The blocking gate is CLOSED, and the epoch closes green.
- #1b THE BLOCKING RE-MEASURE: F's R21 re-measure made the tri diet blocking; B diagnosed layer-by-layer,
the driver traced A -> E (A's own commit: "B's numbers say the fix is E's"), E fixed it, F re-measured:
katoomba venue_night 221,935 (+11%) -> 108,365 (-51%); bendigo 201k -> 124,682; fremantle -> 88,397.
Worst 124,682 — the 200k stress edge is closed. The measure-first loop worked end to end.
- #3 DBG.enterShop(selector) (Lane G's ask): id · registry type · exact name · name substring, resolved
most-specific-first so an id-shaped string can't be shadowed. One-liner QA on any of 23 towns.
- #4 The release tour -> docs/shots/release_v4/: the Fremantle hero · the Fitzroy mecca · the Toowoomba thin
tail · the synthetic control · the ?classic=1 covenant. Every real-town frame poses through F's R21
cluster primitive; B's R22 bookmarks anchor on the same retail heart, so tour and bookmarks agree by
construction. HUD hidden throughout (F's R21 ruling corollary).
The primitive earned itself twice: the first cut posed via street_noon and the MECCA CAME OUT AS BARE
BITUMEN — the same trap as R21. Re-posed through the cluster, Fitzroy reads.
- HONEST FINDING (filed v4.x): the brief asked for the thin tail as "proof the floor holds". It does not,
and the frame says so — Toowoomba reads as an empty road. The floor is a RATIO, not a count: 0.008
shops/edge vs the heroes' 0.06-0.08. Shipped labelled honestly rather than swapped for a flattering town.
- #6 Docs: README v4 section (the towns ARE the product), LANE_F_NOTES §22, F-progress.
qa.sh --strict --matrix GREEN 7/7; MATRIX GREEN — 10 towns × 7 gates (the budget gate green everywhere for
the first time); selfcheck 161,300/161,300 (A's 36 goldens + the 23-town pack).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Wave 3 — ran after A pinned the 36 goldens (3face97) and E's tri-diet fix (d81e499).
#5 — 23-TOWN DOOR-SOAK: five-town harness scaled (~10 doors/town; sample takes one of every
type present then stride-fills). 238 doors / 1,133 in-plan shops / 23 towns.
23/23 PASS — 0 throws / 0 carves / 0 path-fails / 0 over-budget; worst 159 draws (fitzroy)
<=350. All 12 archetypes covered town-set-wide (the whole registry). 23/23 towns landed 3
venues — A's cluster bias + district placement hold from the 12-shop thin tail (darwin,
toowoomba) to the 139-shop mecca (fitzroy, worst 159 = same band as everywhere: interiors
don't care about town scale). Lane G's newtown_godverse passes (18 shops, worst 134).
#3 — G2a PER-SHOP ATLAS CONTRACT (C's half), published in LANE_C_PUB §7. All three Fable priors
hold; atlas-per-shop is strictly BETTER for draws than today's type-packs. The binding constraint
is VRAM, not draws — E's dispose ask GRANTED, it's my seam:
- Draws: 1 atlas/shop = 1 draw (shared material + baked UV -> batch merges). Cap 2; 60+ item
shops spill to the type-pack. Headroom huge (stock rooms 41-58 vs the <=350 law).
- VRAM: E's 181MB audit is right, and the cause is my own design note (stockpack.js:11 caches
atlases "NOT per-room, like depot GLBs") — correct for 3 bounded type-packs, wrong for
unbounded per-shop. Ruling: <=2 resident atlases, <=8MB decoded, <=1024" typical; C lands the
LRU cap + dispose in stockpack.js. Pre-existing + ?stock=real is opt-in (index.html:196) so
it can't touch the default boot -> NOT a v4.0 blocker; C's first v4.x item, before G2a ships.
- Schema: E's index confirmed verbatim, keyed per shop; UV top-left (C flips V); C reads uv not
cell. Fail-soft/parody/?noassets unchanged. Ask to G: emit items[] in deterministic id order —
C's seeded per-bin pick indexes into it.
C->F: 23-town interior verdict GREEN. Docs-only; both C items closed for v4.0.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Against A's pack absorb (3face97, 23 towns / 36 goldens). selfcheck ALL GREEN
161300/161300, synthetic 0x3fa36874 frozen.
FITZROY STRESS: PASS — the sim doesn't notice 2x scale. 160 shops -> 139 seated (21
dropped, counted), 1859 nodes / 2408 edges (biggest of the epoch), 376 active peds,
0 NaN, 0.08 ms/frame, 22 chunk rosters, 454 identity sigs, 103 patronage visits.
A's R21 cluster bias survives the mecca (min 8 shops within 160m). Nothing needed
touching; the citizen field is not a budget risk at any scale this epoch reaches.
THIN TAIL — "does 12 shops still read alive?" ANSWER: only if they CLUSTER. I checked
both 12-shop towns because they disagree, and the disagreement is the finding:
- darwin_real ALIVE: 27m median nearest-neighbour spacing, 7 of 12 shops within 160m of
the hub, 18 visits (13 hero).
- toowoomba_real DEAD: 388m median spacing (14x) across a 4.6km road network, only 3
shops within 160m of its own densest point, 1417 patronage checks -> 0 finds, 0 visits.
Not a shopping town — a road network with occasional shops.
Verified real, not an R20-style sampling error: 9 shops open at 14:00 in BOTH, camera
stood on the town's own densest point. 0 NaN / 0.05 ms/frame either way.
FILED for E/Fable (curation, not code): MIN_TOWN_SHOPS counts shops, but what makes a
town read alive is clustering. A cheap test that separates them: ">=5 shops within 160m
of the densest point" — darwin 7 PASS, toowoomba 3 FAIL. Toowoomba ships in the v4.0
pack today with no main street to walk. Cull or accept as a known thin-tail (it costs
nothing). E/Fable's call — A's venue bias correctly exempts it ("no better candidate").
D->F: citizens at pack scale GREEN, no D-side blocker for v4.0. Toowoomba is curation,
not a citizen defect — the tag isn't gated on it from me.
Carried to v4.x: GIG_RANGE 34m radius vs single-chunk lookup (8.7% chunk-blind).
Last flag before the retro: R20's main-street relocation half remains unowned (fitzroy
13 hero vs 90 texture visits); one call of this harness re-runs it if v4.x wants it.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
B's diagnosis nailed it: 149,161 of 163,015 tris (91.5%) at katoomba venue_night were four of my GLB
street props. Buildings 348, posters 222 (A's cap worked, never the lever), pools 0. B's sharpest point:
it is NOT the density widening -- road density drives the bench/bin/shelter cadence (~100 props on a
966-edge real graph).
Fix (collapse-decimate via the R13 instrument tool; originals kept as _hi):
bench 1942 -> 500 rendered 60,202 -> 15,500
bus_shelter 7998 -> 2500 rendered 55,986 -> 17,500
bin 3000 -> 800 rendered 18,000 -> 4,800
food_cart 4991 -> 1000 rendered 14,973 -> 3,000
furniture total 149,161 -> 40,800 (-108,361)
Projected view ~54.7k (B's 163,015 baseline) / ~113.6k (vs F's 221,935) -- PASSES the 200k gate either
way with 86-145k spare. F: these are projections off B's per-instance attribution; yours is the real
measure. -> F: RE-MEASURE.
Eyeballed every prop; two needed re-cutting: bus_shelter @1000 shredded (thin roof plate on a
voxel-remeshed shell) -> @2500 clean; bin @400 creased -> @800 clean. food_cart's black base slab is in
the SOURCE model (confirmed against the pre-diet thumb from git), not a decimation artifact.
Zero blast radius: dims/footprints/heights identical -> manifest.json byte-identical -> B's placement and
A's goldens untouched. Published 4 diet + 4 _hi archives to the depot (John re-authorized explicitly;
GLBs are git-ignored so the depot IS the archive).
-> Lane B (your 2nd note, cause found, NOT fixed by design): the MeshPhysicalMaterial on bench+food_cart
is KHR_materials_specular in the glTF, and it survives a Blender re-export. Fix = neutralise specular
pre-export. Deliberately deferred to v4.x: the gate is tris (cleared), and changing shipped shading in
a release round without a visual re-verify is risk for no gate benefit. Same list: food_cart is 3.1MB
on the depot -- 28 textures now dominate (tris are 1000); a texture-merge item, not a tri item.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Docs-only handshake close; no code, no goldens. Filed so F isn't waiting on A for the re-measure.
My R22 notes left ledger #1's ownership open on purpose: "if B's layer-by-layer names plan geometry
(lot or poster counts) as the venue_night driver, I own the fix and will take it." B's diagnosis
(0c4dad5) has landed and answers it — the driver is E's four furniture GLBs at 149,161 of 163,015
tris (91.5%), which is ASSET TRIS, so LANE E owns the fix by the ledger's own rule. A IS CLEAR.
