tower / files / decisions.md

decisions.md

Blame: the commit that last changed each line · file · history

`866c7f96` 2026-09-07 Pavel Safronov	1	# Tower — Decisions
	2
	3	Standing record of choices made and why, so they can be revisited deliberately
	4	rather than rediscovered. Newest sections append at the end.
	5
	6	---
	7
	8	## D-001 — Cylinder geometry: r = 50 km, L = 540 km
	9
	10	Decided. Radius 50 km, length 540 km, giving 169,646 km² of inner surface.
	11
	12	Considered three shapes, all matching Wisconsin's area at 1 g:
	13
	14	\| \| radius \| length \| spin period \| character \|
	15	\|---\|---\|---\|---\|---\|
	16	\| Long & narrow \| 8 km \| 3,375 km \| 3.0 min \| ceiling, not sky; absurd aspect ratio \|
	17	\| Balanced \| 50 km \| 540 km \| 7.5 min \| far side visible overhead through haze \|
	18	\| Wide & short \| 150 km \| 180 km \| 13.0 min \| a drum; far cap looms as a wall \|
	19
	20	Why balanced. It is the only one where the sky reads as sky. At 8 km the
	21	roof is 16 km up and oppressive; at 150 km the far side is 300 km away and lost
	22	entirely. At 50 km the opposite land is 100 km overhead — far enough to be pale
	23	and dreamlike, near enough to be unmistakably land above you. That image is the
	24	point of the whole project, and this geometry is what produces it.
	25
	26	The 7.5-minute spin also sits in a good window: fast enough that star motion is
	27	obvious in a few seconds of watching, slow enough not to look like a fairground
	28	ride.
	29
	30	---
	31
	32	## D-002 — Slats sized so land area equals Wisconsin's land area
	33
	34	Decided. 8 slats, 6.80 km wide, leaving 140,268 km² of land.
	35
	36	The brief specified a Wisconsin-sized inner surface. Wisconsin has two areas —
	37	total 169,635 km² and land 140,268 km². Sizing the slats to consume exactly the
	38	difference makes both true at once: **Tower's total inner surface is Wisconsin's
	39	total area, and Tower's habitable land is Wisconsin's land area. The slats are its
	40	lakes.**
	41
	42	This is a coincidence discovered while computing, not a constraint imposed. It
	43	costs nothing and it is too good to pass up.
	44
	45	*Superseded by D-017: there are no slats, so Tower's land is its entire inner
	46	surface, 169,646 km² — Wisconsin's total area. The second coincidence recorded
	47	below no longer applies.*
	48
`cf37225c` 2026-09-07 Pavel Safronov	49	Slat count is left as a tunable constant. *The runtime control is disabled and
	50	hidden as of D-033; the derivation below is unchanged and still tested.* The width
	51	follows from the count:
`866c7f96` 2026-09-07 Pavel Safronov	52
	53	\| slats \| each \| land strips between \|
	54	\|---\|---\|---\|
	55	\| 6 \| 9.07 km \| 43.3 km \|
	56	\| 8 \| 6.80 km \| 32.5 km \|
	57	\| 12 \| 4.53 km \| 21.7 km \|
	58	\| 24 \| 2.27 km \| 10.8 km \|
	59
	60	8 chosen because 32.5 km strips read as regions — a day's travel across, room
	61	for real geography. At 24 the world becomes narrow ribbons where every point is
	62	within 5 km of a wall of stars, which is striking but claustrophobic. Revisit
	63	after seeing the map view; this is cheap to change.
	64
	65	*"Revisit after seeing the map view" is now literal — D-014 makes the count a
	66	runtime control from 0 to 16, and 8 is the default rather than a fixed value.*
	67
	68	---
	69
	70	## D-003 — Widen the beam to 10 km; the physics now works unbroken
	71
	72	Decided. The beam is 10 km in circumference, carrying 54.3 TW at 5,000×
	73	concentration. No physical law is bent. *This supersedes an earlier version of
	74	D-003 that accepted an étendue violation in order to keep a 1 km beam.*
	75
	76	The mistake in the first pass. I treated the 1 km beam as fixed and concluded
	77	that Earth-like brightness was impossible without an active, physics-breaking
	78	waveguide. That was wrong, because it conflated two independent parameters:
	79
	80	- Collector aperture sets how much power is gathered. This was never limited
	81	by anything but engineering scale.
	82	- Beam cross-section sets how hard the light must be concentrated to fit
	83	through. This is where étendue binds — passive optics cannot exceed the sun's own
	84	surface radiance, ~46,223× at 1 AU, or 62.9 MW/m².
	85
	86	Holding the beam at 1 km forced 313,000× and broke the second law. Widening the
	87	beam relaxes the concentration requirement quadratically and costs nothing else.
	88
	89	Where the number came from. Targeting 54.3 TW — Earth clear-sky noon at local
	90	midday, see design §1.3 — against candidate concentrations:
	91
	92	\| concentration \| beam circumference needed \| status \|
	93	\|---\|---\|---\|
	94	\| 46,223× (étendue limit) \| 3.29 km \| thermodynamically impossible to exceed \|
	95	\| 10,000× \| 7.08 km \| beyond current practice \|
	96	\| 5,000× \| 10.01 km \| Odeillo solar furnace reaches ~10 MW/m² today \|
	97	\| 1,000× \| 22.39 km \| routine, but a conspicuously fat beam \|
	98
	99	5,000× chosen. It lands the beam on a round 10 km, it is demonstrated
	100	terrestrial technology rather than an extrapolation, and it sits at 11 % of the
	101	étendue limit — 9.2× of headroom in reserve. Diffuser mirrors at 99.9 %
	102	reflectivity reach thermal equilibrium at 316 °C in that beam, an ordinary
	103	radiator problem rather than an exotic one.
	104
	105	Cost to the imagery: none. At 1,592 m radius seen from 50 km below, the beam is
	106	3.65° wide — about seven sun-widths. It still reads as a luminous rod running the
	107	length of the sky, which is the image the 1 km figure was protecting.
	108
	109	The unlocked consequence. Because Tower now has surplus capability rather
	110	than a deficit, brightness becomes a setting rather than a fixed value. The
	111	diffusers are software-controlled; illumination can run anywhere from 30 W/m²
	112	permanent dusk to 320 W/m² clear-sky noon. The "embrace the dimness" option
	113	weighed and rejected in the first pass is no longer an alternative to this
	114	decision — it is a position on its slider. The app should expose that control.
	115
	116	## D-004 — Terminator orbit, actively precessed
	117
	118	Decided. Orbital plane held normal to the Earth–Sun line, precessing 360°/year.
	119
	120	The brief requires that one cap face the sun permanently. Any orbit whose plane
	121	contains the sun line puts Tower through Earth's shadow once per revolution, which
	122	would interrupt the beam. A terminator orbit never enters shadow.
	123
	124	The cost is that the plane must be reoriented ~1°/day against the gyroscopic
	125	stiffness of a spinning 540 km cylinder. This is enormous and continuous. Canon
	126	adopts it as a feature: Tower's existence is a permanent act of maintenance, not a
	127	stable configuration. That is more interesting than a free lunch.
	128
	129	---
	130
	131	## D-005 — Vanilla TypeScript, no UI framework
	132
	133	Decided. Vite + TypeScript + Three.js, no React/Vue/Svelte.
	134
	135	The app is three canvases, a picker and a control panel, sharing one small state
	136	object. A framework would add build weight and a rendering model that fights
	137	Three.js's own loop, in exchange for conveniences this UI does not need. Revisit
	138	only if the control panel grows genuinely complex.
	139
	140	Canvas 2D rather than Three.js for the map: it is a flat unrolled rectangle with no
	141	projection distortion, and 2D canvas draws that natively.
	142
	143	---
	144
	145	## D-006 — Interior fidelity: real geometry, simple ground
	146
	147	Decided. The shell, curvature, haze, beam, diffusers, slats and stars are all
	148	genuinely modelled and correct. The ground is procedural terrain and texture only —
	149	no scenery meshes, no settlements, no vegetation.
	150
	151	The illusion of standing inside Tower comes almost entirely from the sky: land
	152	curving up into blue, the lit shaft overhead, stars wheeling past every 7.5
	153	minutes. Detailed ground scenery adds cost proportional to realism and contributes
	154	little to that illusion. Deferred, not abandoned.
	155
	156	---
	157
	158	## D-007 — All canon derived from four constants
	159
	160	Decided. `canon/constants.ts` exposes radius, length, slat count and target
	161	land area as free parameters. Everything else — spin period, rim speed, Coriolis
	162	parameter, slat width, angular spacing, axis pressure, beam power — is computed.
	163
	164	The alternative, transcribing the figures in `design.md` as literals, would drift
	165	the moment any parameter is revisited (and D-002 and D-003 both flag likely
	166	revisits). Derivation makes the whole app answer coherently to a changed radius.
	167
	168	---
	169
	170	## D-008 — Diffusers are a 1.2 cm swarm; the beam glows from air, not from them
	171
`59c12827` 2026-09-28 Pavel Safronov	172	*Amended by D-074: individually the mirrorlets stay invisible, but by day the swarm
	173	gathers into one shoal on purpose, and that shoal is Tower's sun.*
	174
`866c7f96` 2026-09-07 Pavel Safronov	175	Decided. 78 billion 1.2 cm mirrorlets, and the beam's visible glow is
	176	atmospheric scattering that does not depend on them.
	177
	178	The question. Would a modest number of large diffusers read as a string of
	179	separate suns rather than a continuous glowing rod?
	180
	181	Yes. At 50 km, naked-eye resolution is 14.5 m; anything larger shows a visible
	182	disc. But shrinking below 14.5 m is not sufficient, which was the non-obvious part.
	183	A flat mirror preserves radiance, so an aimed mirrorlet shows the *sun's own
	184	surface brightness* over a small angle — unresolved but blinding, exactly as stars
	185	are unresolved and plainly visible. Magnitudes: 1 m → −13.6 (brighter than a full
	186	moon), 10 cm → −8.6, 1.2 cm → −4.0, which finally sits under the daylight
	187	visibility threshold. The binding constraint is flux, not angular size, and it is
	188	two orders of magnitude tighter than the resolution limit implies.
	189
	190	The better answer, found while checking. The beam glows regardless. Rayleigh
	191	scattering of a 6.8 MW/m² beam through 3.18 km of 5.2 kPa axis air gives 779
	192	W/m²/sr — 26× blue sky, 25,700× below the solar disc. The continuous-glow imagery
	193	is therefore a robust property of beam plus atmosphere, not a consequence of
	194	getting the diffuser design right. Canon leans on this; the swarm sizing is then
	195	only about keeping individual glints out of the sky.
	196
	197	Consequences taken. 11.3 km² of mirror at 1.2 cm is 78 billion units massing
	198	158 tonnes, spaced 3.8 m apart. Optical depth 1.41 leaves 24 % of the light
	199	reaching the far cap, which satisfies the brief's "shines to the far end" without
	200	stipulating it. *The passive drift-and-recirculate mechanism first recorded here is
	201	superseded by D-009 — it does not work.*
	202
	203	Story consequence. 78 billion independently aimable mirrors make illumination a
	204	render target rather than a schedule. Sunsets are compositing, not optics.
	205
	206	---
	207
	208	## D-009 — Tower flies every foil itself, because nothing else works
	209
	210	Decided. All 78 billion foils are under continuous individual control by Tower.
	211	There is no recirculation machinery at either cap. *This supersedes the passive
	212	drift mechanism in D-008.*
	213
	214	Origin. Proposed by the author as characterisation — Tower is an intelligent
	215	being, so it controls each foil directly and moves the swarm partly for its own
	216	enjoyment. Checking the physics turned this from a flourish into a requirement.
	217
	218	A passive swarm is impossible. D-008 assumed the axis was effectively
	219	weightless. It is not. At the beam's edge, 1,592 m out, g = ω²s = 0.312 m/s². A
	220	2 mg foil sinks outward at 0.26 m/s and falls clear of the beam in 1.7 hours. A
	221	drifting swarm dissolves in an afternoon. Continuous active flight is the only
	222	configuration that holds together at all.
	223
	224	They can easily afford it. Each foil intercepts 979 W; harvesting 0.1 % yields
	225	~1 W on a 2 mg body — 489,600 W/kg, ~50× a jet engine. Flying up-beam at 1.5 m/s
	226	costs 15 µW against 14,400 µW harvestable from ambient world-light: **a 1000×
	227	margin, without touching the beam.** So the foils fly home unaided, no structure is
	228	required at the far cap, and the exterior render is simpler than D-008 implied.
	229
	230	The interesting part is the slack. With 1000× of margin, efficiency is not a
	231	constraint on Tower's behaviour. It can move the swarm in spirals, shoals and
	232	patterns that cost energy for no reason except preference, and nothing about the
	233	world degrades. Play is not a luxury Tower affords despite the physics — it is
	234	budget the physics hands over for free. The weather is therefore not a system
	235	Tower operates but something closer to body language.
	236
	237	Falling out of this: aiming is human-scale. A 1.2 cm mirror is
	238	diffraction-limited to 4.6×10⁻⁵ rad, giving a 2.3 m spot at 50 km. Tower can
	239	place a glint on one person that a bystander five metres away cannot see. Offered
	240	to the author as a story mechanic; canon records the optics either way.
	241
	242	And: mirrorfall. A failed foil sinks out of the beam, takes ~2 days to descend,
	243	and lands at 0.48 m/s — snowflake speed. A permanent slow glitter falls out of the
	244	sky.
	245
	246	---
	247
	248	## D-010 — The app is a companion to a novel, not a tech demo
	249
	250	Decided. Tower is the setting of a novel the author is writing. The app serves
	251	that purpose first.
	252
	253	Two consequences, recorded because they change priorities rather than mechanics:
	254
	255	- Coherence outranks spectacle. These documents are series-bible material and
	256	the prose is downstream of them. A number that looks better but contradicts
	257	something is worse than a duller number that holds. This is already how the
	258	physics has been handled; it is now the stated rule.
	259	- Shareable permalinks to a point move into scope (phase 4, previously
	260	deferred). "Stand here and look at this" is the entire value of a companion
	261	piece, and a link is how a reader receives it. Named locations from the novel
	262	should be markable on the map for the same reason.
	263
	264	Not yet decided, and deliberately left open: whether the app should ever surface
	265	narrative content — place names, events, characters — or remain a pure physical
	266	model that the prose refers to from outside. That is the author's call and does not
	267	need answering to proceed.
	268
	269	---
	270
	271	## D-011 — Writer mode: the app is an instrument, and the world file is git
	272
	273	Decided. Two modes over one committed `world.json`. Readers get a finished map;
	274	the author gets editing plus a set of measuring instruments. No backend.
