Checked and not taken**, which reads as "NDW is closed". The note's own measurements are the case for the additive version: - The A2 point south of Utrecht has **30 reporting loops within 1 km**, 2 within 390 m. - Rotterdam Coolsingel has **2 within the 260 m placement radius** — the note keeps this as "the honest counterexample". - The loops are "dense on motorway mainlines", and `cartopolis_geo::traffic::drive_speed` already gives `motorway` the top speed (26.0 m/s) — so the invented cars are fastest and densest *exactly where the coverage is*. - The note itself writes: "where a loop exists the measurement is better than the invention, and there is no loop where anybody is standing." The first half of that sentence is an additive feature and is left hanging. The coverage objection does not reach this shape at all, because nothing is emptied — the generated cars stay everywhere, and measurement only modulates them where it exists. ## What a user would expect instead One of two things, neither of which is a build: 1. A sentence in "Which value settled it" closing the additive option explicitly — e.g. that a national 52 MB re-slice every minute is too much to pay for motorway density alone, which is a fair answer the note has the material for. Or 2. A candidates row for the additive version, so it is asked once more with a narrower scope. As written, the next person who asks "could the motorway traffic be real?" finds a note that answers a different question and has to redo the whole check — which is the exact cost the "Checked and not taken" table exists to prevent. The note is strong enough that this is worth one sentence rather than a re-run. ## Where the seam is `docs/notes/ndw-traffic.md`, the **"Which value settled it"** section, and the NDW row in `docs/direction.md`. ## What I verified while here The layer this decision preserves does work: a headless capture over the Grote Markt (`--at 53.2194,6.5665,200 --look=-30,0 --dump-state`) reports `traffic_cars=180`, `traffic_walkers=220` — the full pool placed and visible.

viberfox-agent added the

qa-gap

label

2026-09-01 01:40:08 +00:00

viberfox-agent added this to the Viberfox project

2026-09-01 01:40:15 +00:00

viberfox-agent referenced this issue

2026-09-01 01:42:39 +00:00

QA: what #186 shipped — NDW traffic: is it worth wiring in? #251

viberfox-agent referenced this issue

2026-09-03 00:56:45 +00:00

The QA lane has never walked docs/qa/targets.md — 14 of 14 passes re-verified a shipped ticket, 9 of them documentation-only verdicts #280