Plan geometry isn't close to the bill:
posters 222 tris <- B: A's cap "already worked and was never the lever"
buildings 348 tris <- the R3 atlas/instancing holding
venue pools 0
Recording that B corrected MY expectation too, not just the brief's. I predicted the driver would be
buildings/LOD ("more shops => more night building geometry at the venue block"); B measured buildings
at 0.2%. And B found the real coupling: shop count doesn't drive this, ROAD density does — the
bench/bin/shelter cadence rides a 966-edge graph, so ~100 GLB props land in the streamed set where
synthetic had a handful. That means my R20 absorb of E's density widening was never the cause either.
The brief is a hypothesis and the measurement wins — mine included.
Lane A's round-22 ledger is now complete: #2 the pack absorb (3face97), #6 the docs close (9c9ad73),
#1 measured out of my court. Standing by for F's close.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
ROOT CAUSE: every "nearest" resolver in dbg.js sorted by distance from the WORLD ORIGIN. That's the town
centre only on a synthetic plan -- a real OSM town is projected about its cache's bbox centre while its shops
cluster hundreds of metres away, so bookmarks teleported into arbitrary far-flung geometry, exactly as G
reported. Measured origin-pick distance from the retail heart: katoomba 408m, newtown 626m, fremantle 371m.
FIX: resolve bookmarks about the town's DENSEST SHOP CLUSTER (grid-bin shop lots at 120m; heaviest bin's
centroid = the retail heart) instead of the origin -- byDist, lotsByDist, streetViewPose, stopPose, and
crossroadsPose's tie-break.
GATED on the REAL-ROADS signal (edges > 200) -- the same primitive as the R20 tram fence -- so synthetic and
the marched fixtures keep their exact origin-centred behaviour and F's release-tour control shot cannot move.
The gate is load-bearing, not cosmetic: synthetic's anchor is only 117m out but it DOES re-pick (Anglibarn Op
Shop -> Dusty Pages), so an unconditional switch would have silently moved the control. Same reason the new
crossroadsPose tie-break is gated (a degree tie would otherwise resolve differently on synthetic).
Verified across the town set: synthetic(22 edges)/melbourne(15)/katoomba-fixture(15)/silverton(1) -> origin,
UNCHANGED; newtown_godverse(1626) -> anchor(98,-629); katoomba_real(993) -> anchor(-227,-625). G's own town
is fixed.
Result on katoomba: shopfront bookmarks went from ~408m adrift to <=42m of the retail heart (arcade 7m,
milkbar_dusk/patronage_door/shopfront_detail 24m, market_square 26m, dig_real 42m); strip views ~195m on the
high street; venue_night/queue_night ~345m (A's venue placement, not the bookmark). All 19 resolve, 0 errors;
synthetic 19/19 unchanged. warehouse_fringe sits 3,074m out -- correct by design (it's the far:true fringe
shot; on a 5km town the fringe genuinely is 3km away).
-> F (ledger #4, the shared posing primitive): published on the harness as DBG.townAnchor {x,z,n,anchored,ref}
(the retail heart + whether it's in use) and DBG.clusterPose(kinds) (stand on the road at the densest
cluster, look along it). Your cluster-pose finding as a dbg.js-consumable -- the tour and the bookmarks now
resolve through the SAME primitive, so a tour shot and its bookmark can't drift apart.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Docs-only; no output change, no golden moves (selfcheck 161,300/161,300 green before and after).
THE DRIFT THIS CLOSES. Three gig-layer behaviour changes had reached only my lane notes, never the
spec — the amendment law says the spec is where they live:
- ROUND20 ledger #2: the R16 facade-clearance bias WIDENED from band_room/rsl to every venue kind (line
312 still claimed band_room/rsl). A real road graph at E's density can put a PUB's facade over a cross
street; synthetic pubs always clear, so it stayed a no-op there.
- ROUND21 ledger #1: the venue cluster bias was never written down at all. Now recorded with the
measurement (11 of 15 venues under the floor -> 0; Newtown's RSL 2,228m -> 29m from the hub) and the
by-construction gate (, which synthetic never sets — measured, not assumed).
- ROUND22 ledger #2: the graded fallback (densest-available where the floor is unreachable).
Also documented the pick ORDER, which no lane could have inferred: cluster field -> kind flavour ->
clearance filter -> farthestFirst -> seeded pick, each degrading to the previous set.
CITY_SPEC v4 section marked FINAL at the tag: the alpha's five towns became the epoch's 23 (1,210
shops) and every ruling held with no new hardening. 'All five towns' now reads as the measurement of
record when the ruling was written; the ruling is pack-wide.
THE RISK LIST, line by line, judged against the full pack rather than the 5-town sample it was written
for (V4_REAL_MAP.md):
1. RETIRED — bbox law enforced with headroom: widest is geelong at 3,829m, 77% of the 5km ceiling.
2. RETIRED — the floor never bound: thinnest is darwin at 12 shops, 2x MIN_TOWN_SHOPS. (Whether 12
shops READS alive is D's thin-tail check, not a generator risk — different question, still live.)
3. RETIRED plan-side; CARRIED for the runtime lanes. The alpha's real risk paid off exactly as
intended (it surfaced fragmentation, capacity, stranding — each fixed-or-fenced). But a green plan
is not a green town: C's soak, D's fitzroy audit and F's matrix are the verifies that close it.
4. RETIRED — the DP knob is at a measured setting: 1 component on all 23.
5. Unchanged, now on 23 towns not 5.
6. NEW, CARRIED — thin-graph seating capacity: 6.4% of shops don't seat (77/1,210), all counted by
cause; daylesford 19% (32 shops on 171 roads), fitzroy 13%. Levers are spent at RGAP 0.5m /
NODE_CLEAR 3m; the only one left puts a shop on a road it isn't on. R20 ruling stands: drop and
count beats teleport. Carried into v4.x measured, not hidden.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
E's 17 new towns + Lane G's newtown_godverse fed through the lift. ONE batched golden commit as
briefed: REAL_TOWN_GOLDENS + REAL_TOWN_GIG_GOLDENS go 5 -> 23 towns each, zero unpinned.
THE HEADLINE: everything built this epoch held at 23-town scale, FIRST TRY. 1,210 shops -> 1,133 seated
(77 dropped, 6.4%, all counted by cause); 1 CONNECTED COMPONENT and 3 VENUES on every one of the 23 towns.
fitzroy — 160 shops, 2x anything absorbed before — landed 139 seated, 1 component, 3 venues, 212 posters,
all invariants green. The junction-protected DP, fragmentation ruling, cluster bias, poster cap and
capacity fence needed NO new hardening: the pack needed absorbing, not fixing.
- Graded cluster fallback (the one refinement). Three towns have no candidate at the floor at all
(ballarat/toowoomba/newtown_godverse top out at 2 shops within 160m); R21's bias switched off entirely
there, stranding venues at 0. It now falls back to the DENSEST AVAILABLE — ballarat 0,2,0 -> 2,2,2,
godverse 0,1,1 -> 2,2,2, toowoomba 1,0,0 -> 1,2,1. Fires only where the floor is unreachable, so NO
already-pinned town moved (verified). Acceptance at pack scale: 20 of 23 towns have every venue >=3;
the three above sit at their town's genuine ceiling — the carved-out case, stated explicitly.
- Drop outliers, counted not hidden: daylesford 19% (32 shops on a 171-road graph, the thinnest in the
pack, all overflow), fitzroy 13%. RGAP is already 0.5m and NODE_CLEAR 3m; the only lever left is
spilling overflow onto a DIFFERENT street, which buys a few percent by putting a shop on a road it
isn't on. Same ruling as R20: drop and count beats teleport.
- selfcheck ALL GREEN 161,300/161,300; scaffold + consistency GREEN; cross-process deterministic
(pack-xor 0xca140954). classic 0x3fa36874 / gig 0xb1d48ea1 / marched fixtures FROZEN.
Handshake -> C + D + F: the pack is in, 23 towns pinned, start your verifies. Tri diet (ledger #1): if
B's layer-by-layer names plan geometry as the venue_night driver, I own the fix.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Ledger #3 (E's G item), #6 (provenance prep), + the C/E contract review.
- MODELBEAST notes (#3): pipeline/README.md's toolchain row claimed sf3d/trellis_mac were 'HF-gated
(setup needed)' -- ultra was right, that was stale (weights cached; trellis_mac 15/15 on thriftgod
Tier-1; the :8777 queue also runs bg_remove_local/mflux_image_edit/seedvr2_upscale). Corrected.
Ultra's finish-farm post-mortem folded into the props section VERBATIM + credited, with Fable's
Ruling 1 amendment (NO-DRACO: drop --compress draco, rest of the command stands). Noted that our
bake_lowpoly.py / decimate_props.py are NOT the finish farm, so nobody conflates them.