	275
`9d5a061c` 2026-09-27 Pavel Safronov	276	*Superseded in part by D-072: there is one mode now, and the world is edited in
	277	`content/world.json` rather than in the browser. "The world file is git" and "no
	278	backend" stand.*
	279
`866c7f96` 2026-09-07 Pavel Safronov	280	Origin. The author intends to use the app to write — placing things and asking
	281	questions like "can the far face be seen from the midpoint?", "where does a central
	282	embassy go?", "can a lake overhang?". That makes measurement, not rendering, the
	283	core feature. D-010 said the app serves the novel; this is what that costs.
	284
	285	Why those questions justify the build. Two of the three had non-obvious answers
	286	that only fell out of computation, and both became canon (§1.8): the far cap is
	287	invisible from the midpoint at 4 % transmission, and — unexpectedly — the far side
	288	of the world overhead is clearer than the ground at the same distance (78 % vs
	289	31 % at 100 km), because a sightline across the interior rises through thin axis
	290	air. An author cannot guess these. The instrument earns its place immediately.
	291
	292	Static, file-based, no backend. GitLab Pages serves static files, so: edit in
	293	browser → export → commit → publish. The world lives in git beside the prose,
	294	diffable and versioned, with no service to run or pay for. Drafts sit in
	295	`localStorage` between exports.
	296
	297	Coordinates in the file are canon (θ, z), never map pixels, so the world
	298	survives any change to how the map is drawn. Consistent with D-007.
	299
`a088e5d5` 2026-09-17 Pavel Safronov	300	*Correction (D-068): as built, writer mode has always been the default and `?reader=1`
	301	turns it off; `?writer=1` does nothing. The reasoning below holds either way.*
	302
`866c7f96` 2026-09-07 Pavel Safronov	303	Writer mode is a UI mode, not a security boundary. `?writer=1` is guessable and
	304	that is fine — an editor with no persistence and no backend can only alter the
	305	visitor's own copy. Recorded explicitly so it is never mistaken for access control
	306	and never load-bearing for privacy.
	307
	308	The real privacy risk is the repo, not the flag. Working notes must not ship to
	309	readers, but anything committed to a public repository is public whatever a
	310	`"public": false` field says. Hence two export files: `world.json` (public entries,
	311	committed) and `world.private.json` (everything, gitignored). Whether private notes
	312	should be version-controlled anyway — private repo, or a second one — is left to
	313	the author and does not block anything.
	314
	315	Sequenced early. The sightline probe lands in phase 6 but should be built first
	316	within it: it is useful the day it exists, and §1.8 is evidence the author is
	317	already asking questions only it can answer.
	318
	319	---
	320
	321	## D-012 — Ids follow the name until pinned by hand
	322
	323	Decided. A place's id is re-derived from its name while it still matches it,
	324	and freezes permanently once the author edits the id field.
	325
`9d5a061c` 2026-09-27 Pavel Safronov	326	*Retired by D-072 along with the editor. Ids are written by hand in the file now, and
	327	the reason below — a permalink is worthless if the id churns — still governs changing
	328	one.*
	329
`866c7f96` 2026-09-07 Pavel Safronov	330	The tension. design.md §3.2 makes ids permalink targets, which argues for
	331	never changing them. But places are created from a placeholder name ("New city"),
	332	so a never-changing id produces a world file of `new-city`, `new-city-2`,
	333	`new-city-7` — unreadable in git diffs and unusable for hand-editing, both of which
	334	D-011 counts as reasons for the file-based design in the first place.
	335
	336	Resolution. During world-building, readability wins; after publication,
	337	stability wins; and the author is the only one who knows which phase a place is in.
	338	So the id tracks the name automatically until they say otherwise, and the id field
	339	is shown in the editor so saying otherwise is one edit. `isAutoId()` in
	340	`src/world/schema.ts` is the test, and is unit-tested directly.
	341
	342	Found by running it. The first implementation derived the id once at creation
	343	and never updated it, which the browser check caught immediately — every exported
	344	place was called `new-city`. Two further bugs surfaced the same way: renaming
	345	blanked the editor (the store re-rendered before the selection moved to the new id),
	346	and clicking an existing marker while a creation tool was active dragged the old
	347	place instead of creating a new one. None were visible in the unit tests.
	348
	349	---
	350
	351	## D-013 — The map is bounded, and cut at a slat centre
	352
	353	Decided. The map shows the inner surface once, as a bordered rectangle, rather
	354	than tiling vertically. The cut falls at the centre of the last slat.
	355
	356	Why not tile. The circumference wraps, so tiling is the honest rendering and it
	357	makes places near the seam continuous. But it also makes the world look boundless:
	358	with no edge, the author cannot see the extent of what they are building, and
	359	repeated strip labels read as more strips rather than as repeats. For a map — a
	360	thing whose job is to show where things are relative to the whole — legible extent
	361	beats a hidden seam.
	362
	363	Why cut there. Any bounded map needs an edge, and the edge destroys whatever it
	364	falls on. Land strip 0 is centred on θ = 0, so the obvious cut would slice a whole
	365	strip in half across the top and bottom margins. Cutting at a slat centre instead
	366	leaves every land strip whole and splits only a band of stars, which carries no
	367	geography. This is exactly why Earth's maps are cut at the antimeridian. With zero
	368	slats there is nothing to protect, so the cut returns to θ = 0.
	369
	370	Consequences taken. Panning is clamped so the surface cannot be lost off-screen
	371	— without tiling there is nothing to pan back into. A sightline whose short route
	372	crosses the edge is drawn as two segments, entering and leaving opposite edges, so
	373	the drawn path always matches the measured distance rather than contradicting it.
	374
	375	---
	376
	377	## D-014 — Slat count is a runtime control, 0 to 16
	378
	379	Decided. `SLAT_COUNT` stops being a compile-time constant. The count lives in
	380	`src/canon/slats.ts`, adjustable from 0 to 16, persisted per browser, with every
	381	slat-derived figure computed from it.
	382
	383	Why. D-002 already called this "the most tunable number in canon" and said to
	384	revisit it after seeing the map. Making it a slider makes that revisiting real
	385	rather than a code edit, and the author is choosing the shape of a world rather
	386	than configuring software.
	387
	388	The invariant that makes it safe. D-002 fixes the total slat width from the
	389	land-area target, and that total does not depend on how many slats it is divided
	390	into. So **land area stays exactly Wisconsin's 140,268 km² at every count from 1 to
	391	16** — the coincidence survives the whole range, and only the per-slat width moves
	392	(18.1 km at 3 slats, 3.40 km at 16).
	393
	394	Zero is a genuinely different world, and the app says so. With no slats the
	395	land becomes the entire inner surface, 169,646 km², and nobody in Tower can ever
	396	see a star. That breaks the D-002 coincidence, so the control states it plainly
	397	instead of silently reporting a changed number. Rounding that away would have been
	398	the easy option and the wrong one — it is exactly the kind of quiet inconsistency
	399	the documentation gate exists to catch.
	400
	401	Cost. Slat geometry moved from constants to functions, which rippled through
	402	`geometry.ts`, the map, the instruments and the tests. Worth it: D-007's principle
	403	was always that canon is derived rather than transcribed, and this is that
	404	principle extended one parameter further.
	405
	406	---
	407
	408	## D-015 — Hydrology: shaped terrain, no sideways Coriolis, climate from the slats
	409
	410	Decided. Recorded as canon in design.md §1.9, and it amends §1.4.
	411
	412	The provable part. The floor's tangent plane is spanned by θ̂ and ẑ and the spin
	413	axis is ẑ, so surface flow gets `ẑ × ẑ = 0` axially and a purely radial `ẑ × θ̂ = −ŝ`
	414	circumferentially. The Coriolis force has no component inside the surface at any
	415	heading. Tower's rivers do not curve, and Earth's entire rotational repertoire —
	416	hemispheres, Baer's law, gyres, cyclones — is absent. This is not a simplification;
	417	it falls out of the geometry.
	418
	419	Where the force goes instead. Into vertical motion. Waterfalls lean retrograde
	420	as h^1.5 (133 m over a 1 km drop, 377 m over 2 km), and rain hangs at a permanent
	421	1.47° slant that never varies. Rising things lean the opposite way. A world where
	422	every falling thing drifts one way and every rising thing the other is a strong,
	423	cheap, always-visible signature.
	424
	425	The absence that matters more. Potential depends only on radius, so the floor is
	426	one equipotential and a perfect Tower has no rivers at all. Every river exists
	427	because the floor was shaped, there is no tectonic uplift, and erosion is therefore
	428	one-way — ~30 Myr to erase a 3 km range. Tower's hydrology is maintenance, which is
	429	D-004's theme arriving from a new direction.
	430
	431	An inconsistency found and fixed, not smoothed over. §1.4 specified a
	432	glass/diamond composite slat fill. Diamond conducts ~2,000 W/m·K; such a slat
	433	would radiate 385 W/m² to space, exceeding Tower's entire 320 W/m² mean insolation
	434	and freezing the world. Canon now specifies low-conductivity glass, which loses
	435	~28 W/m² through 5 m. Raised explicitly per the documentation gate rather than
	436	quietly retconned.
	437
	438	The gift in that loss. Tower has no latitude — every point is equidistant from
	439	the light, so there is no natural climate gradient whatsoever, which is a real
	440	worldbuilding hole. The leaking slats fill it: eight cool belts running the full
	441	540 km, parallel to the axis. Tower's climate bands run lengthwise, not around,
	442	and the slat count sets the climate as well as the geography — further support for
	443	D-014 making it a control.
	444
	445	---
	446
	447	## D-016 — Slat orientation: axial or rings — OPEN
	448
	449	Not decided. Canon keeps axial strips for now. Rings are a live alternative and
	450	this records what the choice costs, so it can be made deliberately.
	451
	452	Why it matters more than it looks. Slat layout turns out to set three things at
	453	once — the shape of Tower's countries, its entire climate map, and what its people
	454	can see of the universe. It is not a detail of the hull.
	455
	456	Geometry. At equal total area: rings are 11.69 km wide with 55.8 km land bands.
	457	Each land region is then 314 × 55.8 km rather than 32.5 × 540 km — the same area at
	458	aspect 5.6:1 instead of 16.6:1, a country instead of a corridor. Ring bands are also
	459	closed loops: you can walk 314 km around your whole region and come home without
	460	crossing a window. Axial strips are ribbons with two dead ends at the caps.
	461
	462	Climate. Axial strips put the cool belts lengthwise, so a 540 km river runs its
	463	whole length through identical country. Rings put them crosswise, giving Tower a
	464	genuine latitude analogue — a long river crosses all eight bands like the Nile.
	465
	466	The sky, which is the deep one. The spin axis is the celestial pole, so stars
	467	move along the celestial equator. Orientation decides whether a window lies across
	468	that motion or along it:
	469
	470	- Axial: 3.90° across the motion — each star crosses in 4.9 s, but the strip
	471	spans pole to pole so essentially the whole sky passes through, strobing.
	472	- Rings: 6.70° along the motion — a star rides the band horizon to horizon over
	473	3.7 minutes, but only the 5.8 % of sky within 3.35° of the celestial equator
	474	is ever visible. The rest is never seen at all.
	475
	476	Earth follows the same rule: a five-second flicker under axial slats, a slow bright
	477	traveller under rings. The sun is at the celestial pole and invisible through both.
	478
	479	The cost of rings. Hoop stress dominates the structure — ρv² is 785 MPa in a
	480	composite shell plus ~500 MPa of air pressure — and it runs along θ. Axial slats sit
	481	across that path, so ribs bridge the gap and the glass carries only local pressure.
	482	A ring runs along the path and is continuous in θ, so **the window itself must carry
	483	~1.2 GPa**. Glass cannot; it would need hoop cables through every pane, putting bars
	484	across every view forever. Characterful, but a real price axial slats do not pay.
	485
	486	Consequence for the app if rings are chosen. D-013 cuts the map at a slat centre
	487	because there is an empty θ to cut at. With rings there is not — every θ is land —
	488	so the map seam would have to slice through inhabited country, or the map would need
	489	reorienting. Worth knowing before the choice is made rather than after.
	490
	491	---
	492
	493	## D-017 — The hull is sealed; the only views out are at the caps
	494
	495	Decided by the author. No slats, in any orientation. The cylinder wall is
	496	opaque and unbroken for its full 540 km. Supersedes D-002's slat sizing, D-014's
	497	default, and closes D-016.
	498
	499	What it costs. The wheeling star field was canon's answer to "what is most alien
	500	about standing inside Tower" (§1.2) and the interior view's headline feature. It is
	501	gone. Tower also loses its only passive climate driver and its only natural source
	502	of geographic variation.
	503
`da6bcb0e` 2026-09-08 Pavel Safronov	504	*Partly recovered by D-045, and worth reading as a correction to the paragraph
	505	above rather than a reversal of the decision. The wall is still opaque and unbroken
	506	for its full 540 km — that is what D-017 decided and it stands. What came back is
	507	the pinwheel, through the far cap window, where §1.4 had always said it would be:
	508	one sky, at one end of the world, at night. It is no longer everywhere, which is
	509	the whole difference between the slats and a window.*
	510
`866c7f96` 2026-09-07 Pavel Safronov	511	What it buys, which is more.
	512
	513	- One unbroken surface. 169,646 km² of continuous land — Wisconsin's total
	514	area, the coincidence in §1.1 now standing alone and cleaner. Nothing divides the
	515	world into strips; you can walk anywhere, and a river can circle the whole 314 km
	516	circumference and close on itself.
	517	- The spin inverts. With no windows and a co-rotating far side, *nothing in the
	518	sky moves at all*. But rain still hangs at a permanent 1.47° slant and a 1 km
	519	waterfall still lands 133 m off its cliff. **The sky is still and the water is
	520	not** — a stranger signature than wheeling stars, and one that shows up in every
	521	puddle rather than only when you look up.
	522	- The caps become the only view, and it is a good one. A cap window looks along
	523	the spin axis — the celestial pole — so stars do not sweep but *rotate about a
	524	fixed centre*, one turn per 7.5 minutes, one motionless star at the middle. The
	525	caps are also walls rather than floors: gravity there is tangent to their surface,
	526	so each is a 50 km cliff at the end of the world with its windows at the base.
	527	- Nobody can ever see Earth. The terminator orbit (D-004) puts Earth ~90° off
	528	the axis, out the side, which is now solid hull. A habitat in Earth orbit whose
	529	people have never seen the planet they orbit.
	530
	531	The consequence the author accepted explicitly. With no latitude and no windows,
	532	nothing breaks the thermal uniformity, so there is no passive weather whatsoever.