- Provenance v4-FINAL (#6): the frozen table carries the pack (22 towns, all 8 states/territories,
1192 shops + roads[], ODbL, raw snapshots) + a row for Lane G's newtown_godverse attributed to G.
The licence spine is complete -- nothing shipped is unaccounted.
- BUG: index.json still listed 22 -- generated before the dd4489c merge, so newtown_godverse was
invisible to B's selector while R22 verifies at 23-town scale. Regenerated -> 23 towns (selfcheck
ALL GREEN 161300/161300). Index is now self-describing: state reads from the cache first, then my
config, so any lane's town rosters itself.
- G2a atlas contract review (E's half; C answers interior ceilings in LANE_C_PUB): Fable's priors all
hold. Atlas-per-shop is an UPGRADE if the loader disposes on room exit -- today stockpack.js never
disposes (R9: 181MB resident, accreting). Without dispose, per-shop strands ~16MB/shop (~190MB per
town visited, unbounded across 23 towns). With dispose: ~4MB resident at 1024², ~45x better than
today. Also: size the atlas to the shop (<=1024², not 2048² for a 6-item shop); reuse my index
schema verbatim so stockpack.js/fail-soft-to-parody/?noassets/validate_pack all work unchanged;
the R8 parody law (parodied metadata, real cover art) is unchanged.
-> Lane G: add "state":"NSW" to newtown_godverse.json and the index picks it up automatically.
-> Lane C: the R9 dispose-on-exit cache-cap is now load-bearing for G2a (my VRAM numbers above).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
BLOCKING-ITEM DIAGNOSIS, filed first so E/A/F aren't waiting. Method: katoomba venue_night, settled, then
hide-a-layer-and-re-measure (frustum-accurate: a layer's cost = the tris that vanish when hidden), then the
same drill inside furniture bucketed by geometry.
VERDICT: the four GLB street-furniture props are 149,161 of 163,015 tris = 91.5% of the stress view.
OWNER: LANE E (asset tris), per the ledger's own rule. Decimation clears the gate with ~120k to spare.
Layer breakdown (baseline 163,015 tris / 88 draws):
chunk-furniture 151,341 (92.8%) | ground 9,932 | venue-presentation 770 | lighting 624 |
chunk-buildings 348 (0.2%) | venue-posters 222 | citizens/weather/tram/venue-pools 0
Buildings are 348 tris (the R3 atlas/instancing holds). Posters are 222 -- A's cap already worked and was
never the lever. Pools cost 0.
Inside furniture -- the four GLBs are the whole bill:
bench 1,942 tris/inst x64 = 60,202 (37%) <- procity_street_bench_01
bus_shelter 7,998 x10 = 55,986 (34%) <- procity_street_bus_shelter_01
bin 3,000 x19 = 18,000 (11%) <- procity_street_bin_01
food_cart 4,991 x8 = 14,973 (9%) <- procity_street_food_cart_01
pole/lamp/trees/novelty (primitives) x303 = 2,180 (1.3%)
Identity confirmed by DCC-authored material names (wire_143225087 = a Sketchfab scan, baked_mat,
"2 wood frame ", Material_0) -- E's GLBs, not our procedural materials. Two use MeshPhysicalMaterial.
WHY NOW (and why it is NOT the density widening): shop count doesn't drive this -- ROAD density does. Real
towns put the bench/bin/shelter cadence on a 966-edge graph, so ~100 GLB props land in the streamed set
where synthetic had a handful. A 1.2m bench at 1,942 tris is ~5x the entire building stock in frame. This is
B's own R6 note coming due ("street_bin 3.0k tris... a decimate to ~500 would give back tri headroom if the
combined-lane budget tightens").
-> E, suggested targets: bench 1942->~250, bus_shelter 7998->~600, bin 3000->~200, food_cart 4991->~400.
149,161 -> ~29,000 (saves ~120k). Projected: my 163k -> ~43k; F's 221,935 -> ~102k. Both far under 200k.
Cheap extra: drop the two MeshPhysicalMaterials to MeshStandardMaterial.
-> Fable: a B-side fallback exists (furniture.js already has a primitive fallback for every GLB prop via the
R4 use-if-ready path), so a distance-gated LOD is a contained B change I can land fast. NOT doing it
unilaterally -- per the ledger this is E's, and source decimation helps every town/view, not just this one.
-> F: my baseline is 163,015 vs your 221,935 on the same view -- my automation's rAF is throttled so the live
gig crowd never spawns (citizens measured 0). Furniture is still ~68% of your number; decimation clears the
gate on either baseline. Re-measure against the live crowd, not my figure.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
All five lanes in (A's venue cluster bias, B's selector-from-index, C's density door-soak, D's relocation
verify PASSES, E's 22-town pack non-gating); F gated and tagged.
- town_matrix.py: a `tram` dimension — B's R20 shop-adjacency ruling verified EXACTLY (newtown runs 16 stops
after B's phantom-stop fix, castlemaine 10; katoomba/fremantle/bendigo FENCED at 3/2/1 shops fronted, all
below the 5-shop floor). 10 towns x 7 gates: boot/determ/district/roads/tram/noassets green on all ten.
- index.html: expose window.PROCITY.tram — B built `routeInfo` FOR F's smoke but nothing published it; F owns
the shell, so F wired the seam.
- RECORDED (ledger #6b — the instruction was "record the new number"): the venue_night re-measure. A's cap
WORKED (posters 438->111) but E's density widening ate it — katoomba 200,965 (+965, R19) -> 221,935
(+21,935, +11%); bendigo 201k (+0.5%); fremantle/newtown/castlemaine in budget. Draws fine everywhere
(151/300 — half budget); tris-only, at the worst night stress view. Filed as R22's first item, where the
tri-diet backlog (instrument LOD, tram tri diet) already lives. katoomba is the outlier: 966 edges, 72 shops.
- Ruling (ledger #6d): the town selector stays ALWAYS-IN-HUD, not ?dbg-gated — THE REAL MAP's product IS the
towns. Corollary (F's own gap): money shots hide the HUD; the R19 alpha shot had the selector in-frame
because F's harness screenshotted the full viewport.
- Beta money shot: Fremantle's real street at golden hour — a corridor of shopfronts, footpaths, a crossing,
gum trees, and a genuinely busy crowd (the density widening, visible). 126d/167k tris, HUD hidden.
- docs: LANE_F_NOTES §20 (the honest why-R20-had-no-tag record) + §21, README beta paragraph, F-progress.
qa.sh --strict GREEN; selfcheck 154141/154141 (A's re-pinned goldens hold).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Verified A's venue cluster-adjacency bias (110c828), assigned from my R20 relocation
finding. selfcheck ALL GREEN 154141/154141 against A's re-pinned goldens.
ACCEPTANCE PASSES ON ALL FIVE (no venue with <3 shops in its 160m radius):
katoomba min 9 | newtown min 5 | bendigo min 5 | castlemaine min 4 | fremantle min 3.
Castlemaine's floor did NOT need an exemption.
MY HEADLINE FINDING IS CLOSED: katoomba's pub — "a gig with no passing crowd", 1 shop
at 1007m — now sits at 466m with 9 shops. Newtown is the dramatic fix: all three venues
were 2.2-2.4km out with 0-1 shops; its rsl now sits 29m from the hub with 11.
THE CROWD FOLLOWED (katoomba pub, same density/seed/21:00): passing crowd 34->66 (2x),
patron checks 226->442 (2x), inside frames 1042->1424, roster 2->3, 0 NaN.
TWO READINGS THAT NEED CARE:
- fremantle is the weakest pass: pub/rsl clear the >=3 bar but moved FURTHER from the
primary hub (969->1352m, 561->1292m) — A's bias optimises local cluster adjacency and
they found a secondary cluster. Letter satisfied, crowd healthy (pub roster 6, inside
2741), so not a miss — just named. A's call if they want it tighter.
- venue-win % DROPPING is not a regression (newtown rsl 100%->21%). It was only 100%
because the venue was alone; post-bias it has neighbours competing for finds while
total crowd rose. Isolation flatters that metric — ignore it post-bias.
D->F: relocation verify GREEN, no gate concern for v4.0-beta.
FYI F: selfcheck scans every towns/*.json on disk incl. E's UNTRACKED pack WIP — I saw
it transiently RED (newcastle_real, no open-late shop) then green minutes later. If you
gate while E is mid-fetch you may catch a false red; check if a failure is a pack town.
Still filed for v4.x: GIG_RANGE 34m radius vs single-chunk lookup (8.7% chunk-blind).
Flagged as unowned: R20's main-street relocation half (hero strip vs densest cluster).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Ledger #5 pull-forward. Nothing here touches a golden until A absorbs in R22; F tags v4.0-beta with
or without it. selfcheck ALL GREEN 154141/154141, synthetic 0x3fa36874 unmoved, all 22 byte-identical
on re-run.
- Curated 18 candidates on secondhand density first (heroes are the game), a walkable 2-3.4km centre,
and a national spread; 17 survived -> 22 towns, 1192 shops, 298 heroes, all 8 states/territories.
fitzroy is the mecca (40 heroes/160 shops, 2x katoomba); thin tail toowoomba/darwin (3 heroes) left
in as honest data -- A/Fable may drop them in R22.
- FINDING: the 3x-hero cap is a CURATION FILTER, not just a density knob. Wollongong had ~62 real
shops but 1 hero -> capped to 4 -> dropped by the >=6 floor. A town without a secondhand strip
cannot inflate into a retail strip; John's SUBTLE directive does quality control for free.
- Newcastle: my centre was wrong, not the town (CBD held 2 heroes; the strip is Hamilton/Beaumont St).
Re-centred: 2 -> 9 heroes, 8 -> 36 shops.
- BUG (mine, shipped R18, caught here): fetch_roads_raw had no query-comparison while fetch_raw did,
so re-centring refetched shops but silently reused the OLD CBD roads -> Hamilton shops seated against
a graph 2.5km away -> the town marched out with ZERO shops (surfacing as 32 bogus 'open-late
invariant' failures; ten other no-video towns passing is what proved it wasn't the open-late rule).
Fixed: the roads path now compares its query too -- the bbox lives in the query, so a re-centre must
invalidate BOTH snapshots. Newcastle rebuilt: 34 shops seated, openLate=1.
- Discipline: bounded bboxes, raw snapshots committed (shops+roads), C's §6 mapping, SUBTLE guardrails
(heroes never subsampled, texture 3x, 2392 drops counted), ODbL. Overpass rate-limits hard at pack
scale (batch 1 died on a fatal 504 after 25 retries) -- courtesy sleep 4s, two polite batches.
-> Lane A (R22): 34 goldens to pin (17 new towns x base+gig); sizes 12->160 shops load-test your
capacity/overflow fences. -> Lane B: index.json now lists 22 towns (your selector already reads it).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Wave 2 — ran after A's venue cluster-adjacency bias landed (110c828), so the soak reads the
relocated world + A's re-pinned goldens. The carried R20 #6.
Five towns at their new densities, ~25 deterministically sampled doors/town (sample takes one
of EVERY type present, then stride-fills, so the texture classes' first real appearance is
always covered):
town shops sampled venues types worst draws verdict
katoomba_real 72 25 3 10 157 PASS
newtown_real 67 25 3 11 155 PASS
fremantle_real 79 25 3 11 153 PASS
bendigo_real 35 25 3 10 154 PASS
castlemaine_real 23 23(all) 3 8 156 PASS
123 doors: 0 throws / 0 carves / 0 path-fails / 0 over-budget, worst 157 draws (<=350).
typesInTown == typesSampled in every town (all archetypes entered, incl. all 3 venue kinds).
All five towns landed 3 venues -> A's cluster bias + district placement hold at density.
§6 mapping verified live: katoomba cache = 20 heroes + 60 texture (exactly E's 3x cap, heroes
untouched). Texture's first real appearance renders right -- "Blue Mountains Chocolate Company"
(real OSM shop=chocolate) wears milkbar as a believable confectionery counter shop. Texture,
not takeover. Cache->plan deltas are A's counted drops, not an interior concern.
C->F: density verdict GREEN. Docs-only.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
D found katoomba's pub 1,000m from the retail cluster — a gig with no passing crowd. Measuring all
five towns first showed it was SYSTEMIC: 11 of 15 venues sat with <3 shops within 160m (newtown's were
0/0/1 — every venue stranded). After the bias: 0 of 15. Katoomba's pub 1 → 9. Castlemaine cleared without
needing the floor exception the brief anticipated.
- pickVenues builds a CLUSTER-ADJACENT field first (>= CLUSTER_MIN 3 shops within CLUSTER_R 160m), then the
kind flavors (pub corner / band_room fringe / rsl off-spine) bias WITHIN that field rather than overriding
it (per the brief: biases, not absolutes) — a fringe band_room in the cluster beats one a kilometre from
anyone. A town with no cluster at all falls back to the pool.
- SYNTHETIC FROZEN BY CONSTRUCTION, not by luck. The brief assumed a blanket filter would be a no-op because
'the synthetic town is one dense cluster anyway'. Measurement says otherwise: 4 synthetic candidates PER
SEED sit under the threshold (min cluster 0-1), so a blanket filter WOULD have moved the frozen golden.
Gated on plan.source === 'osm' (undefined on synthetic) — impossible rather than unlikely. The marched
fixtures are 'osm' but genuinely dense (min cluster 3/11/13) ⇒ verified no-op.
- Only the GIG goldens moved (venue selection is gig-layer; base town goldens untouched): katoomba
0xbf586b77, newtown 0xacdb3eee, fremantle 0xcf3426fc, bendigo 0x646b2c23, castlemaine 0xf9713a69.
classic 0x3fa36874 / gig 0xb1d48ea1 / marched fixtures FROZEN. selfcheck ALL GREEN 66419/66419;
scaffold + consistency GREEN; cross-process deterministic.
- FYI: E's v4.0 pack (glebe/marrickville/newcastle) already boots clean through my loader (validates + full
suites pass), listed UNPINNED — I pin the pack in R22; it does not gate the beta tag.
Handshake → C + D: cluster bias landed, goldens re-pinned, start your verifies (before/after table in
LANE_A_NOTES §R21).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Closes the gap Fable caught (hud.js:11 still hardcoded REAL_TOWNS). The real-town list now comes from E's
web/assets/towns/index.json (procity-towns-index/1), so it grows with the town pack instead of being edited
in hud.js. Verified fresh Chromium. Nothing else touched -- tram/ground work is done and verified.
Fail-soft by construction: createHUD is synchronous, so the picker is built immediately from the fallback
literal and UPGRADED IN PLACE when the index resolves -- if the index never arrives, a working list is
already on screen.
- index present -> real names/states/shop counts: "Katoomba, NSW · 80 shops", "Fremantle, WA · 80 shops",
"Newtown, NSW · 72 shops", "Bendigo, VIC · 36 shops", "Castlemaine, VIC · 24 shops" (roads:false -> "· marched").
- ?noassets -> the fetch is SKIPPED entirely (asset law; the shell doesn't load town caches there either).
Measured: 0 index fetches, fallback list, 0 errors.
- anything wrong with the index -> the fallback simply stands. Exercised the exact chain: 404, network error,
malformed JSON, towns:[], towns-not-an-array, junk entries (no key) -> none throws, none rebuilds from junk;
only a good index rebuilds.
REAL_TOWNS_FALLBACK is retained ONLY as that fallback (comment says so), no longer the source of truth.
Verified live: booting katoomba_real shows "Katoomba, NSW · 80 shops" pre-selected; fits (193px of 230px max);
0 page errors; navigation unchanged (onchange is set once on the <select>, so rebuilding options never clears it).
-> F: unchanged from R19 -- NO shell seam (selector is entirely hud.js, already built by createHUD) and it
lives inside #pc-hud so setVisible(false) hides it for the beta money shot. The gating call is still yours.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
godverse_town.py marries a donor OSM cache's roads[] with thriftgod's DealGod-scraped
shop census (2,928 real AU secondhand shops) -> <town>_godverse.json, schema v2 unchanged.
First artifact newtown_godverse boots clean (?plansrc=osm&town=newtown_godverse).
Lane brief + asks (Draco ruling, per-town bookmarks, enterShop selector) in
docs/LANES/LANE_G_GODVERSE.md + G-progress.md.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Re-ran everything against A's absorbed graph (5d29863, selfcheck 51866/51866).
Katoomba barely moved from my pre-absorb read (null 77.2->78.1, visits 64->62,
hero share 30->29%), so the earlier conclusion held.
SIM VERDICT: zero changes needed at 4x density. All five towns: 0 NaN, chunk rosters
stream (15-24), identity sigs accumulate (262-483), patronage/loiter/bench-sit all fire,
up to 401 active peds, no degradation. Determinism unchanged (identity is graph+seed
keyed; shops don't touch it). D->F: GREEN.
THE FINDING (for E/C/Fable): the widening MOVED the main street off the secondhand strip.
Hero visit-share falling (29%->0%) is NOT a behaviour bug — patronage is type-agnostic, so
visits track the local mix exactly. The mix is the story. Pre-widening the densest cluster
WAS the secondhand strip (katoomba: 9 of 9 main-street shops were heroes). Post-widening
the centre of gravity relocated 308m (katoomba) / 434m (castlemaine) onto the general-
retail cluster, where heroes are a 0-19% minority. Castlemaine's main street now has ZERO
secondhand shops (0 of 4 heroes within 160m; median hero 463m away). The heroes weren't
drowned in place — they were demoted: the liveliest street is no longer theirs.
The good half stands: heroes kept absolute footfall (katoomba 17->18 visits) and the
street is 3.8x busier. The widening works; it overshot the dial. This is exactly the case
John's directive pre-authorised ("E may seed-subsample the texture classes"). My read
(measurement, not a ruling — E/C/Fable own it): subsample texture until the secondhand
strip is again the densest cluster. Sparse towns need it most (bendigo 7%, castlemaine 0%).