	533	Every cloud, every rainfall and every jet stream is Tower deliberately aiming light.
	534	This is D-009 reaching its end point: the swarm was already Tower's body language,
	535	and is now its lungs. There is no autonomic layer beneath. **If Tower's attention
	536	lapses, the rain stops.**
	537
	538	---
	539
	540	## D-018 — Rivers are a closed, pumped circulation, and Tower can easily afford it
	541
	542	Decided. Tower's rivers follow the architecture the author proposed: reservoir →
	543	shaped descent → lake → lift back to the reservoir. Source and sink are one system.
	544
	545	Why it is the only option. The floor is an equipotential (D-015) so there is no
	546	natural slope, and the sealed hull (D-017) leaves no natural water cycle. Water in
	547	Tower moves only because something moves it.
	548
	549	Why it is not a compromise. Costed against the 54.3 TW budget:
	550
	551	\| \| \| \|
	552	\|---\|---\|---\|
	553	\| Earth-like rainfall, 1,000 mm/yr \| 12.2 TW \| 22.4 % — Earth spends the same share \|
	554	\| Recirculating all runoff (1,792 m³/s, a Rhine) against 500 m \| 8.8 GW \| 0.016 % \|
	555
	556	Pumping the entire hydrology of the world is a rounding error — 6,000× headroom.
	557	And full Earth-intensity weather is affordable at the same fraction of budget Earth
	558	itself spends. Tower does not have to choose.
	559
	560	Canon uses both, because they give different rivers. A pumped river *begins at a
	561	machine* — walk to its source and find a wall of water leaving a structure. A
	562	rain-fed river begins at a cloud Tower made over the highlands, and looks entirely
	563	natural while being entirely deliberate. Made weather runs the world's general
	564	hydrology; pumps run the watercourses Tower wants under precise control.
	565
	566	Two consequences worth keeping. A closed loop has no net descent, so a river
	567	circling the world must be pumped — which is the same architecture, closed the whole
	568	way round. And since nothing evaporates to space, **Tower's water inventory is
	569	conserved and exactly known, forever.**
	570
	571	---
	572
	573	## D-019 — Geography is authored as data; terrain is generated from it
	574
	575	Decided. `content/geography.json` holds the authored geography. `npm run terrain`
	576	(tools/terrain.mjs) reads it and emits a height map, a colour map, a corridor detail
	577	render, and a seed `world.json` of places.
	578
	579	Why data first. The author asked to work the geography out before returning to
	580	app code. A JSON source plus a generator means the map is re-derivable from one
	581	readable file: move the river by editing eighteen waypoints, not by repainting.
	582	Coordinates are canon (θ, z) throughout, consistent with D-007 and D-011.
	583
	584	No dependencies. Node's `zlib` is enough to write PNGs directly (tools/png.mjs,
	585	~60 lines for 16-bit greyscale and 8-bit RGB), so the tool adds nothing to
	586	`package.json`. Terrain noise is a small value-noise/fBm module.
	587
	588	Three bugs the renders caught that no test would have.
	589
	590	- A float hash. The noise hash multiplied large integers in floating point,
	591	silently losing precision and streaking the whole map with axis-aligned lines.
	592	Fixed with `Math.imul` throughout.
	593	- A fractional wrap period. The circumference is not a round multiple of any
	594	useful feature size, so the noise lattice did not close on itself and tore at the
	595	seam. Sample coordinates on the θ axis are now snapped to a whole number of
	596	cells; the resulting few-percent anisotropy is invisible.
	597	- A discontinuous river. Valley geometry switched between reaches abruptly,
	598	stepping the valley from 5.4 km wide to 18.2 km in one pixel column and drawing a
	599	visible wall across the map. Reach parameters now blend along the river. A
	600	`--probe` flag prints elevation jumps along a row, which is how it was found.
	601
	602	Known gap, deliberately not fixed here. The seed `world.json` contains
	603	`polyline` geometry (the river, the canyon) and kinds the app does not yet know.
	604	`parseWorld` accepts only points and polygons, so those two places would be dropped
	605	silently on load. App work is paused by request; this is the first thing to fix when
	606	it resumes. Closed by D-022.
	607
	608	---
	609
	610	## D-020 — The world is belts, and the desert rings it
	611
	612	Decided by the author. The desert and its three ridges ring the world; the
	613	capital's water becomes a sea running along the axis; both end caps get ring oceans;
	614	the range feeds four parallel rivers.
	615
	616	The ring desert is a barrier no Earth desert can be. 22,000 km², 13% of the
	617	surface — modest against Earth's ~33% — but on a cylinder there is no way around a
	618	band. Combined with three ringing ridges it is three concentric walls across Tower's
	619	waist.
	620
	621	The drainage was the hard part, and Earth answered it. Many rivers each cutting
	622	their own canyon would destroy the chokepoint. But on Earth most drainages do not
	623	cut through a rising ridge — they are defeated and diverted, and only an antecedent
	624	river keeps its course. So three of the four rivers are turned by First Ridge, run
	625	along its front gathering each other, and leave with the fourth through the one gap.
	626	The Narrows carries the entire world's runoff, ~392 m³/s. The chokepoint is
	627	strengthened by the extra rivers rather than diluted, and the hydrology is more
	628	realistic than a single river would have been.
	629
	630	The Deep's orientation is the decision, not its size. Long down the axis and
	631	narrow around, so it is crossed the short way — around θ, where the floor curves
	632	up and the far shore stands 2.6 km above eye level. Lengthwise it would be dead
	633	flat. It also gives Tower back a thermal flywheel: 19× the diurnal heat storage of
	634	bare ground, 5.9 days per kelvin at full power. D-017's "if Tower blinks the rain
	635	stops" becomes "the world has about a week of momentum" — a lapse now has time to
	636	become visible before it becomes fatal.
	637
	638	The cap seas were not chosen. Gravity on an end cap is tangent to its surface,
	639	so water cannot rest there and runs to the rim. Both ring oceans are consequences,
	640	not designs.
	641
	642	Skiing came out of §1.9.2 unprompted. A descending skier has radial velocity and
	643	therefore real Coriolis: a steady antispinward push on axial fall lines, and a
	644	weight change of up to 11% on circumferential ones. A retrograde run gets lighter as
	645	it accelerates; a prograde run grips harder. Every slope in Tower has a handedness,
	646	which is a sport that cannot exist on Earth.
	647
	648	Tool work this required: `ring` footprints, a `flood` op for troughs and bands,
	649	and a `tarn` op that fills from the bowl floor rather than the rim — the first
	650	version took its level from the rim mean, so water spilled downhill and every
	651	mountain lake rendered as a half-moon.
	652
	653	---
	654
	655	## D-021 — The terrain tool reports progress, and got 13× faster doing it
	656
	657	Prompted by the author: `npm run terrain` ran for a long time saying nothing.
	658	Adding progress output meant first working out where the time went, which turned a
	659	logging task into a performance one.
	660
	661	86 s → 6.4 s. Three causes, all found by measuring rather than guessing:
	662
	663	- No spatial index on the river carve. `opCarve` tested all 592 smoothed river
	664	segments against all 2.4 million pixels — ~2.9 billion distance tests across both
	665	renders. Segments are now binned into a 4 km grid and each pixel tests only its
	666	own cell. 86 s → 17.8 s.
	667	- Deflate level 9 on the colour maps. Hillshade noise compresses poorly, so
	668	level 9 spent 4.2 s on one map for a few percent of size. RGB now encodes at
	669	level 6: 8× faster for 10% larger files, which is the right trade for a tool run
	670	on every edit. The 16-bit height map stays at level 9 — smooth terrain under the
	671	Up filter is nearly all zero deltas, and it costs 0.1 s. 17.8 s → 8 s.
	672	- Domain warps computed before range checks. Several ops evaluated two noise
	673	functions per pixel then discovered the pixel was nowhere near the footprint.
	674	A conservative unwarped-z rejection now runs first. 8 s → 6.4 s.
	675
	676	The log is a table, not a stream. One row per relief op with its timing, then
	677	relief/colourise totals per view, then each PNG with its size, then the seed. The
	678	point is to see which feature is expensive when the file grows, not just that
	679	something is happening. `--quiet` collapses per-feature rows.
	680
	681	Worth recording as a habit: the request was for logging, and the logging is what
	682	exposed that the colour PNG cost more than the entire mountain range. Instrumenting
	683	first and optimising second found three real problems in one pass.
	684
	685	---
	686
	687	## D-022 — The app catches up with the decisions
	688
	689	Implemented. Five outstanding decisions, plus two bugs found while doing it.
	690
	691	The schema now matches the world (closes D-019's gap). `polyline` geometry
	692	joins point and polygon — a polygon needs three vertices for an interior, a
	693	polyline only needs two to go somewhere — and the kind list grows from six to
	694	thirteen (`mountain`, `river`, `canyon`, `forest`, `desert`, `tower`, `town`), with
	695	`AREA_KINDS` / `LINE_KINDS` deciding whether a traced shape closes. Before this the
	696	seed silently dropped The Long Water and The Narrows. A test now loads
`62458672` 2026-09-07 Pavel Safronov	697	`static/world.json` and asserts nothing is lost, so the two files cannot drift apart
`866c7f96` 2026-09-07 Pavel Safronov	698	again unnoticed.
	699
	700	The rendered terrain is drawn under the map, which is what the whole tool was
	701	for. Area regions become outlines rather than fills when it is showing: the raster
	702	already says where the desert is, and a translucent rectangle over it only fights.
	703
	704	A desync that would have been baffling. The app cut the map at θ = 0 with no
	705	slats; `content/geography.json` renders against θ = π. The raster would have landed
	706	half a world out of register with its own place markers. Both now use
	707	`MAP_CUT_WITHOUT_SLATS = π` — opposite the capital, which is D-013's antimeridian
	708	principle applied to habitation rather than geology — and a test asserts the two
	709	files agree.
	710
	711	Reader and writer modes (D-011). `?writer=1` reveals the tools, editor, canon
	712	controls and import/export; readers get the map, the place list, a read-only detail
	713	card and the sightline probe. A bug worth recording: `renderEditor` un-hid the
	714	editor whenever a place was selected, overriding the mode — so readers were shown
	715	the edit fields including private notes. Mode is now decided inside
	716	`renderEditor`, not applied around it. Hiding a panel elsewhere is not the same as
	717	never rendering it.
	718
	719	Permalinks (D-010). `#place=<id>` selects and centres a place; `#at=θ,z` restores
	720	a sightline endpoint. This is why D-012 made ids follow names only until pinned —
	721	a permalink is worthless if the id churns.
	722
	723	Labels are placed by priority and skipped on collision. Settlements outrank
	724	regions; overlapping labels are dropped rather than drawn on top of each other. With
	725	22 seeded places the mountain stations were an unreadable smear.
`5f38efd0` 2026-09-07 Pavel Safronov	726
	727	---
	728
	729	## D-023 — The exterior view, and what it is honest about
	730
	731	Implemented. Phase 3: Tower from outside, with the hull, both caps, the
	732	collector, the converging beam, the spin, and picking on the hull.
	733
	734	Canon numbers are the code's numbers. Scene units are kilometres and the axis
	735	lies along Z with the sun cap at z = 0, so a canon point is literally
	736	`(R cos θ, R sin θ, z)`. No conversion layer to get wrong.
	737
	738	Picking has to undo the spin. The hull turns, so a ray hit is in world space
	739	while the answer wanted is canon θ, which is fixed to the hull. Resolving the hit
	740	through `worldToLocal` handles it. Verified by clicking the same screen pixel at two
	741	spin phases 34° apart and confirming the reported θ moved by 0.588 rad against a
	742	0.593 rad phase change, with z identical — the pixel really is landing on different
	743	ground as Tower turns beneath it.
	744
`11a13dbf` 2026-09-07 Pavel Safronov	745	Three departures from physics, all stated in the code. Sunlight runs parallel to
`5f38efd0` 2026-09-07 Pavel Safronov	746	the axis, so the flank is edge-on and correctly renders almost black; fill lights
	747	make it readable. And in vacuum the beam between collector and cap has nothing to
	748	scatter off, so it would be invisible; it is drawn faintly because it is the point
	749	of the view. Same licence as the map's hillshade, and worth naming rather than
	750	letting a reader mistake either for a physical claim.
	751
	752	Three.js loads on first use of the view. It is by far the largest dependency and
	753	starts an animation loop; a reader who only opens the map should not pay for it.
	754
	755	Earth is not drawn. It is ~90° off the axis and hundreds of thousands of km
	756	away; at true scale it is invisible, and at any visible scale it is a lie about
	757	distance. Given how much of this world came out of taking distances seriously, a
	758	compressed Earth would be the wrong kind of shortcut. The sunward glow is honest —
	759	the sun really is that way — and D-017's point stands: from inside, Earth is never
	760	visible at all.
`11a13dbf` 2026-09-07 Pavel Safronov	761
	762	---
	763
	764	## D-024 — The terrain is wrapped onto the hull as an x-ray
	765
	766	Implemented at the author's request. The exterior view paints the rendered inner
	767	surface onto the outside of the cylinder, alongside the place markers.
	768
	769	It is a projection, and the code says so. From outside, Tower is structure; the
	770	texture shows what is inside. That makes it the frankest of the view's three
	771	reading aids — nobody could mistake a forest on the outer hull for a physical claim,
	772	where the fill lighting and the visible beam are subtler about it. It earns its
	773	place because the hull is opaque: this is the only view that shows the whole world
	774	at once.
	775
	776	The UVs are rewritten, not derived. Three's cylinder runs u around the
	777	circumference and v along the axis; the raster is the other way round, x = z and
	778	y = theta. The obvious approach — recovering the angle from each vertex position —
	779	is wrong, because the seam vertices are duplicated and `atan2` gives both the same
	780	angle, smearing one column of triangles across the entire image. Remapping the
	781	original uv attribute stays linear and lets RepeatWrapping absorb the offset.
	782
	783	Alignment was verified, not eyeballed. The raster was sampled at each place's
	784	computed position and the colour classified: four of four mountains landed on rock,
	785	both cap seas on water, the desert on desert. Judging this by eye would have been
	786	hopeless — the whole map is bands of green and tan, and a half-world error looks
	787	much like a correct one from most angles.
	788
	789	Two fixes the terrain forced. Small bright dots read fine against a bare hull
	790	and vanish into forest and desert, so markers gained a dark outline shell; and they
	791	are now scaled per frame to a roughly constant screen size, because 4 km is
	792	sub-pixel at the 1,200 km the default view sits at. The terrain also needed an
	793	emissive map: sunlight runs along the axis, so most of the hull faces away from
	794	every light and the map would otherwise be legible only on a sliver.
	795
	796	Also fixed here: selecting a place from the list moved the panel but not the
	797	camera — only permalinks called `focusOn`. Both paths now do.
`3b520afd` 2026-09-07 Pavel Safronov	798
	799	---
	800
	801	## D-025 — An empty draft is not a draft
	802
	803	Fixed after the author hit it. The map showed one place where twenty-two were
	804	expected. The local draft was empty and, because any draft beat the seed, the
	805	published world had been shadowed since the first time the app was opened.
	806
	807	How it happened. `public/world.json` originally shipped as `{"places": []}`.
	808	Opening the app once in that window persisted an empty world to `localStorage` —
	809	`pagehide` flushes on close — and from then on `load()` saw a draft, took it, and
	810	never fetched the seed again. Twenty-two places were sitting in the served file, and
	811	nothing in the UI suggested anything was wrong: an empty world and a deliberately
	812	emptied world look identical.
	813
	814	The fix. A stored world with no places is treated as no draft, so the seed
	815	loads. There is nothing to protect in an empty world, and the opposite failure — a
	816	deliberately emptied world refilling on reload — is visible and one click to undo,
	817	where this one was silent and permanent.
	818
	819	Plus an explicit way back. Reload published world in the Data section discards
	820	the draft and refetches `world.json`, for the case where a draft has drifted rather
	821	than emptied. Verified end to end: an empty draft now recovers 22 places on reload,
	822	and reseeding after a local edit returns 23 to 22.
	823
	824	The general shape is worth remembering. Local-first storage that silently wins
	825	over shipped content will, sooner or later, win with nothing in it. Precedence rules
	826	between a cache and its source need a defined answer for the empty case, and a
	827	manual override for every other case.
`04d6779e` 2026-09-07 Pavel Safronov	828
	829	---
	830
	831	## D-026 — Two grids that did not agree, both removed
	832
	833	Removed at the author's request. The exterior had structural ribs every 20 km;
	834	the map had a graticule every 50 km. Neither was wrong on its own, but side by side
	835	they invited a comparison that does not hold — twenty rings against ten lines,
	836	counting nothing in common.
	837
	838	Both are gone. Neither was load-bearing: the map keeps its scale bar and the
	839	exterior its spin readout, and with the terrain drawn the ground itself carries the
	840	sense of distance far better than either grid did. A reference grid that has to be
	841	mentally reconciled with another reference grid is worse than none.
	842
	843	Worth generalising: two independent scales shown in two views of the same object
	844	must agree or one must go. It is not enough for each to be individually correct.
	845
	846	---
	847
	848	## D-027 — A band is its own geometry, not a polygon
	849
	850	Fixed. Regions that ring the world — the desert, the forest, both cap seas — are
	851	now `{ type: 'band', zFrom, zTo }` rather than four-cornered polygons.
	852
	853	Found while removing the graticule. With the grid lines gone, a stray hairline
	854	was left across the middle of the map. A band written as a polygon runs theta from
	855	0 to 2pi, and both of those normalise to the same map row — so all four corners
	856	land on one line and the region collapses into a one-pixel streak. It had been
	857	hiding among the graticule lines the whole time.
	858
	859	Why a new type rather than a workaround. On a cylinder, "the whole circumference
	860	between two z" is a genuinely different shape from a polygon: it has no theta
	861	extent to give, because it has all of it. Tower's geography is built out of these —
	862	D-020 made the desert ring the world precisely because a band cannot be walked
	863	around — so it deserves to be a first-class shape rather than a polygon that happens
	864	to wrap. The map draws it as a full-height rectangle, hit-testing is a z range, and
	865	the editor reports it as rings the world, z 148.0 km to 218.0 km (70.0 km deep).
	866
	867	Tests assert the four ringing regions survive as bands through parse and public
	868	export, and that a band with unusable bounds is rejected like any other geometry.
`8b6287ba` 2026-09-07 Pavel Safronov	869
	870	---
	871
	872	## D-028 — The beam was inverted, and cones fringe
	873
	874	Found because the author asked what the brown cone and the white circle were.
	875	That question is its own answer: the two most important objects in the exterior view
	876	were unlabelled and one of them was drawn backwards.
	877
	878	The beam pointed the wrong way. `CylinderGeometry(radiusTop, radiusBottom)`
	879	combined with `rotateX(-pi/2)` — which sends +Y to −Z — put the first radius at
	880	the collector. Writing the pair in the order the eye reads them gave a cone wide at
	881	the sun cap and narrow at the mirror: light converging on the collector instead
	882	of into the aperture, with a 122 km mouth swallowing a 50 km hull. It had been that
	883	way since the view was built, and the flat brown wash it rendered as hid it.
	884
	885	The brown was a second bug. A uniform additive layer at 0.085 opacity over a
	886	black sky sums to about RGB(43,39,30) — a dull brown that reads as a solid object,
	887	which is the opposite of a beam.
	888
	889	Cones fringe, so the beam is rays now. Brightening it exposed bright dashes
	890	scattered across the collector's rim. Isolated by bisection: removing the cone
	891	cleared the disc, removing the stars did not, and shortening the cone moved the
	892	dashes with its mouth. At the mouth the cone's shell is edge-on, so its triangles
	893	project to slivers and the additive blend stacks them. Rays from the mirror to the
	894	aperture have no silhouette to alias, need no alpha ramp to fake convergence, and
	895	read more like light than a translucent solid ever did.
	896
	897	Everything is captioned now. Collector, beam, sun cap, far cap, as screen-space
	898	labels. If the person who commissioned the world cannot identify an object in it, a
	899	reader has no chance.
	900
	901	---
	902
	903	## D-029 — The view persists, which surfaced a boot-order bug
	904
	905	Implemented at the author's request. Map or exterior is remembered in
	906	`localStorage` and carried in the permalink hash, so a shared link can say *look at
	907	this from outside*. The hash wins; storage is the fallback for a bare URL.
	908
	909	And it immediately broke the exterior on load. Restoring that view un-hides the
	910	canvas and constructs the renderer in the same task, before the browser has laid the
	911	element out — so `resize()` measured a zero-sized box, bailed out, and the renderer
	912	kept its default 300×150 buffer stretched across a full-size canvas. Clicking into
	913	the view had always worked because layout had long since settled by then; only
	914	restoring it on boot hit the window.
	915
	916	Fixed with a `ResizeObserver` on the canvas rather than another guess about timing,
	917	which also covers the sidebar and window resizing for free.
`15d933c9` 2026-09-07 Pavel Safronov	918
	919	---
	920
	921	## D-030 — The camera persists, and the caps are places
	922
	923	Both at the author's request.
	924
	925	Camera. Position and target are stored per browser and restored on load, so a
	926	reload returns to the vantage you left rather than the default framing. Every field
	927	is validated before use — three finite numbers each, and a distance inside the
	928	controls' own limits — because a corrupt entry here would leave the view pointing at
	929	nothing with no obvious way back, and silently-bad persisted state has already cost
	930	this project once (D-025). Saves are debounced: OrbitControls fires `change` every
	931	frame of a drag.
	932
`ada068bc` 2026-09-09 Pavel Safronov	933	Amended: a heading is an angle, so it validates by wrapping, not by range. The
	934	interior view persists the same way, and its saved heading was range-checked like a
	935	distance — rejected outside ±100 radians, which the panning loop reached after about
	936	sixteen turns in one direction, at which point a reload silently put the observer back
	937	facing deepward. Nothing was corrupt; the check was asking the wrong question. The
	938	heading is now wrapped into [0, 2pi) at every write, and on restore any finite value is
	939	accepted and wrapped, because an unwrapped heading from an older save is still a
	940	perfectly good direction to face.
	941
`15d933c9` 2026-09-07 Pavel Safronov	942	Caps. The sun cap and far cap are now features in `content/geography.json`, so
	943	they appear in the place list and can be focused like anywhere else. Two things had
	944	to be special-cased, both for the same underlying reason — *a cap is not a patch of
	945	ground*:
	946
	947	- Its marker belongs on the axis. Placed on the hull at theta = 0 like every
	948	other marker, it sat on the cap's rim and rode round the edge as Tower turned.
	949	A cap's location is its centre.
	950	- Its framing is different. The usual from-the-side shot shows an end cap edge-on
	951	as a line. Standing off along the axis instead is worse — it looks down the barrel
	952	and foreshortens all 540 km behind an unlit disc. What works is the default view's
	953	oblique angle slid onto the cap, and a light from deepward so the far cap's outer
	954	face is not a black hole exactly when you focus on it.
	955
	956	A note on the debugging. Verifying camera persistence took several wrong turns
	957	because the test kept invalidating itself: `location.reload()` fires `pagehide`,
	958	which saved the live camera over the value the test had just planted. The
	959	persistence was working the whole time. Worth remembering that a `pagehide` save
	960	makes "set state, reload, check state" an invalid experiment.
`0e0c2dba` 2026-09-07 Pavel Safronov	961
	962	---
	963
	964	## D-031 — Sidebar: an accordion, and a label that stopped lying
	965
	966	**Five UI corrections from the author, all of them about the panel saying the wrong
	967	thing or being in the wrong place.**
	968
	969	"Canon" was a section heading for one slider. It read Canon → Star slats — 0,
	970	which framed an exploration control as if the whole section were canon. The section
	971	is now Star slats, its value is a Count, and canon is a chip on the
	972	summary that appears only when the count is 0 — because that is the one value canon
	973	actually asserts (D-017). A label should describe what it labels; "canon" is a
	974	property of a value here, not a category of setting.
	975
	976	Order follows scope. View moved to the top: it changes the whole app, not one
	977	panel. About moved to the bottom, where the least-used thing belongs.
	978
`5e405272` 2026-09-07 Pavel Safronov	979	Foldable sections. Seven panels in a fixed column pushed the useful ones
	980	off-screen. They are `<details>` elements now — native disclosure, keyboard and
	981	screen-reader behaviour for free. *This first shipped as an accordion, one open at a
	982	time, with the Selected panel opening itself on selection; both were removed — see
	983	D-032.*
`0e0c2dba` 2026-09-07 Pavel Safronov	984
	985	Mode gating moved from a CSS class to the `hidden` property, since `<details>`
	986	already carries open/closed state and stacking a `.hidden` class on top of it was a
	987	second, conflicting notion of visibility.
	988
	989	And a claim removed. The About text said the canon was "worked out in the open",
	990	with links to design.md and decisions.md. Those documents are the author's working
	991	notes on an unpublished novel, not a public invitation, and the links went to a repo
	992	a reader may not have. Now it says what is true and no more: *Tower is the setting
	993	of a novel. This map is its companion piece.*
`5e405272` 2026-09-07 Pavel Safronov	994
	995	---
	996
	997	## D-032 — No automatic opening or closing
	998
	999	Reverted at the author's request: the accordion was annoying, and nothing should
	1000	open or close by itself.
	1001
	1002	The two behaviours were one mistake. Exclusive opening meant that clicking a
	1003	place could collapse the panel about to describe it, so the Selected panel had to
	1004	open itself to compensate. Removing the accordion removed the need for the
	1005	compensation: sections the author opens stay open, so a selection simply appears in
	1006	a panel that is already there. A workaround that only exists to undo another
	1007	behaviour is a sign the first behaviour is wrong.
	1008
	1009	What remains is state, not behaviour. View and World start open, and Selected
	1010	starts open but stays hidden until something is selected — so the first selection
	1011	shows without anything having moved. After that the panels are wherever the author
	1012	put them: closing the editor keeps it closed through every later selection.
	1013
	1014	The general rule this is worth remembering by: an interface that rearranges
	1015	itself in response to a click costs the user their place. Prefer a sensible initial
	1016	state and then hold still.
`cf37225c` 2026-09-07 Pavel Safronov	1017
	1018	---
	1019
	1020	## D-033 — The slat control is disabled and its UI removed
	1021
	1022	Removed at the author's request: changing the count did not fully work. It did
	1023	not, and for two independent reasons, both worth writing down because this is
	1024	deferred rather than abandoned.
	1025
	1026	The raster has no slats in it. `terrain-color.png` is generated from
	1027	`content/geography.json` at canon's 0 slats. Raising the count drew slat bands as a
	1028	vector overlay on top of terrain that shows unbroken land underneath them — two
	1029	layers asserting different worlds. The corridor is authored the same way: Snowcrown,
	1030	the rivers and The Narrows all assume no slats, so bands above 0 cut straight
	1031	through them.
	1032
	1033	And the map's cut angle moves with the count. `mapThetaOrigin()` returns
	1034	`slatCentre(n-1)` when slats exist and π when they do not (D-013, D-024). The raster
	1035	is drawn against π, so every non-zero count rotates the terrain out of register with
	1036	its own place markers — by 120° at 3 slats, 157° at 8, up to 137 km of
	1037	displacement. This is the same desync D-024 fixed for the zero case, and the test
	1038	added there only guards the zero case.
	1039
	1040	What was done. The section is gone from the sidebar and `loadSlatCount()` is no
	1041	longer called, so a stored non-zero count from before cannot leak back in. What was
	1042	kept is everything that makes the count correct: `src/canon/slats.ts`, the
	1043	derivation, and the tests covering counts 1 to 16 — including that land stays
	1044	exactly Wisconsin's 140,268 km² across the whole range, and that the map cut lands
	1045	on a slat centre at every count.
	1046
	1047	What re-enabling would need. The terrain tool would have to bake slats into the
	1048	raster from the same count the app uses, which means the count becomes an input to
	1049	`npm run terrain` rather than a runtime toggle — and the geography would need
	1050	re-authoring around wherever the bands fall. That is a content decision as much as a
	1051	code one, which is the real reason it is parked.
`28f9a6bc` 2026-09-07 Pavel Safronov	1052
	1053	---
	1054
	1055	## D-034 — The interior view: real geometry, and haze doing the work
	1056
	1057	Phase 5, and the piece design.md §2.4 said would carry the illusion. The scope held:
	1058	real geometry, simple ground. The shell is the actual cylinder, displaced by the
	1059	height raster and painted with the colour raster; there is no scenery, no buildings
	1060	and no vegetation.
	1061
	1062	Nothing in the scene rotates, and that is the point. The observer, the ground and
	1063	the air all co-rotate, so from inside a sealed hull the spin is genuinely invisible.
	1064	No stars wheel overhead because there are no stars (§1.4). What is left is
	1065	`src/canon/motion.ts`, and it says something the design did not anticipate:
	1066
	1067	\| \| lean, antispinward \|
	1068	\|---\|---\|
	1069	\| rain at terminal velocity \| 1.5° — fixed, whatever the cloud height \|
	1070	\| a 1 km waterfall, at the bottom \| 11.3°, and 133 m of drift \|
	1071
	1072	The waterfall figure is the velocity at the bottom, which is the angle a streak of
	1073	falling water is drawn at. §1.9.2 tabulates 7.6° for the same drop, which is the
	1074	tilt of the path — where it lands against how far it fell. Both are right and they
	1075	differ by exactly half again in tangent (`freeFallPathLeanRad` alongside
	1076	`freeFallLeanRad`, pinned in test/motion.test.ts), and noticing they were not the
	1077	same number is the only reason this is written down rather than silently
	1078	contradicting the canon.