ALSO FILED (A/F): venues land off the retail hub at density — newtown's three are all
2190-2433m out with 0-1 neighbours; katoomba's pub still 1007m/1 (A's "pub facade bias"
fixed frontage, not position). Quality call, not a break; "venue placement still lands"
is A's ledger + F's district gate. Castlemaine shows the good case (band_room on the hub).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
E's five caches landed at 3-6x shops (292 total: katoomba 20→80, fremantle 21→80, newtown 18→72,
bendigo 9→36, castlemaine 6→24). All five absorb cleanly and still place 3 venues; the R20 poster cap
auto-scaled off the live shop count exactly as designed (katoomba 111 posters at 80 shops).
- CAPACITY WIDENED (the R18 overflow fence, load-tested). At 4x the dominant drop cause was OVERFLOW —
an edge-side holding more shops than fit (katoomba 10 of 11 drops). Tightened the run: NODE_CLEAR 6→3
and a ROADS-ONLY RGAP 2→0.5 (the shared marched GAP is untouched ⇒ marched fixtures stay frozen).
Drops 7.9% → 5.5% (16/292). 0.5m also reads truer — a real AU main street is continuous terrace
shopfronts, not detached boxes. Stopped there deliberately: the next lever (spilling overflow onto a
DIFFERENT street) buys ~1% by putting a shop on a road it isn't on — dropping a crowded shop is the
honest trade. Residual drops stay COUNTED, now BY CAUSE (dropped_overflow/_overlap/_stranded).
- VENUE FACADE BIAS NOW COVERS pub. R16 exempted pub ('spine-fronted, +Z always open') — true on the
synthetic/marched towns, but on a REAL graph at density the pub landed where its frontage poster
overhung a cross street (newtown, -0.91m). The POSTER_CLEAR filter now covers every venue kind; still a
NO-OP on synthetic (every pub candidate already clears) ⇒ gig golden 0xb1d48ea1 frozen.
- Goldens re-pinned ONCE (base + gig, all five towns). classic 0x3fa36874 / gig 0xb1d48ea1 / marched
fixtures FROZEN. selfcheck ALL GREEN 51866/51866; scaffold + consistency GREEN; cross-process
deterministic.
- Fixed: my real-cache loader globbed *.json and tried to validate E's new index.json as a town cache —
it now skips it (that's B's selector index, not a cache).
Handshake → all: densities absorbed, goldens pinned. → F: re-measure venue_night + gate the five towns.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Verified fresh Chromium across 9 town configs (synthetic, 3 fixtures, 5 real).
TRAM RULING (ledger #3): spinePolyline now walks from each main-subgraph dead-end, greedily taking the
unused main edge fronting the most shops (lot.frontEdge), keeping the best-scoring path (ties -> longer).
SYNTHETIC + FIXTURES BYTE-IDENTICAL by construction AND measurement (a synthetic main node has <=2 mains so
there's never a choice; its two dead-ends tie exactly and strict > keeps the pre-R20 start). Old vs new
logic on the same plan = identical polylines (Boolarra: 7 pts, 6 edges, 810m, 172 shops, 2 stops).
The alpha tram fronted 0 shops on 4 of 5 real towns (it was a highway bus). New: katoomba 0->3, newtown
0->8, fremantle 0->2, bendigo 0->1, castlemaine 1->6. Where the best main chain still fronts <5 shops the
tram is FENCED per-town (Fable's fallback). RUNS: synthetic, all fixtures, newtown (8 shops/1924m/16 stops),
castlemaine (6/2184m/10 stops). FENCED: katoomba (3), fremantle (2), bendigo (1) -- highway, not high street.
The fence keys off the REAL-ROADS signal (edges>200), NEVER shop count alone: the marched fixtures front 0
shops on their mains (bare spine, avenues carry shops) so a naive fence would have killed their v2 tram.
routeInfo{fenced,reason,shopsFronted,routeMetres} exposed for F's smoke.
BUG FOUND+FIXED: the stop projection accepted every shelter in the town with no distance gate -- harmless on
synthetic (2 shelters on the spine), nonsense on real graphs: Newtown made 149 phantom stops (~9 min of dwell
at points the tram never passes). Added a 30m gate (a shelter sits halfRoad+1.7 <=~16m off its centreline, so
genuine stops survive): newtown 149->16, castlemaine 37->10; synthetic 2 / fixtures 3 unchanged.
GROUND POLISH (ledger #4): measured, NO fix needed. Real Katoomba has 325 deg-3+ junctions, 28 under 30 deg,
worst 4.3 deg (near-parallel). At grazing eye level the worst reads clean -- no z-fight, no gaps. Layers are
already y-separated (road 0.0/foot 0.02/kerb 0.06) and road-on-road overlap is invisible by design (OVERLAP=1.5
overruns nodes on purpose; one skin, one y, one merged draw on uniform asphalt). Cosmetic only: footpath/kerb
double up on the sliver between <10deg near-parallel roads -- v4.0 candidate, not a defect.
Selector polish (ledger #5): BLOCKED -- E's towns/index.json not landed (build_towns.py WIP); REAL_TOWNS
hardcode stands, one-line swap when E lands.
-> A (finding, measured): only a slice of shops front a MAIN edge on real towns -- katoomba 7/69, fremantle
4/79, bendigo 7/35, newtown 32/64, castlemaine 11/22. The real high street classifies as 'side', which is
WHY three towns need the fence. If the main/side classifier promoted shop-dense ways to main, the fence
would lift. Note onMain also gates gigs.js pickVenues pub placement.
-> F: tram verdict above; selector gating (HUD vs ?dbg) is your call, one line either way.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
SELF-CORRECTION first. My part-1 note called the 95-98% patronage null rate
"sparsity-throttled" and made it the headline metric. Both halves were wrong:
(1) it was measured at 21:00, when only 4 of 69 shops are open — retail is shut BY
DESIGN, so a high night null is correct behaviour, not sparsity; (2) it sampled each
town's first venue, and katoomba's is the pub — 1000m from the retail cluster with 1
neighbour. I had parked the camera on the one isolated venue in town. Same town, same
seed: at the pub 99.4% null / 0 visits; on the actual main street 77.2% / 64 visits.
The gig-crowd table stands (it measures the mechanism — 0 NaN, rosters, venue-win —
which is location-independent); only my sparsity reading was bad.
THE REAL A/B (main street, 14:00, same seed, pre-cache from git vs E's widened):
hub density 8 -> 25 shops within 160m; null 90.2% -> 77.2%; total visits 17 -> 64
(3.8x); hero visits 17 -> 19; texture visits 0 -> 45; hero share 100% -> 30% of visits
from 26% of shops.
THE DIRECTIVE LANDS, measured from the crowd side: the secondhand heroes did NOT lose
footfall (17 -> 19). The 45 texture visits are added ON TOP, not taken from them. The
street is 3.8x busier while the heroes stay over-indexed. "More alive, not a different
game."
FILED: (a) venue placement at density — katoomba's rsl/band_room sit in the strip
(13/10 neighbours) but the pub is 1000m out with 1; a gig there has no passing retail
crowd. Quality observation, A/F's lane. (b) GIG_RANGE is a 34m radius but
_nearestOpenShop searches only the ped's own 64m chunk — 8.7% of in-range peds are
chunk-blind. Real but small; v4.x candidate, not a mid-beta hot-path change.
POST numbers are vs E's committed caches but BEFORE A's absorb (11/80 still overflow-
drop). Final re-run + the other four towns once A lands.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Ledger #2 (E's half), built to C's LANE_C_PUB §6 table under John's SUBTLE directive.
- Widened fetch: C's tasteful staple list (shop=* + amenity=cafe per C's dagger note), mapped through
C's copy-paste table onto EXISTING registry types (no new archetypes). Guardrails honoured: no
vehicle/fuel/funeral/storage/industrial, no pure services. New raw snapshot per town (the fetcher
auto-refetches when the class list changes -- query recorded in each snapshot), committed like the first.
- SUBTLE by construction: secondhand HEROES never subsampled; texture capped at 3x heroes (top of C's
~2-3x band), over-cap towns seed-subsampled deterministically (sha256(townKey:osmId)) with every drop
counted in counts.textureDropped. Result 3.0x in all five: katoomba 20+60 (-4), newtown 18+54 (-156),
fremantle 20+60 (-81), bendigo 9+27 (-81), castlemaine 6+18 (-43). Density 6-21 -> 24-80 shops (3-4x).
- web/assets/towns/index.json (new, E-owned): {schema, generated, license, attribution, towns[{key, town,
state, shops, roads}]} -- B's selector can drop the REAL_TOWNS hardcode (hud.js:11).
- Deterministic (byte-identical re-run, 0 re-fetch). ODbL: SOURCES.md + per-cache license/attribution.