	1079
	1080	Rain barely leans, and it leans by a constant angle because drag holds the fall
	1081	speed constant. Only free fall keeps accelerating, so only free fall keeps leaning.
	1082	Design.md §2.4 offered "slanted rain and a leaning waterfall" as a pair; they are not
	1083	a pair. The waterfall is the compass. The rain is drawn, and the readout prints
	1084	1.5° beside it so the smallness reads as the finding it is rather than as a bug.
	1085
	1086	Haze is integrated per fragment, 24 midpoint samples through the rotational
	1087	density profile, and it reproduces the §1.8 table from the other direction: the
	1088	readout prints 78.1% straight up against 31.3% for the same 100 km along the ground.
	1089	The extinction is carried per channel (Rayleigh goes as λ⁻⁴, so red survives twice
	1090	as far as green and blue half as far), which is what makes distance blue rather than
	1091	merely grey.
	1092
	1093	The airlight is neutral, deliberately. Tinting it sky blue is the obvious move
	1094	and it is wrong: blue already scatters 2.4× harder in the per-channel extinction, so
	1095	a blue source colours the haze twice and every distant thing turns mauve regardless
	1096	of what it is made of — desert included, which is how the bug was spotted. With a
	1097	neutral source the blue falls out of the physics and the far side stays recognisably
	1098	forest and desert at the 78% canon promises.
	1099
	1100	## D-035 — Standing on the ground requires the ground on the CPU
	1101
	1102	The shell is displaced in the vertex shader, which is fine until somebody has to
	1103	stand on it. The first build put the camera 1.7 m above the undisplaced cylinder —
	1104	underground everywhere the land is higher than 1.7 m, which is everywhere. From
	1105	inside the rock the ground is invisible: its underside is a front face and the shell
	1106	is drawn back-side only. The view showed sky below the horizon and nothing else, and
	1107	at 366 m of eye height it looked perfect, which is what made it puzzling.
	1108
	1109	There is no way to ask the GPU where the ground is, so `terrain-height.png` is now
	1110	read back through a canvas and sampled here. Two mismatches had to be handled:
	1111
	1112	- The mesh is coarser than the raster. Cells are 0.7 km, texels are 0.26 km, so
	1113	the surface actually drawn is the raster sampled at the cell corners and stretched
	1114	flat between them. The lookup interpolates the same four corners rather than
	1115	sampling the raster directly, which would put the observer's feet tens of metres
	1116	off in rough country.
	1117	- Underground is much worse than hovering. Half the spread across the cell is
	1118	added as clearance: nil on the plains, and on a ridge it lifts you clear of the
	1119	facet instead of dropping you inside it.
	1120
	1121	The readback is 8-bit — browsers decode 16-bit PNGs down to 8 — so elevations
	1122	quantise to 19.5 m steps. Both the shader and this lookup read the same quantised
	1123	data, so they agree with each other, which is what matters for not being buried.
	1124
	1125	## D-036 — A custom ShaderMaterial writes linear colour into an sRGB canvas
	1126
	1127	Three inserts its colour-space handling into its own materials and never into a
	1128	custom one, and the first build silently got both halves wrong: the colour raster
	1129	was sampled as if it were linear, and linear output was written straight to the
	1130	canvas. Everything read dark and oversaturated, and the far side of the world came
	1131	out navy where it should have been hazy green.
	1132
	1133	The fix is both halves explicitly — decode the raster with a `pow(c, 2.2)` and end
	1134	each fragment shader with `#include <colorspace_fragment>`. The trap on the way:
	1135	adding `#include <colorspace_pars_fragment>` too fails to link, because Three
	1136	already puts that chunk in the prefix it builds for every material in order to
	1137	define `linearToOutputTexel`. The duplicate definitions produced no console error at
	1138	all — only `INVALID_OPERATION: no valid shader program in use` once per frame, and a
	1139	window filled with clear colour.
	1140
	1141	## D-037 — A shared `#view=` link was erased before it was read
	1142
	1143	Found while opening the interior view by URL. `App`'s constructor rendered before it
	1144	decided which view to open, and `render()` rewrites the hash from current state — so
	1145	`#view=interior` was replaced by an empty hash a few lines before `initialView()`
	1146	went looking for it. Every shared link that named a view had been landing on the map
	1147	since D-029, and it went unnoticed because the last-used view is also persisted in
	1148	`localStorage`, which restored the right view for whoever had visited before. The
	1149	view is now decided before the first render.
	1150
	1151	## D-038 — Selecting a place walks you there
	1152
	1153	Turning to face a place was the wrong default, and standing in the view for a minute
	1154	makes it obvious: at ground level the haze horizon closes at 100–150 km (§1.8), so
	1155	"face the Deepward Sea" from the capital points at 270 km of nothing. Selecting a
	1156	place now moves the observer to it, which is what "focus" already meant in the
	1157	other two views — the map centres, the exterior swings round, the interior walks.
	1158
	1159	Places with extent also get a heading: stand at the head of a river and look down
	1160	it, taking the centroid of the traced points as the aim. A point has no direction of
	1161	its own, so arriving is the whole answer and the heading is left alone. The separate
	1162	"Stand at the selected place" button is gone, since selection now does exactly that.
	1163
	1164	## D-039 — The height raster gets a declared working range
	1165
	1166	`terrain-height.png` was normalised 0..5,000 m against a fixed `maxElevationM`, and
	1167	both halves of that were wrong.
	1168
	1169	The floor mattered more than the ceiling. Clamping at zero flattened everything
	1170	below the datum, so every sea and lake floor was a plane — the Deepward Sea read as
	1171	0 m where it is actually −320 m. `content/geography.json` now declares
	1172	`minHeightM` / `maxHeightM` and the raster is stretched across that range.
	1173
	1174	The range is declared, not fitted. Fitting it to the terrain would silently
	1175	change what every grey level means whenever the geography is edited, which is the
	1176	kind of drift D-024 came from. Instead the generator prints the range it found
	1177	against the range declared, warns and clamps if the terrain has outgrown it, and
	1178	test/raster.test.ts asserts the declared range still contains the terrain.
	1179
	1180	The precision win is smaller than it looks, and worth stating plainly: the
	1181	terrain spans −508..4,392 m, so tightening the range to −520..4,400 m moves the step
	1182	from 19.6 m to 19.3 m. The quantisation is not set by the range at all — it is set
	1183	by browsers decoding a 16-bit PNG down to 8 bits, which leaves 256 levels however
	1184	wide the span. The fix for that is a second byte in a second channel, which is a
	1185	change to the encoder and the shader, and is not made here.
	1186
	1187	## D-040 — The axial sightline names the distance it measured
	1188
	1189	The readout compared "100 km up" against "100 km on" — except the along-axis probe
	1190	was clamped to the far cap, so standing 22 km from the end it measured 22 km and
	1191	still called it 100. It reported 77.5% where the honest figure for 100 km is 31%,
	1192	which is exactly backwards from the point the line exists to make. It now measures
	1193	whichever axial direction has the most room and prints the distance it used.
	1194
	1195	## D-041 — The caps get a pattern, and it is polar
	1196
`c954a775` 2026-09-28 Pavel Safronov	1197	Superseded by D-078 for the pattern: the caps' frames are identical triangles, at this pitch.
	1198
`da6bcb0e` 2026-09-08 Pavel Safronov	1199	*Half superseded by D-045: the far cap is now a window, and this pattern survives
	1200	on it as the window's mullions rather than as a filled checker. Everything below
	1201	still holds — the tessellation, the pitch, the mipmapping and the reason for all
	1202	three are unchanged, and the sun cap is still exactly this.*
	1203
`28f9a6bc` 2026-09-07 Pavel Safronov	1204	The end caps were flat tinted discs. They are the only surfaces in Tower that are
	1205	not landscape — 100 km of structure closing each end — and blank they read as fog
	1206	with an edge. A checker fixes that, and does something more useful besides: it is
	1207	the one thing in the world with a known cell size, so a glance at the far cap
	1208	tells you roughly how far away it is. That is worth having in a world where §1.8
	1209	says distance is otherwise unreadable.
	1210
	1211	Polar, not square. A cap is a disc on a spinning cylinder, so its natural
	1212	coordinates are radius and angle; a square grid meets the rim at every possible
	1213	angle and reads as something laid over the geometry rather than part of it. Cells
	1214	stay roughly square by doubling the sector count each time the radius doubles, which
	1215	leaves a seam at each doubling — which is what a segmented diaphragm would have
	1216	anyway. The alternative is slivers at the hub or slabs at the rim.
	1217
	1218	One texture, both views. Drawn once into a canvas and shared, so the exterior
	1219	and the interior cannot drift apart. Three maps UVs across the bounding box for both
	1220	`CircleGeometry` and `RingGeometry`, so the same image lands correctly on the far
	1221	cap's full disc and on the sun cap's annulus.
	1222
	1223	Two things had to be walked back after looking at it:
	1224
	1225	- It must be mipmapped. A high-contrast geometric pattern is the worst case for
	1226	point sampling. My first pass disabled mipmaps on the theory that resolving the
	1227	checker at distance was the whole point; minified, the far cap moiréd into a dark
	1228	cross-hatch that read as damage. Averaging to grey once the cells fall below a
	1229	pixel is both correct and honest — you cannot resolve them from 540 km either.
	1230	- Six rings, not ten. 5 km cells looked better face-on from inside and collapsed
	1231	from outside, where D-030 deliberately frames the caps near edge-on and
	1232	foreshortens them ten to one. A pattern has to survive its worst viewing angle.
	1233
	1234	The cap framing also moved to ~40° off the axis rather than ~17°, since there is now
	1235	something on the face worth turning to see, and the caps carry an emissive copy of
	1236	the pattern for the same reason the hull carries one of the terrain: sunlight runs
	1237	along the axis, so a cap is edge-on to every real light and renders as a dark disc
	1238	exactly when it is being looked at.
`fc11b064` 2026-09-08 Pavel Safronov	1239
`6364748a` 2026-09-08 Pavel Safronov	1240	## D-042 — Screenshots are taken through the UI, and diffed per shot
	1241
	1242	`npm run screenshots` drives the real application in a headless browser and
	1243	photographs it: nine views across two viewports, plus a contact sheet of all nine.
	1244
	1245	Through the UI, on purpose. Every shot clicks the view buttons, picks places out
	1246	of the list, and drags to orbit and to look up, rather than posing a renderer
	1247	directly. Posing is easier and it defeats the point. A screenshot taken by using the
	1248	application fails when the application fails, so a black canvas or a sidebar that
	1249	has swallowed the window ends up in the picture instead of hiding behind a passing
	1250	unit test. The tests already pin the canon numbers; nothing was checking that the
	1251	thing still draws.
	1252
	1253	The contact sheet is for a human. `overview.png` exists to be looked at once,
	1254	before committing — one image that answers "is it still there?". It is built by
	1255	laying the shots out as an HTML page and screenshotting that, because the browser is
	1256	already running and already knows how to lay images out; `tools/png.mjs` only
	1257	encodes, and this project has no reason to gain a PNG decoder.
	1258
	1259	Diffed per shot, not per sheet. The first version measured divergence on the
	1260	contact sheet, and that was the wrong unit: rewriting two of the nine panels
	1261	outright moved the whole-sheet mean by 0.6%, because a mean over nine panels on a
	1262	dark page dilutes everything. Per shot the same change reads in tens of percent.
	1263
	1264	The 3% threshold is measured, at both ends. Seven of the ten shots come out
	1265	byte-identical run to run — the rain is the only randomised thing in Tower and is
	1266	seeded on the way in, the exterior spin is switched off, and each shot gets a fresh
	1267	browser context so no camera leaks into the next. The two exterior shots are the
	1268	exception, drifting up to 0.98% because OrbitControls damps toward its target
	1269	asymptotically and lands a hair short of the same place each time. Against that,
	1270	hiding both 3D canvases scored 7–31%; a view going entirely black scores 12% to 39%.
	1271	3% is three times the worst noise and less than half the mildest real failure.
	1272
	1273	It is a smoke alarm, not a verdict: when it fires the answer is to look at the
	1274	sheet. And it only catches regressions — a view that never worked scores 0% when
	1275	it breaks further, which is exactly how mobile-exterior behaved during the sabotage
	1276	test that calibrated this, because at the time it was already a black slot.
	1277
	1278	Two browsers, one tour. The contact sheet gets its own browser instance. In the
	1279	one that has just taken the tour — nine WebGL contexts and two Three.js scenes —
	1280	`Page.captureScreenshot` on the sheet fails outright with "Unable to capture
	1281	screenshot", and the sheet is plain HTML and images, so the only thing wrong with it
	1282	is its neighbours.
	1283
	1284	What it found immediately. The mobile row: the sidebar was a fixed 340 px
	1285	against a 390 px viewport, so the world got about fifty pixels and both 3D views
	1286	rendered into a black slot. Fixed in D-043, which the same tool then verified —
	1287	the six desktop shots moved 0.00% and the three mobile ones moved 10–33%, which is
	1288	exactly the shape a layout fix should have.
	1289
`fc11b064` 2026-09-08 Pavel Safronov	1290	## D-043 — On a phone the sidebar is a drawer
	1291
	1292	The layout was one grid, `1fr 340px`, with no media query anywhere. On a 390 px
	1293	phone that is a 340 px panel and fifty pixels of world, and nothing to hide the
	1294	panel with — the application was, on a phone, a control surface for something you
	1295	could not see.
	1296
	1297	A drawer, not a stack. The other obvious move is to put the panel below the
	1298	world in a single column and let the page scroll. That is wrong here: the world is a
	1299	`position: absolute` canvas filling `main`, all three views are drag-to-look, and
	1300	`touch-action: none` means every gesture over the world belongs to the world. A
	1301	scrolling page and a canvas that eats touches are in permanent conflict. A drawer
	1302	keeps the canvas full-bleed and the gesture unambiguous.
	1303
	1304	Below 720 px, where the panel would otherwise be taking more than half the
`6364748a` 2026-09-08 Pavel Safronov	1305	window. Above it nothing changes at all — the desktop shots come out of the
	1306	screenshot tour byte-identical.
`fc11b064` 2026-09-08 Pavel Safronov	1307
	1308	Never the full width. `min(340px, 86vw)` leaves a strip of world showing, which
	1309	is what tells you the panel is over the map rather than a page you navigated to.
	1310	That strip is live canvas, though, and in writer mode a stray tap on it drops a
	1311	marker, so a scrim covers it and closes the drawer instead.
	1312
	1313	Two buttons, not one. The opener is fixed at the top right, opposite the view
	1314	switcher. The closer is in the panel header, in normal flow. A single fixed
	1315	toggle would have to float over the panel's own scrolling content once open, and
	1316	would occlude whatever scrolled under it; this way each button is only ever over the
	1317	thing it belongs to. The spin readout, which also lives at the top right, drops to
	1318	`top: 54px` to clear the opener.
	1319
	1320	Picking a place closes it. In every view selecting a place moves you: the map
	1321	centres, the exterior swings round, the interior walks there (D-038). On a phone the
	1322	list is the drawer, so staying open means watching none of it happen.
	1323
	1324	The closed drawer is `inert`. It is only translated off-screen, so without that
	1325	its fields stay in the tab order — typing into a panel nobody can see, scrolling it
	1326	under the world. It is set from the same media query the CSS uses, and crossing that
	1327	breakpoint (a phone turned on its side) resets to closed, which is also what clears
	1328	`inert` on the way back to the desktop column.
	1329
	1330	One thing came along with it. The standing readout is ~65 monospace characters
	1331	of aligned columns and overran 390 px at 11 px type. It shrinks to 9 px on the same
	1332	breakpoint rather than wrapping — wrapping those columns destroys the thing they are
	1333	for.
`de65ab10` 2026-09-08 Pavel Safronov	1334
	1335	## D-044 — Tower's wall is a kilometre thick
	1336
	1337	Canon had no figure for it. Both 3D views drew the hull, the caps and the collector
	1338	as zero-thickness surfaces, which is invisible until something is seen edge-on and
	1339	then reads as a paper cutout rather than a structure.
	1340
	1341	1 km, floor to outer skin — 2% of the radius. The alternatives were a 200 m
	1342	technical minimum, which is defensible on shielding and services and is sub-pixel at
	1343	every exterior zoom, and 5 km, which is unmistakable and a great deal of mass to
	1344	justify later. A kilometre reads as built without making Tower a pipe.
	1345
	1346	It is only visible at the openings, which is worth stating because it took a
	1347	false start to notice. A closed can has no visible thickness from outside — the
	1348	silhouette of a 50 km cylinder does not change when you give its wall a kilometre.
	1349	Depth shows only where there is a hole: the beam aperture, and now the far cap
	1350	window. So the caps sit at the outer face of the wall with a 1 km collar
	1351	standing between them and the inner surface, and it is the collar you see down.
	1352
	1353	The exterior camera had to be let closer. A kilometre is 1% of the cap's
	1354	diameter, so at the old 120 km minimum distance the collar was two pixels and the
	1355	near plane sat at 8 km — the depth was being drawn correctly and could not be got
	1356	close enough to to see. 60 km and a 2 km near plane make the choice legible; without
	1357	that change the constant would have been a number in a file and nothing else.
`392bb8a7` 2026-09-08 Pavel Safronov	1358
`769726ea` 2026-09-08 Pavel Safronov	1359	## D-045 — The far cap is a window, and the checker becomes its frame
	1360
`c954a775` 2026-09-28 Pavel Safronov	1361	Amended by D-078: the mullions are a triangle lattice, not the polar tessellation.
	1362
`769726ea` 2026-09-08 Pavel Safronov	1363	§1.4 has said since the first draft that the caps are the only views out of Tower
	1364	and that the sky through them turns rather than sweeps — a pinwheel about the axis
	1365	with one motionless star at the middle, one revolution every 7 min 29 s. It was
	1366	written and never built: §2.4 said flatly "no stars", and D-041 had meanwhile made
	1367	the far cap a checkered structural plate.
	1368
	1369	Glazed whole, not windows at its foot. §1.4 put the windows at the base of a
	1370	50 km cliff. A strip at the bottom of something 540 km away is nearly nothing to
	1371	look at, and the canon fact worth having in the view is the pinwheel, which needs
	1372	sky. So the whole far cap is a window. The sun cap is not — it faces the sun and
	1373	needs glare management rather than a view, and that asymmetry is already canon.
	1374
	1375	The checker survives as the mullions. Deleting it would have retired the only
	1376	object in Tower with a known cell size, which is the only rule anyone has for
	1377	judging distance (D-041, §1.8). Instead the same polar tessellation at the same
	1378	pitch is drawn as the frame around transparent panes: the cells are exactly where
	1379	they were, and you look through them rather than at them. The rule now reads against
	1380	the sky, which is a better place for it than against a plate.
	1381
	1382	The physics fell out of the shader that was already there. A pane is black with
	1383	stars in it, so the existing haze integration does the rest by itself: by day the
	1384	airlight term swamps everything and the cap is a disc of blue, which is what §1.5
	1385	already said would happen to anything 540 km off; at night the airlight collapses
	1386	with the beam and what is left is the sky. Nothing needed special-casing.
	1387
	1388	The stars are not scaled by the exposure, and that is the whole point. Nothing
	1389	Tower does to its own beam changes how bright Vega is. What does change is the eye,
	1390	so a separate gain carries that half: full at night and below, 2% at temperate. The
	1391	stars are there at noon and unseeable, which is a fact about eyes and not glass.
	1392
	1393	Two projections were wrong before the third. Gnomonic (xy/z) about the axis puts
	1394	the pole star at the origin and blows up as the ray goes perpendicular to the axis —
	1395	which is the ordinary grazing view from the foot of the cap, not an edge case.
	1396	Painting the field on the cap plane instead is stable everywhere and has no fixed
	1397	angular scale: from a kilometre away the entire 60° field lands inside one star
	1398	cell and the sky comes out empty, which is exactly what the first screenshot showed.
	1399	Azimuthal-equidistant has neither failure. It is defined for every direction but
	1400	exactly sunward, stars keep the same angular size however far off the window is —
	1401	which is what being at infinity means — and the pinwheel is not animated but falls
	1402	out of the projection, because azimuth plus spin phase is the rotation.
	1403
	1404	The pole star is placed rather than left to the hash. It is the fixed point the
	1405	whole phenomenon is read against, and a sky that happened not to roll one there
	1406	would be missing the only part of it anybody in Tower would have named.
	1407
`da6bcb0e` 2026-09-08 Pavel Safronov	1408	Settled by D-048: the surviving quarter of the beam does not go through the
	1409	window at all. It arrives collimated in a 3.2 km spot — a thousandth of the cap —
	1410	and that spot is a mirror, not glass.
`769726ea` 2026-09-08 Pavel Safronov	1411
`392bb8a7` 2026-09-08 Pavel Safronov	1412	## D-046 — Night is the beam banked, not the swarm aimed away
	1413
`59c12827` 2026-09-28 Pavel Safronov	1414	Still true under D-074, where night begins when the sun goes out at the far cap.
	1415
`392bb8a7` 2026-09-08 Pavel Safronov	1416	The obvious way to make night in a habitat lit by a steerable mirror swarm is to
	1417	aim the swarm off. It does not work here, and §1.3's own numbers say so.
	1418
	1419	The beam glows by itself: §1.3 gives the glow to the beam and the air rather
	1420	than to the diffusers, and says outright it "would be there if the beam were empty".
	1421	779 W/m²/sr over a strip 3.65° wide running horizon to horizon is about 0.2 sr, and
	1422	with the mean cosine of a strip through the zenith that is **~100 W/m² on the
	1423	floor** — half of temperate. A lit beam is a light source no matter where the
	1424	mirrorlets are pointed, so a dark world requires a dark beam. There is a test for
	1425	this figure in `test/canon.test.ts`, because the conclusion rests entirely on it.
	1426
	1427	So night is the beam banked to a thousandth of full, 0.32 W/m²: a rod you can
	1428	still find overhead and a world you can just make out under it. And **dark is the
	1429	beam shut down** — no airlight, no lit ground, no rod, and the only thing left in
	1430	the world is the sky through the far cap (D-045). The two settings are genuinely
	1431	different states rather than two points on a dimmer, which is what made both worth
	1432	having.
	1433
	1434	The beam's brightness scales with the exposure already, so it banks and goes out on
	1435	its own; neither setting needed the beam shader touched.
`a3f01bd7` 2026-09-08 Pavel Safronov	1436
	1437	## D-047 — The exterior hull reads two ways, and they are different claims
	1438
`ab6ae94f` 2026-09-28 Pavel Safronov	1439	Amended by D-076, D-077 and D-081: Rock, Piping, and the two together are on trial beside True and X-ray.
`908833aa` 2026-09-28 Pavel Safronov	1440
`a3f01bd7` 2026-09-08 Pavel Safronov	1441	The exterior had a "Show terrain" checkbox that wrapped the raster onto the hull.
	1442	It was doing something stronger than a checkbox admits: from outside Tower is
	1443	structure, so terrain on the outside is not what Tower looks like — it is a diagram
	1444	of what is inside, drawn on the outside because the wall is opaque. Useful, and not
	1445	a view. The two readings are different claims about the same object, so the control
	1446	is now a two-way switch that names both:
	1447
	1448	- True — the outer skin. Bare hull, checkered sun cap, glazed far cap.
	1449	- X-ray — the inner surface projected onto the outside.
	1450
	1451	The wall is drawn as two meshes now, not one double-sided one. Its faces show
	1452	different things: the outer skin changes with the mode and the inner surface is
	1453	always the land, and a single `DoubleSide` material has no way to say so. The inner
	1454	mesh is emissive for the same reason the x-ray always was — it is lit by a beam this
	1455	scene does not model, and unlit it reads as a hole rather than a world.
	1456
	1457	This is what made the far cap window worth having from outside. D-045 glazed it
	1458	for the sake of the interior, where the payoff is the pinwheel. From outside the
	1459	payoff is different and better: the panes are clear, so in true mode you look in
	1460	through the glass and the world is really there, mountains and desert and lakes
	1461	receding 540 km down the tube behind a grid of mullions. It is the only line of
	1462	sight into Tower that is not a projection, and it exists only because the wall got a
	1463	window and the hull got two faces in the same week.
	1464
	1465	The dark backing disc the first version of D-045 put behind the glazing is gone. It
	1466	was there to stop the panes showing the inside of the tube — which turns out to be
	1467	exactly the thing worth showing.
`da6bcb0e` 2026-09-08 Pavel Safronov	1468
	1469	## D-048 — The far cap's bullseye is a mirror
	1470
	1471	Decided by the author, from four options: mirror it, let it out, absorb it
	1472	behind fins, or grade the swarm so it never arrives.
	1473
	1474	The shape of the load is the argument. §1.3.1 leaves 24% of the beam
	1475	unscattered after 540 km, and it survives collimated — so this was never 13 TW
	1476	spread over a 7,854 km² cap. It is 1.7 MW/m² in a 7.96 km² spot, about 1,200
	1477	suns, and an absorber there sits at 2,330 K: survivable for tungsten, and for
	1478	no structure. But the spot is a thousandth of the cap's area, which means the
	1479	whole problem lives on a 3.2 km bullseye and 99.9% of the far cap needs nothing
	1480	done to it at all. That is what makes D-045's window and this coexist.
	1481
	1482	A mirror is the cheap thermal answer, not the expensive one. At 99.9% it
	1483	absorbs 1,665 W/m² — about 1.2 Earth solar constants, ordinary sunlight. The thing
	1484	that would have glowed white is merely warm. The absorber-with-fins option needed
	1485	the entire cap radiating at 400 K and 13 TW moved 50 km sideways from the spot to
	1486	get there; it was strictly worse than every alternative.
	1487
	1488	It also recovers the energy, and fixes a number that was already wrong. The
	1489	return trip makes the optical column 1,080 km rather than 540, so utilisation is
	1490	1 - e^(-2τ) = 94% against 76%. That matters more than it looks, because canon
	1491	had a contradiction sitting in it: §1.3 sets full brightness at 54.3 TW ÷ 169,646
	1492	km² = 320 W/m², which assumes the whole beam reaches the inner surface, while
	1493	§1.3.1's optical depth says only 76% does — implying 243 W/m². Both were
	1494	printed, and a test pinned 320. With the cavity closed at both ends (the mirror at
	1495	one, the collector at the other, returning what still escapes) essentially all
	1496	54.3 TW is eventually absorbed, and 320 is right as it stands. No test moved.
	1497
	1498	What it does not do is make weather. Aiming light at a region to warm it stays a
	1499	separate lever, and §1.9.4 already says that is how every climate in Tower is made.
	1500	The mirror is about not wasting the beam; the Dry Reach is about where the heat is
	1501	wanted. Keeping them separate means neither constrains the other.
	1502
	1503	A geometric trap, recorded because I fell into it. The mirror sits on the axis,
	1504	so it looks like it must occult the pinwheel's motionless centre — I wrote that into
	1505	canon before checking. It does not. The pinwheel turns about the direction
	1506	parallel to the axis, and that direction lands on the window at the observer's
	1507	own radius, not at the cap's middle. From the floor the spin axis is horizontal,
	1508	so the centre of Tower's only moving sky is a point on the horizon, dead ahead
	1509	at eye level — while the mirror is ~10° above it from 270 km. The two coincide only
	1510	for an observer standing on the axis, and nobody can stand there.
`e27aaf20` 2026-09-08 Pavel Safronov	1511
	1512	## D-049 — Rain is not drawn any more
	1513
	1514	Rain existed in the interior view for one reason: with the hull sealed it was the
	1515	only moving thing in the sky, and therefore the only evidence Tower turns. §1.9.2
	1516	then undercut it — drag pins a drop's speed, so rain leans a fixed 1.5° whatever
	1517	the cloud, against a waterfall's 11°. The readout printed the angle to stop the
	1518	smallness reading as a bug, which is an admission that the picture was not carrying
	1519	the claim.
	1520
	1521	D-045 gave the view a whole sky going round once every 7 min 29 s. Against that, a
	1522	1.5° lean is not evidence of anything, and 6,000 line segments were being stepped
	1523	every frame to fail to show it.
	1524
	1525	The physics stays. `canon/motion.ts` still computes the lean and
	1526	`test/motion.test.ts` still checks it — it is a real fact about how anything falls
	1527	in here, and the waterfall is still the compass. It is only the rendering that is
	1528	gone, along with the toggle, the seeded raindrop field, and the six Three imports
	1529	that existed for it. The readout line it occupied now prints the pinwheel's period,
	1530	which is the interval you can actually time against the sky.
`9c891d2e` 2026-09-08 Pavel Safronov	1531
	1532	## D-050 — The sun cap is a window too, and Tower never sees the sun
	1533
	1534	§1.4 was wrong about this cap, and the number that shows it was already in canon.