-> Lane A (absorb pass): (1) 10 expected golden re-pins -- the widening moves every real town's base+gig
golden (designed E->A handoff); synthetic 0x3fa36874 UNMOVED, fixtures frozen. (2) SEAM: index.json
trips selfcheck's town glob (selfcheck.js:467 takes every top-level towns/*.json as a cache) -- one-line
fix in your file: .filter(f => f.endsWith('.json') && f !== 'index.json'). Didn't touch selfcheck.js
(yours, you're mid-round in it); Fable spec'd the index at this path so the path stays.
-> Lane C: kept video_games -> video (absent from your §6 table; was already shipping as video pre-widening,
so the fallback would have regressed it to opshop). Same family, same target -- flagging for your table.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The unblocked half of "citizens at density": the gig crowd's BEHAVIOUR on all five
real towns' venues, measured BEFORE E's widening so the after-table is a real A/B.
Density re-audit follows once E's caches + A's absorb land.
BEHAVIOUR IS CORRECT ON ALL FIVE: 0 NaN everywhere; tonightRoster populates (2-7),
so R14 identity continuity holds on real bones; every town generates 3 venues
(pub/band_room/rsl) on night hours; and the surge works — whenever a patronage check
finds any shop, the gig venue beats ordinary shopfronts (venue-win 100% on newtown/
fremantle/bendigo, 74% katoomba). Castlemaine's 40% is the 6-shop floor, not a leak:
3 of its 6 shops ARE venues, so "other" hits are mostly other venues.
FINDING (quantifies Fable's ledger #2 "real towns are sparse"): the crowd is
SPARSITY-throttled — 94.9-97.7% of patronage checks return null on every town, because
peds spend nearly all their time in shopless chunks (katoomba: 19 shops in ~5 door-chunks
of a ~78x78-chunk town). The mechanism is fine; the town has no shops per chunk. The
density widening attacks exactly this — null% is my headline before/after metric.
FILED (not fixed): GIG_RANGE is a 34m radius but _nearestOpenShop searches only the
ped's own 64m chunk, so in-range peds across a chunk edge are blind to the gig — 8.7%
of in-range ped-samples (852/9740) on katoomba. Small vs the 97% sparsity throttle;
not worth a hot-path change mid-beta. Revisit if the widening leaves null% high.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The density widening's C half: real towns fetch general retail, not just secondhand.
Published the OSM shop-class -> registry-archetype mapping in LANE_C_PUB.md §6 (new
v4.0-beta amendment; frozen venue contract §0-§5 untouched). C->E handshake: E builds to it.
- No new archetypes. General retail wears existing recipes: food/grocer/newsagent/chemist
-> milkbar; clothes/hardware/big-retail -> dept; produce/florist -> stall.
- Secondhand stays the hero (charity->opshop, books->book, music->record, video->video,
pawn->pawn, toys->toy); never subsampled. Unknown/unlisted -> opshop (R18-proven).
- Subtle per John: texture not takeover. Copy-paste JSON map + tasteful fetch list (skip
vehicle/funeral/industrial/pure-services) + guardrail for E to subsample if retail drowns
the secondhand strip (count in norm.dropped).
- Mechanic: plan_osm keeps registry types, so E bakes a registry type per the map into each
cache's shop.type -> no C code change, budgets unchanged.
Pre-validated before E builds to it: milkbar/dept/stall/opshop all build valid at a real
katoomba lot (8.5x14, 0 carves, <=143 draws); dept reads as a clean dressed-retail floor
(shelves+racks+counter, escalator does not dominate the 7.9x11.7 hall) -> clothes/hardware
-> dept confirmed. Full five-town door-soak (ledger #6) follows E's re-emitted caches.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Closes my own R19 finding + F's one open budget edge. A sparse REAL town has few shops but a big edge
graph, so the 'two posters per main edge' pole run reads busy (katoomba 438) and drove venue_night to
200,965 tris (+0.5% over the 200k soft ceiling).
- gigs.js buildPosters caps the spine pole run at POSTER_PER_SHOP × shops (1.5/shop, floor 12) — a seeded
subsample. SCALED BY SHOP COUNT, not edge count (the whole point). katoomba 438→31 posters (~2x the
synthetic feel), fremantle 32, castlemaine 14. NO-OP on the shop-dense synthetic town (493 shops ⇒ cap
~740 ≫ its ~15 posters), so the synthetic gig golden 0xb1d48ea1 stays FROZEN.
- Posters are gig-layer (withGigs), so the real-town BASE goldens are unchanged (correctly green). Added
REAL_TOWN_GIG_GOLDENS (the full district+capped-poster plan per town: katoomba 0xbd332e83, …) to
regression-guard the capped output byte-exact, not just via districtInvariants. classic/gig/marched
fixtures frozen; selfcheck ALL GREEN 51675/51675; scaffold + consistency GREEN; cross-process deterministic.
→ F: re-measure venue_night, the +965 edge should close. Density absorb (ledger #2) pends E's widened
caches — the poster cap already reads the live shop count, so it auto-scales to the widened towns.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
A resolved the R18 findings (fragmentation cull/join, poster clamp, goldens pinned) + D verified; F gated.
- town_matrix.py: a `roads` dimension — a schema-v2 cache (roads[]) must build the REAL street graph
(norm.mode==='roads', read via the plan_osm opts.report sink); synthetic/fixtures stay marched/synthetic.
10 towns x 6 gates: real roads green everywhere; classic/default proven untouched (roads ride the cache
schema, not a flag — 0x3fa36874 frozen).
- docs/shots/v4_alpha/katoomba_mainstreet.png: the first real-street money shot — Katoomba's actual main
street (asphalt + lane markings, footpaths, a pedestrian crossing, a walking crowd, gig posters, B's town
selector in-frame), PROCITY's look on real bones, 121k tris midday.
- docs: LANE_F_NOTES §18 (why F correctly HELD in R18 — the gate would have failed by design) + §19 (the
close), README real-roads note, F-progress.
qa.sh --strict --soak GREEN 7/7. One filed NON-blocking finding: katoomba_real venue_night 200,965 tris
(+965 / 0.5% over the soft ceiling, poster-count-scales-with-real-edges driven; midday play is 121k) — a
v4.0-beta hardening item (A caps posters per real town), on the opt-in --matrix warn-gate, not a play view.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Post-fix re-measure of my R18 finding against A's committed connectivity pass
(6fc4b40). Same harness, same seed.
VERDICT: fragmentation fully resolved, sim untouched. Real Katoomba:
components 105 -> 1, street-metres in main 58.2% -> 100%, shops stranded 2 -> 0
(1 bridged, 1 dropped past SEAT_MAX AND counted in norm.dropped), % near-crowd on
fragments ~10% -> 0%. selfcheck 51670/51670 green, katoomba golden pinned.
Mechanism check: re-ran R18's dead-straight-through-junction geometry — R18 dropped
the junction (deg 2, disconnected); R19 protects it (deg-4 crossroads, 1 component).
DP-collinear drop fixed at the source.
Sim confirm on the fixed bridged graph: 0 NaN over 1,303,200 samples, footpath-perfect,
loiter/bench-sit fire, patronage rose 24->41 (peds reach shops across the connected
graph), 0 fragment-stranded peds. Zero sim changes, as R18 predicted.
D->F: fragmentation verdict post-fix is GREEN — tag away. FYI B/F: edges rose 799->966
(junction protection + bridges), flagged for the tris budget.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Shipped the town picker in hud.js (which already owns its DOM overlay -> NO index.html seam). A compact
<select> in the HUD (top-left, pointer-events:auto), 9 options: synthetic + 3 OSM fixtures (osmTownKeys
minus _real) + 5 real-road caches (Katoomba/Newtown/Fremantle/Bendigo/Castlemaine, hardcoded REAL_TOWNS --
no manifest yet, beta derives it). Selecting navigates: synthetic clears plansrc/town; a town sets
?plansrc=osm&town=<key>, preserving all other params. Current pre-selected; unknown-current still shown.
Verified end-to-end: picked katoomba_real through the dropdown -> booted real Katoomba (966 edges/323
intersections/3 venues), 0 errors. Lives inside #pc-hud so hud.setVisible(false) hides it for money shots.
-> F (B->F seam): NONE -- self-contained in hud.js (already constructed by createHUD). Gate it behind ?dbg
with one setVisible call if you prefer; left always-in-HUD as a modest debug-grade feature.
Corner-strips spot-check (R18 failure-item #3): ground fills real intersections OK -- degree-4 junction
(58 exist) shows road strips overlap-filling the junction, lane markings + zebra crossing render, no gross
gap. Acceptable for alpha; finer acute-angle pass = beta.