	1535	It demanded "heavy glare management" on the grounds that the sun cap faces the sun.
	1536	It does not face the sun. It faces the collector, which is 244 km across at 320
	1537	km — 41.8° wide against the sun's 0.53°, seventy-nine times its diameter. The
	1538	collector sits between Tower and the sun by construction, so it occults it entirely,
	1539	and the umbra is 120 km across where Tower's 51 km sits in it. There is no direct
	1540	sunlight on the sun cap and there never has been. Nothing to manage.
	1541
	1542	So this cap is glazed on the same terms as the far one, and what fills it is the
	1543	back of the collector: a black disc, because its front is a mirror and its back
	1544	radiates 9.6 TW of waste heat at about 245 K, which is infrared. Around the edge of
	1545	it, stars. It does not turn with them — the collector flies in formation rather than
	1546	spinning — so the sun cap window is a still black disc against a wheeling sky.
	1547
	1548	Nobody in Tower can see the sun. They already could not see Earth (§1.4, the
	1549	terminator orbit puts it out the side, which is solid hull). A habitat in orbit
	1550	around a planet it has never seen, lit by a star it has never seen.
	1551
	1552	The occultation is computed in the shader rather than modelled: the fragment casts
	1553	its own ray on to the collector's plane and asks whether it lands on the disc. That
	1554	is the same trick the far cap uses for its stars, and it means the interior does not
	1555	have to carry a 244 km mesh 320 km outside the world to draw one black circle.
	1556
	1557	The interior's checkered cap material is gone with this. Both caps are windows now,
	1558	and nothing inside was still using it — the checker survives as the exterior's sun
	1559	cap and as both windows' mullions.
`03607270` 2026-09-08 Pavel Safronov	1560
	1561	## D-051 — The range rings the world
	1562
	1563	The mountains were "a 140 km arc, not a ring" — four named domes covering about a
	1564	third of the circumference, with the other two thirds simply unauthored. Asked to
	1565	close it, and closing it is better than the arc was: a ring of peaks is a watershed
	1566	that divides the whole world rather than a wall you can walk around the end of.
	1567
	1568	The ridge op grows optional ridged noise, which is what turns a ridge into a
	1569	range. Without it a ring band is a smooth swell all the way round — right for the
	1570	three low walls across the waist, wrong for peaks.
	1571
	1572	It blends with `max` rather than adding, and that is not a detail. Added, The
	1573	Girdle's 2,900 m would stack onto Snowcrown's 4,300 and push the massif to 6,900 —
	1574	past the raster's 4,400 m ceiling, where it would clip flat. Taking the greater of
	1575	the two leaves the named summits as the high points of their own range, which is
	1576	what they are: Snowcrown is not a separate mountain from The Girdle, it is the tall
	1577	part of it.
	1578
	1579	Peaks reach about 3,900 m, so they cross the 3,000 m snowline and the ring reads as
	1580	snow the whole way round. The rivers still cut it, because carve runs after relief.
`9a621732` 2026-09-08 Pavel Safronov	1581
	1582	## D-052 — A fourth noise octave, and colour that mottles instead of dimming
	1583
	1584	Height. The base had three noise layers — 78 km, 12 km, 3.4 km — and a pixel is
	1585	264 m. Between the finest layer and the raster's ~530 m Nyquist limit there was an
	1586	octave and a half of nothing, so the ground read as smooth swells with no grain on
	1587	them at exactly the scale you look at it from inside. A fourth layer at 1.5 km and
	1588	±26 m fills that band. It goes in the base, deliberately: carve and flood run
	1589	afterwards, so rivers still cut through it and lakes keep flat surfaces rather than
	1590	inheriting a rippled floor.
	1591
	1592	Colour. There was already per-pixel variation, and it was the wrong kind — one
	1593	noise field scaling all three channels together, which dims and brightens without
	1594	ever changing the colour. A second, finer field now mottles: it pushes the
	1595	channels apart, green moving most and blue least, so ground reads as patchy
	1596	vegetation against bare soil rather than as one paint turned up and down. Keeping
	1597	blue nearly fixed is what stops the desert drifting toward the same green as the
	1598	forest.
	1599
	1600	It exposed a live bug, and the fix matters more than the noise does. The extra
	1601	relief pushed the terrain to -521 m against a declared floor of -520 m, so
	1602	every pixel down there was being silently flattened. The generator printed a warning
	1603	and carried on; nothing else noticed. That is D-024's failure mode exactly — a
	1604	clipped raster displaces terrain rather than breaking it, so nothing downstream
	1605	complains.
	1606
	1607	Two changes: the floor moves to -560 m, and `npm run terrain` now **exits
	1608	non-zero** when any pixel is clamped instead of printing a line into a log nobody
	1609	reads. The unit test could never have caught this — it compares the declaration
	1610	against extremes written down when they were last true, not against the terrain that
	1611	just came out — and it now says so where it sits.
`2bb3e0de` 2026-09-08 Pavel Safronov	1612
	1613	## D-053 — The day is a front, not a dimmer
	1614
`59c12827` 2026-09-28 Pavel Safronov	1615	Superseded by D-074: the day is a sun that crosses the world, not a front.
	1616
`2bb3e0de` 2026-09-08 Pavel Safronov	1617	The obvious day-night cycle is a global fade between the light settings. It would
	1618	have worked and it would have thrown away the most interesting thing about how this
	1619	world is lit.
	1620
	1621	What the physics allows. D-046 already established that the swarm cannot make
	1622	night, because §1.3 gives the beam's glow to the beam and the air rather than to the
	1623	diffusers — a lit beam lights the world whatever the mirrorlets are doing. So beam
	1624	power has to ramp, and when it does, the sky brightens everywhere at once. There
	1625	is no way to have one end of the axis get its morning before the other.
	1626
	1627	What the swarm can still do is choose where the light lands. The diffusers hand
	1628	beam light to the ground, and they are individually flown with absurd control
	1629	authority (§1.3.1). So the ground can be lit progressively even though the sky
	1630	cannot, and the natural direction is the one the beam already runs: sun cap to far
	1631	cap.
	1632
	1633	A Tower dawn is therefore a fully lit sky over dark ground, with a line of
	1634	daylight travelling toward you at 225 m/s. That is not a compromise between the two
	1635	options, it is what falls out of taking both constraints seriously — and it is a
	1636	better image than either.
	1637
	1638	It gives Tower longitude. §1.9.4 says the world has no latitude: every point is
	1639	equidistant from the light, which is why nothing about it varies. The cycle adds the
	1640	axis back as 40 minutes of time difference between the caps. Dusk runs the same
	1641	direction, so the sun-cap end has the shorter day by a hair.
	1642
	1643	The panel gets a time slider before it gets a clock. Running the cycle is the
	1644	feature, but a 24-hour day at any watchable rate means waiting for dawn, and dawn
	1645	lasts 40 world-minutes out of 1,440. Being able to scrub straight to 06:17 is what
	1646	makes the front usable at all — and the front is a pure function of the phase, so
	1647	scrubbing and ticking are the same operation.
	1648
	1649	The readout names the distance to the line rather than just the hour, because the
	1650	distance is the part you can act on: 27 km sunward is about two minutes' wait.
`49b0d64d` 2026-09-08 Pavel Safronov	1651
	1652	## D-054 — The collector gets a lit rim
	1653
	1654	Drawing the collector as pure black was right about its face and wrong about the
	1655	object. From inside the sun cap window at night you could not see it at all — only
	1656	that a piece of the sky was missing, which reads as a rendering hole rather than as
	1657	a 244 km structure.
	1658
	1659	The physics gives the rim for free. The collector is backlit by definition: its
	1660	mirror face is pointed at the sun and away from Tower. Anything holding 46,938 km²
	1661	flat has structure standing proud of the mirror plane at the edge, and at the edge
	1662	the sun strikes that structure side-on. Add the scatter every real mirror throws at
	1663	grazing incidence and the boundary is a thin sun-coloured ring. That is simply how
	1664	backlit things look, and it is the one part of the collector that can be lit.
	1665
	1666	What it is not is sunlight past the edge, and the distinction matters because the
	1667	first is a property of the object and the second would break canon. Tower sits wholly
	1668	inside the umbra — §1.3.0 — so the sun's disc is occulted from every point in the
	1669	world. Nobody sees a sliver of star; they see lit metal.
	1670
	1671	Rendered as two exponentials: a tight core at the declared 4 km of edge structure,
	1672	and a wide dim skirt for the scatter around it. One alone reads as a drawn outline
	1673	rather than as light. The first attempt was three times too bright and its skirt
	1674	seven times too wide, which washed the whole disc gold instead of tracing it.
	1675
	1676	It does not scale with the beam. The sun does not dim when Tower banks its own
	1677	lamp, so the rim is outside `uExposure`, like the stars. It does fade as the eye
	1678	adapts, but from a much higher floor than a star's — this is sunlit structure, and
	1679	it stays faintly visible at noon where a star cannot.
`472390e2` 2026-09-09 Pavel Safronov	1680
	1681	## D-055 — The collector looks the same from outside as it does from inside
	1682
	1683	The exterior view drew the collector as a near-white disc with a pale blue glow, and
	1684	the sun cap window drew it as black with a lit rim (D-054). Both cannot be right, and
	1685	the window is the one that follows from the physics: the collector's mirror face is
	1686	pointed at the sun and away from Tower, so what the world sees is the back of it, a
	1687	46,938 km² radiator running at 245 K. That is infrared. In visible light it is black.
	1688
	1689	So the exterior draws two faces rather than one. The disc is two meshes sharing a
	1690	plane, one front-facing and one back-facing, and the camera's own position picks which
	1691	one it gets — culling does the choosing, so they never fight over the same depth.
	1692
	1693	- The Tower side is very dark and unlit, with just enough emissive to keep it a
	1694	surface. Not zero: against black sky, a wholly unlit disc is not a dark object, it
	1695	is a missing one.
	1696	- The sun side is bare metal at almost no roughness, which under one directional light
	1697	is dark everywhere except a specular glint. That is what a flat mirror in vacuum
	1698	actually looks like: it shows the empty sky in front of it, not the sun behind you.
	1699
	1700	The rim is what makes it an object. The 4 km of sunlit edge structure from D-054 is
	1701	drawn here as a ring of geometry rather than as two exponentials in a shader, because
	1702	out here it is being seen rather than looked through — and it is placed just outboard
	1703	of the mirror, not overlapping its edge, so the two surfaces never share a depth
	1704	either. It reads from both sides, which is right: it is structure, not a reflection.
	1705
	1706	The cost is that the largest object in the scene is now a dark one. The label, the
	1707	beam converging on it and the lit rim carry it — and being hard to see is the point,
	1708	since Tower has never seen this thing lit.
`c48030fb` 2026-09-09 Pavel Safronov	1709
	1710	## D-056 — Stars are drawn a pixel wide, so the sky stops flickering
	1711
`e331ec2a` 2026-09-28 Pavel Safronov	1712	Superseded by D-075: the stars are now the real catalogue; the pixel-wide gaussian stays.
	1713
`c48030fb` 2026-09-09 Pavel Safronov	1714	The star field was a spike: brightness `(1 - d/r)^4` over a radius of about 0.11°,
	1715	which on a retina panel at the default field is a core roughly one pixel across. The
	1716	sky is sampled once per pixel, at the pixel's centre. So what got drawn was that spike
	1717	read at one arbitrary point along it, and half a pixel of head movement moved the
	1718	reading from the peak to a third of it. Every star did that independently, every
	1719	frame, and the whole sky boiled whenever anyone looked around.
	1720
	1721	That is wrong twice over. Vacuum has no air to twinkle through — the sun cap
	1722	window looks at deep space and the far cap at the anti-sun sky, and neither has an
	1723	atmosphere in front of it. And the interior view's whole claim is that nothing in
	1724	Tower moves except the pinwheel: a sky that seethes contradicts the one thing §1.4
	1725	promises about it.
	1726
	1727	The star becomes a gaussian one pixel wide. Sigma is tied to the pixel footprint,
	1728	so a star is one pixel on any display and at any zoom, and a curve that smooth barely
	1729	changes between neighbouring samples. This is not a compromise with the physics; it is
	1730	closer to it. No optic renders a point source as a point — it renders it as a blob the
	1731	width of its own resolution, and here the resolution is the pixel.
	1732
	1733	The footprint is measured from the view direction rather than from the projected star
	1734	coordinates, because those carry the branch cut of an `atan` and a derivative across
	1735	that seam is a line of garbage down the middle of the sky.
	1736
	1737	**Measured, facing sunward from The Capital at night on a 2x display, turning one
	1738	pixel at a time over sixteen frames:** the light in star-like points varied by 8.1% of
	1739	its mean before and 5.0% after, and the count of them by 5.2% before and 3.9% after.
	1740	About a quarter more stars are visible at any moment — not because the field got
	1741	denser, but because a star no longer has to land near a pixel centre to be seen at its
	1742	true brightness. The variation that remains is mostly stars genuinely entering and
	1743	leaving the frame as the view turns, which is not flicker; per star, the worst case
	1744	went from a factor of three between neighbouring samples to about 16%.
`ada068bc` 2026-09-09 Pavel Safronov	1745
	1746	## D-057 — The height raster carries no mipmaps, because the vertex stage cannot choose one
	1747
	1748	Firefox on Apple Silicon rendered the interior wrongly, intermittently: the terrain
	1749	lost its colour to a flat blue-grey wash and its relief to a coarse grid of creases,
	1750	and it came back on a hard reload. It followed where the view was pointed rather than
	1751	where the observer stood — changing eye height did not clear it — which is what made
	1752	it read as a bug about rotation rather than about a texture.
	1753
	1754	The cause is a texture fetch in a vertex shader that never named its mip level.
	1755	`elevation()` in the ground shader reads the height raster once per cell corner, in
	1756	the vertex stage. A vertex shader has no derivatives, so there is nothing for an
	1757	implicit `texture2D` to compute a level of detail from — the level is whatever the
	1758	driver decides, and the driver is entitled to decide anything. The raster was loaded
	1759	with Three's defaults, which means generated mipmaps, and with the card's maximum
	1760	anisotropy on top: a sampler with eleven levels and an ambiguous rule for picking one.
	1761	On a driver that picks badly the shell is displaced by a downsampled height field,
	1762	which flattens the terrain into blocks and shows the block seams as creases, and the
	1763	colour raster read at the same wrong scale averages into the flat wash. Every symptom
	1764	follows, including the reload: a fresh context re-uploads and re-mips the textures.