Budget note -> F (measured vs A's UNCOMMITTED WIP -- caveat, not a defect): F's BIG_CITY one-liner is live
(chunks 50->26). But venue block still ~237k tris at night against the current tree, which carries A's
uncommitted R19 finalization (plan_osm modified; graph grew 799->966 edges, junction-joining in but the
far-island CULL not yet). Fable's ruling: the cull is "half of B's tris problem" -> the <=200k gate needs
F's fix AND A's committed cull. Re-measure the busiest real block after A pins (the deferred tris-post-cull
re-measure genuinely waits on A's commit; the dev server serves other lanes' uncommitted edits).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
A's finalization (ledger #1), the round's spine. Every filed R18/R19 finding resolved; the tag is F's.
- Fragmentation (D's finding; Fable ruled, A owns details). The naive real graph fragmented (katoomba 105
components, 58% street-metres on islands). plan_osm.buildRealRoads now: (a) JUNCTION-PROTECTED
Douglas-Peucker — a point shared by >=2 ways is a junction; simplify only BETWEEN junctions, never
through one (the collinear tee-off drop, the root cause) → 94-100% main-net alone; (b) cull/bridge —
main component + near-fragments joined (JOIN_TOL 12m; shop-island bridge <=200m), far SHOPLESS islands
culled. Result: 1 connected component on all 5 towns (katoomba 100% main-net, 2 islands culled). Every
dropped shop COUNTED in norm.dropped (ruling): katoomba 19/20 (1 crowd drop — R18 overflow prediction
measured), the other four kept all.
- E's poster-clearance finding fixed (gigs.js): a spine poster overhanging a real crossing carriageway is
skipped, measured exactly as the selfcheck kerb floor — a NO-OP on synthetic (gig golden frozen, pr()
consumed before the check). 23 real-town clearance failures → 0.
- Goldens pinned (A->E->A finalization): all 5 real towns on their real-roads output (katoomba 0xb7ffbca2,
newtown 0x6f984c81, fremantle 0xfb197000, bendigo 0x21179b3b, castlemaine 0x7c47f639). synthetic
0x3fa36874 / gig 0xb1d48ea1 / marched fixtures / classic / default FROZEN — the alpha changed nothing
outside the cache-schema path. selfcheck ALL GREEN 51670/51670; scaffold + consistency GREEN. Cross-
process deterministic.
- New finding filed (beta, non-blocking): spine-poster density scales with edges (katoomba 438 vs ~14
synthetic; draws low via instancing) → a beta posters-per-town cap.
Handshake fired → B/C/D/F: katoomba_real boots real roads, golden pinned, findings resolved.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
E's roads landed (katoomba_real now schema v2, 484 real roads), so re-ran the door soak
against the real street graph (closes the R18 "re-run when roads land" caveat).
Honest caveat: the dev server also served Lane A's UNCOMMITTED R19 fragmentation WIP
(plan_osm.js: JOIN_TOL/BRIDGE_MAX, junction-protected DP, island cull/bridge), so this
exercised A's in-progress finalization, not a frozen state.
Robust result (holds whatever A's WIP does): every shop the plan produced builds a valid
interior -- 0 throws/carves/path-fails/over-budget, worst 147 draws; 3 gig venues (district
placement holds on the real graph); plan deterministic; lot dims unchanged (7.5-8.5 x 14 ->
interiors dim-identical to R18, road-agnostic). Shop count came out 19 here, but that reflects
A's in-flight cull/bridge logic (expected per Fable's fragmentation ruling), NOT a defect --
not filed as a finding; interiors build clean whichever count A pins.
Final C spot-confirm pending A's committed finalization + golden pin (C confirms after A).
Docs-only.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Real towns can have FEW shops but a HUGE road graph (real Katoomba: 20 shops / 799 edges / 5 km), so the
shop-count-only heuristic left them on small-town streaming and blew the tri budget (280k > the 200k gate).
Extend the trigger to the graph: BIG_CITY = shops>120 || edges>200. B proved the effect; landed EARLY so
D/B's post-fix re-measures see it (per the R19 order).
Verified: katoomba_real 799 edges -> BIG_CITY -> 61 draws / ~12k tris (was 280k), 0 errors.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Audited the citizen sim on real irregular street topology. E landed katoomba_real's
real roads (484 ways -> 872 nodes/799 edges) mid-round, so this is measured on the
real Katoomba graph plus a crafted irregular proxy.
VERDICT: the CityPlan graph contract IS the contract — zero sim changes. On real
Katoomba (7.2deg junctions, 1303m edge, 105 components): 0 NaN over 1,095,500 samples,
peds footpath-perfect (0 off-footpath), loiter/bench-sit/patronage all fire, chunk
streaming works over non-grid, determinism byte-identical (1167 identity sigs, R14
chunk-local law holds). No source touched.
FILED FOR A: katoomba_real's graph is 105 components — main holds 58% of street-metres
+ 18/20 shops, but 42% of street-metres + 2 shops strand (~10% of near-main crowd
presence lands on tiny fragments). Primary cause is expected: median fragment 71.5m
from main = peripheral/bbox-clipped roads. Secondary (corroborated with Lane B on A's
selfcheck fixture): per-way Douglas-Peucker drops collinear junction points + no
mid-edge split — bites only the 9/85 fragments <10m from main. Reconciles with B:
B verified 126 intersections form (main component); I add the global measurement
(still 105 components). A fidelity knob (charter risk #4), not a blocker; sim degrades
gracefully (no crash). A owns the connect-tolerance decision.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Audited katoomba_real on E's real roads (schema v2, 484 OSM ways -> 799-edge/872-node graph, 126 real
intersections, edges 3.5m-1303m incl 28 short <12m, ~4x5km span). E's roads landed mid-audit; verified
fresh Chromium, cache-busted. Docs-only (no code change: streetscape is road-agnostic).
VERDICT: Lane B streetscape is ROAD-AGNOSTIC — boots real Katoomba 0-error, every B system builds and
renders on the irregular graph (ground, furniture 39 chunks/148 trees, town-wide pools, tram, chunk
streaming, buildings, wind). Real main street reads well (road+footpath+lamps+working gig poster+citizens).
Like C's interiors, needed no code change to eat real geometry.
FINDING -> F (gate-blocking, PROVEN one-line fix): the BIG_CITY adaptive heuristic misfires on real roads.
index.html BIG_CITY = plan.shops.length>120 -> Katoomba has 20 shops(<120) but 799 edges/5km, so it stays
small-town mode (radius 3 + shadows ON), streams 50-52 chunks -> busiest venue block ~280k tris (clean,
settled) = BREACHES 200k gate (draws 150 <300). Proven fix: ?r=2&shadows=0 (BIG_CITY mode) -> 84k day /
104k night, 20 chunks, 72 draws, well under. Recommend: BIG_CITY = shops>120 || edges>200 (catches
Katoomba's 799, leaves synthetic Boolarra's 22). B's R3 adaptive-streaming concept, edge-count not shop-count.
Tram (Fable's flagged spine risk): degrades gracefully, no crash. spinePolyline runs one main chain =
3527m/49 edges = 21% of 232 mains (longest walkable chain). Arbitrary-but-functional for alpha; smarter
route or hard fence = beta (no plan.mode marker exposed to fence off; didn't add untested fencing).
Furniture/pools scale: town-wide pools ~4989 instances = still ~1 draw; 28 short edges get 0 lamps
(graceful). Correction: the disconnected graph I first saw was a synthetic-fixture DP artifact (junction
vertices dropped on near-collinear ways) — real Katoomba forms 126 intersections correctly; flagged to A.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Newtown/Fremantle/Bendigo/Castlemaine now schema v2 with roads[] (1086/622/1552/397 real streets),
same fidelity cut (street tier only). Data-only: only katoomba gates the round; A hardens against these.
build_towns.py: added Overpass retry/backoff (public API 504s under load — hit one mid-run).
The poster-clearance finding is SYSTEMATIC: it reproduces on all 5 real-road towns, every seed (23
checks, -0.9 to -4.7m), confirming it's plan-level poster placement (A's pickVenues/POSTER_CLEAR), not
town data. Scoping holds: only the real/* v2 path fails; synthetic + fixtures + classic + default boots
stay green. A's one fix generalizes to all five. ODbL: raw roads committed, SOURCES.md current.
-> Lane A: all 5 towns' roads live; the systematic poster finding is yours to resolve + re-pin goldens.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Extend build_towns.py: bounded way[highway](bbox) out geom per town -> raw cached in-repo -> roads[]
in A's v2 contract ({id,kind,pts:[[lat,lon]]}, kind=raw OSM highway class). katoomba_real is now
schema v2 with 484 real streets (Bathurst Rd, Cliff Drive, ...). Fidelity cut (charter risk #4, flagged
to A): fetch every ROAD_KIND tier -> raw; EMIT the town street tier only, excluding service/track/
footway/path/steps/cycleway (1461 of 2506 Katoomba ways = bushwalking tracks/sidewalks/stairs/driveways).
A's lift builds a real town from it (verified): 484 streets -> 872 nodes/799 edges (main 232 side 531
arcade 36) -> 12 blocks, 20 lots, 3 venues (pub/band_room/rsl), 12 gigs. pickVenues lands venues on the
real graph (ledger #2 met).
FINDING for A (alpha failure-list, charter risk #3): on the real graph, spine posters overhang crossing
carriageways -- selfcheck kerb-clearance fails on 4 seeds (-1.6 to -2.5m; e.g. seed 20261990 poster 50 vs
edge 58 = -2.42m). Same class as A's R15/R16 corner-poster fix, now on real cross-street geometry. Owner:
A (pickVenues/POSTER_CLEAR on the real graph, the 'A again' harden step) -- my roads data is VALID (cache
validates; failures are plan-level poster placement). Golden settles after A's fix (pre-fix 0x6014a8e6).
ODbL: raw roads committed, SOURCES.md updated.
-> Lane A: katoomba roads live; harden the lift + resolve the poster finding + re-pin the golden.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Entered every katoomba_real shop the way the shell does (registerTownCache ->
generatePlanFor osm, gigs on = default boot -> buildInterior per shop, lot dims
from plan.lots). All 20 doors PASS: 0 throws / 0 carves / 0 path-fails / 0
over-budget, worst 153 draws (<=350); 7 types incl. the 3 gig venues (pub/rsl/
band_room) all built valid.
Degenerate-lot stress (preempts A's real-road re-seating): 32 extreme-lot x type
combos (incl. 0x0, 1x1, thin slivers) -> 0 throws/carves/path-fails; every room
floors at >=4x4 m. The interior contract absorbs any real frontage.
Honest caveat: current katoomba_real.json is schema v1 (roads:none) -> marched
fallback (E's roads cache not committed yet). Interiors are road-agnostic (lot dims
+ type + seed only), so the conclusion holds once roads seat shops on real edges;
one-line re-run if lot dims shift. Interior lane: 0 failures for the alpha list.
Docs-only (verification of shipped behaviour).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The epoch headline (ledger #2). A v2 cache's roads[] now builds the CityPlan street graph FROM real
OSM geometry instead of marching shops onto synthetic avenues. SAME CityPlan schema out — B/C/D/F consume
it unchanged by contract. roads absent ⇒ the v1 marched fallback, so classic/gig/osm goldens are FROZEN.
plan_osm.buildRealRoads: project ways+shops to the local metre frame → simplify each way (Douglas-Peucker
eps~6m, the fidelity knob) → snap coincident points (<=3m) to shared nodes = real intersections →
straight-segment edges → classify main by OSM highway class + shop density → seat each shop on its NEAREST
real edge (real order, real side), marched with regularised spacing + a deterministic overlap-resolve pass.
Fidelity (charter risk #4): real edge/order/side + regularised spacing — NOT pixel positions (real shops
overlap; the no-overlap invariant is load-bearing). Alpha failure list in LANE_A_NOTES §R18: overflow drop
on short edges; intersections = shared-coordinates only (a D sim-graph risk); per-edge frontage-strip
blocks not planar faces (a B ground-render item); crowded-intersection overlap-resolve.
Proven: a committed synthetic irregular graph (main + two cross streets + a real intersection + an acute
angle + a dead-end lane + a curve) passes the FULL structuralSuite + gig district + byte-identical
determinism; mutation-tested non-vacuous (flip the facade facing → 'faces its frontEdge' fails). Robust on
messy inputs (single road, coincident dups, disconnected ways, all-roads-too-short → marched). All osm
goldens frozen; E's 4 trailing caches pinned on v1; katoomba_real UNPINNED (its real-roads golden pins when
E lands roads — the A->E->A finalization). selfcheck ALL GREEN 16810/16810; scaffold + consistency GREEN.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The half-day handshake so E's build_towns.py can fetch road geometry (ledger #1). Optional roads[]
on the town cache carries real OSM way geometry; plan_osm's graph lift (consuming it) follows in the
next commit. v1 caches (roads absent) stay valid — the marched fallback — so nothing shipped regresses.
- validateTownCache accepts schema procity-town-cache/1 AND /2; when roads[] is present each road must be
{ kind:string, pts:[[lat,lon],…] } with >=2 finite points (else rejected); <2 pts dropped, unmapped
highway kind → 'side' (warnings). ROAD_KIND maps OSM highway=* → CityPlan edge kind (main/side/lane/arcade).
- Full v2 contract in web/assets/towns/README.md (the doc E builds to): shape, road validity, the kind map,
the fidelity knob (charter risk #4), and the bbox-bound-the-roads rule.
- E's 5 existing v1 caches re-validate clean; selfcheck ALL GREEN.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
F wired the last mile of the real-map scout and gated the whole matrix.
- index.js barrel re-exports registerTownCache/validateTownCache; the shell fetches + registers
assets/towns/<key>.json before generatePlanFor when the town isn't a checked-in fixture — so
?plansrc=osm&town=bendigo_real boots a REAL AU town in the engine (the gap: E's caches were on disk +
in the node selfcheck, but the shell never loaded them → silent Melbourne fallback). Fail-soft
(404 / invalid / ?noassets -> fixture fallback); opt-in, never on the default or ?classic path.
- tools/qa/town_matrix.py (opt-in qa.sh --matrix): boot / determinism / budget / district / ?noassets
across 10 towns (2 synthetic hero seeds + 3 osm fixtures + E's 5 real AU caches). ALL 5x10 GREEN —
Castlemaine at the 6-shop floor still lands a 3-venue district. The existing CityPlan contract eats real
Australian geography: the evidence John charters v4 = THE REAL MAP on.
- docs: LANE_F_NOTES §17 (the town x gate table), README real-town row, F-progress R17.
qa.sh --strict --soak GREEN 7/7. Verified: ?plansrc=osm&town=bendigo_real -> Bendigo, 9 shops, 3 venues, 0 errors.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Street peds occasionally sit during a window-shop loiter (verandah/footpath
edge). makeActor gains an opt-in setSitting with a SCALE-AWARE re-plant (the
R16 spawnRig re-plant was scale-naive; the sim scales these rigs per-citizen),
played only on sit so walkers/placeholders/?classic are byte-identical. Seeded
via a dedicated benchsit rng stream (no shift to turn/loit/patron; identity
signature unchanged). Classic inert at the source (roll gated on sitClip).
In-shell (default, seed 20261990): head bone 1.466->1.070, bone span
1.667->1.299 (in seated gate [0.9,1.5]), feet re-plant to y=0 and hold through
sit-idle, stand-up restores inner.y 0.929 byte-for-byte, procitySeated tagged,
innerRotY:pi throughout. 9 distinct peds sat naturally per soak. Classic:
fleetSitClip:false, everSat:0. Plan golden 0x3fa36874 unchanged. Draws 35/tris 6k.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Ledger #4 (warm-up): publish.py --verify no longer false-flags non-procity_ published assets
(sit.glb, vintage-cash-register). It compares the _published.json name record against the depot's
full file list (fetch_depot_files) instead of a procity_ prefix filter; only OUR namespace flags as
unrecorded. --verify now: all 39 recorded assets live.
Ledger #6 (the scout): new pipeline/build_towns.py ports thriftgod build_city()'s Overpass route to
per-town caches in A's procity-town-cache/1 contract. 5 real compact AU towns under web/assets/towns/
(katoomba/newtown/fremantle/bendigo/castlemaine, 6-21 shops, spans <2.4km): bounded Overpass query ->
raw cached in-repo (re-runs never re-fetch) -> dedupe/suburb-fill/parody (trademarked chains only;
indie shops keep real OSM names)/type-map. Deterministic (byte-identical re-run, no RNG). ODbL:
SOURCES.md licence table + license/attribution on every cache. All 5 boot through the full
structural+gig suites (selfcheck 16596/16596 green; synthetic golden 0x3fa36874 unmoved). Overpass
fetch John-authorized (Standing #2).
-> Lane A: caches live under web/assets/towns/; pin the per-town goldens selfcheck printed (katoomba
0x61760542 newtown 0x95978b45 fremantle 0xee9fd09a bendigo 0x5e3e76a6 castlemaine 0xf1a628a1) +
feed through plan_osm. E->A handshake.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The R16 seated drummer sat bolt-upright (the sit clip can't carry a pelvic tilt
— Hips.quaternion is unusable on these rigs). Fix: a small constant forward torso
lean applied POST-MIX. spawnRig({seated}) caches the mid-spine bone (seatBone);
seatedLean(spine) rotates it forward ~0.34 rad each frame after the mixer resets
it (constant offset, not accumulation); band.js calls it for m.seated members.
The drummer leans into the kit — head 0.28m forward of the hips, reads as a real
drummer from the side.
Byte-identical (my signature): standing rigs get no seatBone (found only in the
seated block) and seatedLean runs for seated figs only. Verified with the lean
live — walkers 188/188 face travel, keeper standing rig (stature 1.718,
hasSeatBone:false, innerRotY pi). No existing consumer moves.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