	1765
	1766	The ambiguity is removed at the sampler rather than at the fetch. The obvious
	1767	guard is to name the level in the shader, and it is not available here — the
	1768	vertex-only `texture2DLod` does not survive Three's WebGL2 shader prefix, and asking
	1769	for it fails the program link outright, which trades an intermittent bug for a
	1770	permanent black world. So the height raster is built with `generateMipmaps` off,
	1771	`LinearFilter` at both ends and anisotropy 1: one level, nothing to choose between.
	1772
	1773	This costs nothing. The height raster is read at a fixed rate — one texel per 0.7 km
	1774	cell corner — that has no relationship to how the cell lands on screen, so mipmaps and
	1775	anisotropy were never doing anything for it but occupying a third more texture memory.
	1776	The colour raster keeps both, and should: it is read in the fragment shader, at
	1777	every grazing angle the inside of a cylinder can offer, which is exactly what
	1778	anisotropic filtering is for.
	1779
	1780	Honest about the evidence. The failure was not reproduced on demand — four hundred
	1781	scripted pan steps in Firefox on the same machine rendered correctly throughout, with
	1782	no WebGL warnings and no context loss. The diagnosis is fitted to the artefact: the
	1783	reported view was reproduced exactly from its saved state, and the difference between
	1784	what it should show and what was photographed is specifically the two rasters, not the
	1785	camera, the depth buffer or the geometry. If it recurs, the next thing to drop is the
	1786	anisotropy on the colour raster, which would move the suspicion from the vertex stage
	1787	to the sampler state as a whole.
`d975fc10` 2026-09-09 Pavel Safronov	1788
	1789	## D-058 — The interior keeps a flight recorder
	1790
	1791	D-057 was diagnosed from a photograph. That worked, and it should not have had to:
	1792	the fault was intermittent, it was reproduced by the author and never once by the
	1793	tooling, and every question about it — was the geometry there, was the camera sane,
	1794	were the textures still live — had to be answered by inference from a screenshot
	1795	rather than by reading the number. So the interior now records what it is doing.
	1796
	1797	A ring of the last nine hundred frames, about a quarter of a minute: camera,
	1798	heading, field, where the observer stands, draw calls, triangle count, and every
	1799	fifteenth frame a readback of a 64-pixel block from the middle of the frame. The
	1800	readback has to happen immediately after `render`, while the drawing buffer is still
	1801	the current target; a first attempt read it from a 2D canvas a moment later and got
	1802	nothing but zeroes, which looked for a while like a finding of its own.
	1803
	1804	`towerDump()` on the console prints the lot and copies it to the clipboard. It is
	1805	present in the built site, not only in dev, because this fault has never once appeared
	1806	while anybody was looking for it. It is two properties on `window` that do nothing
	1807	until they are called.
	1808
	1809	The line that actually answers the question draws each surface alone. A flat,
	1810	colourless frame can be the ground shader losing its raster, or the ground missing
	1811	altogether with a cap or the bare clear colour behind it — indistinguishable on
	1812	screen, entirely different faults. Rendering the ground, the two caps, the bore and
	1813	the beam one at a time and measuring each separates them in a single dump. The clear
	1814	colour reads 19 in chroma, so 19 means that surface drew nothing here; the ground
	1815	reads about 30 when it is carrying land. Verified by taking the colour raster away
	1816	from the ground shader mid-flight: ground fell from 31.3 to 19.5 while the far cap
	1817	stayed at 31.3, which is exactly the signature to look for.
	1818
	1819	There is deliberately no automatic alarm, and that is a result rather than a gap.
	1820	The first version warned when the middle of the frame lost its colour, on the theory
	1821	that the fault paints a neutral wash. It does not. It paints the haze tint, and the
	1822	haze tint is itself a blue-grey measuring 19 — while a legitimate view of the far side
	1823	through 200 km of air measures about 20, against about 30 for land underfoot. Any
	1824	threshold that caught the fault would have fired every time anyone looked at the
	1825	horizon. The recorder therefore records and the human decides; a quarter-minute of
	1826	history is long enough that noticing the fault, reaching for the console and typing
	1827	still catches the frames on both sides of the change.
`8cb6755f` 2026-09-09 Pavel Safronov	1828
	1829	## D-059 — Inside, the finger count is the mode
	1830
`3e898b51` 2026-09-28 Pavel Safronov	1831	Amended by D-083: the pinch no longer zooms and the wheel moves rather than zooms; the field of view is fixed at 60°.
	1832
`8cb6755f` 2026-09-09 Pavel Safronov	1833	Until now the interior could not be moved through at all. Dragging looked around,
	1834	the wheel zoomed, and the only way to change where you stood was to select a place
	1835	and be walked there (D-038). That is fine for the places that exist and useless for
	1836	the ground between them, and on a phone it left the whole view as something to be
	1837	turned on the spot.
	1838
	1839	**The gesture had to carry the mode, because a switch would have to live on the
	1840	world.** The alternatives were considered and dropped. A Look/Move toggle in the
	1841	corner is the CAD convention and it is honest, but it is a second pill bar over a
	1842	canvas that is already carrying the view switcher, the standing readout and the
	1843	place captions, and on a phone that is the scarcest surface in the application
	1844	(D-043). A virtual joystick is the mobile-FPS convention and reads as a game rather
	1845	than a map. So: one pointer looks, two move. The count is the mode, nothing is
	1846	drawn, and the gesture is the one every map on the device already uses.
	1847
	1848	The cost is discoverability, and it is paid in the panel. A gesture with no
	1849	chrome is a gesture nobody finds, so the interior hint now leads with it in bold
	1850	rather than mentioning it in passing. This is the known trade and it was taken
	1851	deliberately.
	1852
	1853	A mouse has only ever one pointer, so shift-drag or the middle button stands in
	1854	for the second finger. Without that the feature would exist only on touchscreens,
	1855	which is not what was wrong.
	1856
	1857	Moving means grabbing the ground, not steering. The ray through the pointer is
	1858	met with the ground plane before the move and after it, and the observer is shifted
	1859	by the difference — so the patch of land under the finger stays under the finger.
	1860	The point of doing it this way is that no speed has to be chosen: the distance
	1861	covered falls out of the eye height and the field of view, and the same drag is a
	1862	few metres of pavement at 1.7 m and a hundred kilometres of plain from 10 km up.
	1863	Tower is 540 km deep and 314 km around, and no fixed rate is right at both ends of
	1864	that.
	1865
	1866	Two caps keep it from flinging. A ray at the horizon meets the ground at
	1867	infinity and one above it never does, so the grab distance is capped at 1,200 eye
	1868	heights and at 60 km outright. Both numbers were measured rather than guessed, by
	1869	dragging the running view a screen at a time at five eye heights: a screen of level
	1870	drag carries 1.9 km at a standing 1.7 m, and 56 km from 100 m up, where the ceiling
	1871	has taken over. The ceiling sits above the honest grab distance from the highest
	1872	the eye goes — 25 km up, tilted well down, meets the ground about 60 km out — so it
	1873	only ever bites near the skyline and never lags the finger on a view aimed at the
	1874	ground. A first pass capped at 150 eye heights and a screen of drag moved four
	1875	metres, which is how the numbers came to be measured.
	1876
	1877	The tilt is the gearing, and it was not arranged. Looking level grabs the far
	1878	ground and covers kilometres; looking down grabs what is underfoot and covers
	1879	metres — 26 m to the screen at standing height, tilted 23° down. That falls out of
	1880	grabbing the ground rather than being a control, which is the argument for the
	1881	whole approach: no speed had to be chosen and none can be got wrong.
	1882
	1883	The grab plane is flat, deliberately. Meeting the ray with the terrain instead
	1884	would make the grab point jump as the ray crossed a ridge half a screen away, and
	1885	throw the observer with it. The plane through the observer's feet keeps the drag
	1886	smooth, and the height they actually stand at is resolved afterwards, as it always
	1887	was.
	1888
	1889	Both readings of the pointer set are taken over the same set. The centroid is
	1890	sampled before the moving pointer is folded in and after, never across a finger
	1891	arriving or leaving — otherwise the second finger touching down would jerk the
	1892	world by half the distance between the two.
`6bb061a6` 2026-09-14 Pavel Safronov	1893
	1894	## D-060 — The exterior looks in at the interior's own world
	1895
	1896	Implemented at the author's request, three changes at once.
	1897
	1898	The place markers are gone from the exterior. They were coloured spheres on the
	1899	hull, and they had become a crutch the view no longer needs: the sidebar lists the
	1900	places, the map shows them, and selecting one still swings the exterior camera round
	1901	to it (D-038). With the markers went their outline shells and their per-frame
	1902	scaling. A cap is still framed from 40° off its axis, which now also means looking
	1903	in through its window.
	1904
	1905	Looking in from outside now shows the interior view's world, not a sketch of it.
	1906	Before this the exterior had its own inside: an emissive terrain texture on the back
	1907	of the hull, no haze, and a checkered sun cap that was opaque. The interior had
	1908	glazed that cap long ago (D-050), so the two views disagreed about the one surface
	1909	you could see from both. The fix is not to copy the interior's look but to use it:
	1910	the ground, both windows, the bore and the beam move out of `interior.ts` into
	1911	`interior-model.ts`, and both views add the same meshes with the same shaders.
	1912
	1913	Two things had to change for the shaders to be looked at from outside.
	1914
	1915	- The haze integral clips to the air. It used to integrate the whole segment from
	1916	eye to fragment, which is right when the eye is inside and absurd when it is 300 km
	1917	out in vacuum. The segment is now cut to the part inside the tube — radius out to
	1918	the deepest lake floor, z from the sun cap to the far cap's glass — before it is
	1919	sampled. From inside that is the whole segment, so the interior renders as before;
	1920	the only difference is the density is held at floor value rather than rising past
	1921	it below the datum.
	1922	- The windows are see-through from outside. Inside, a pane paints the computed
	1923	sky and is opaque to the renderer. From outside the same shader writes the pane
	1924	transparent, carrying only the airlight in front of it, and blends with whatever is
	1925	really behind: the land beyond the near cap, and the real collector and stars beyond
	1926	the far one down the tube. Mullions stay opaque and hazed. Alpha is taken on the
	1927	green channel, where canon quotes extinction.
	1928
	1929	The exterior turns its hull and the air turns with it, so the shaders work in the
	1930	tower's frame: `uFrame` carries world into tower coordinates and the exterior updates
	1931	it each frame after the spin. Inside, nothing turns and it is the identity.
	1932
	1933	The blue light on the collector was the fill lights, reflected. D-055 made the
	1934	collector's sun face polished metal under the scene's lights, expecting a single sun
	1935	glint. But the scene also carries two fill lights that exist only to make the hull
	1936	readable (§2.4), and a mirror reflects them as faithfully as it reflects the sun: at
	1937	any angle lined up with the bluish rim fill, a blue-white sweep crossed the disc.
	1938	Reproduced by aiming the camera along that light's reflection. A standard material
	1939	cannot be told to ignore some lights, so the face is now a small shader that knows
	1940	exactly one: the same GGX lobe Three drew for the sun at the same roughness, over the
	1941	same dark base, and nothing for the fills.
	1942
	1943	The consequence is worth writing down, because it surprised the author: **the bright
	1944	band that had been read as the sun's reflection was the fill light too**, white at
	1945	its core where it saturated. The sun is on the axis behind the collector, so its
	1946	real glint only appears when looking nearly straight down that axis from sunward.
	1947
	1948	The exterior's stars are the interior's. The old field was 3,500 points 26 km
	1949	wide at 18,000 km — about a pixel each, and on a retina display simply absent. The
	1950	sky is now a sphere around the scene painted with the windows' own star field
	1951	(D-056), sampled by direction from the camera so the stars are at infinity and hold
	1952	still while it orbits. The pinwheel phase is zero, since the camera is not riding the
	1953	hull, and the gain is full, since there is no daylight out here to adapt to. The
	1954	same pole star sits on the deepward axis in both views.
	1955
	1956	Showing the whole sky exposed that the field was never uniform. It was laid out
	1957	in an azimuthal-equidistant projection about the axis, which spends area as θ dθ dφ
	1958	where the sky has sin θ dθ dφ: stars per square degree grew as θ / sin θ, correct at
	1959	the deepward pole, half again by the equator and without bound sunward, where they
	1960	also shrank into grey dust. The windows had only ever shown narrow parts of it. The
	1961	author saw it at once from outside — sparse bright stars one way, a haze of faint
	1962	ones the other. Stars are now scattered in 3D instead: a cubic lattice of cells each
	1963	rolls a star, and only stars inside a one-cell shell around the eye count, which
	1964	holds the same number of cells per steradian in every direction. The pinwheel is
	1965	kept by turning the direction about z before the lookup, and the density matches
	1966	what the old field had at its pole, so the far cap window looks as it did.
`aa201500` 2026-09-14 Pavel Safronov	1967
	1968	## D-061 — The exterior opens on the true hull, and remembers the choice
	1969
	1970	Implemented at the author's request. X-ray was the default because, before D-060,
	1971	the true hull showed little: a bare cylinder, a checkered cap, and land only through
	1972	the far window. Now both caps are windows onto the hazed interior, so the true hull is
	1973	the better first look, and it is what Tower actually looks like.
	1974
	1975	The choice is remembered in localStorage under `tower.exteriorHull`, alongside the
	1976	active view (D-025) and the camera. It is restored into the Hull control at startup,
	1977	before the exterior is built, because the exterior reads its reading from that control
	1978	when it is first constructed. Switching to another view and back never rebuilds the
	1979	exterior, so the reading survives that as well. An unreadable or unknown stored value
	1980	falls back to true. The screenshot script selects a reading explicitly for each shot,
	1981	in a fresh browser context, so it is unaffected.
`376f2ce6` 2026-09-16 Pavel Safronov	1982
	1983	---
	1984
	1985	## D-062 — The world generates itself, live, from features
	1986
	1987	Decided by the author (2026-09-16), after two rounds of answers below. Full design in
	1988	`world-generation.md`; Phase 1 is D-063.
	1989
	1990	The terrain stops being two baked images. `content/geography.json` and `world.json`
	1991	merge into one world file of typed features — mountains, lakes, rivers, seas, desert,
	1992	forest, city, roads and paths — and the browser generates the world from it in web
	1993	workers, on load and on every edit, streaming detail in around the observer. Trees,
	1994	buildings, water surfaces with waves and flow, and ships on the cap seas become real
	1995	geometry.
	1996
	1997	What it reverses. D-006 chose "real geometry, simple ground" and ruled out scenery,
	1998	settlements and vegetation; this adds all three. D-019's build step, its PNG rasters
	1999	and its no-dependencies rule go, along with D-021, D-035, D-039 and D-057, which exist
	2000	only because of the rasters. `design.md` §2.4's "water has no surface" and §4's
	2001	"procedural scenery, settlements" out-of-scope line are reversed. These are listed in
	2002	`world-generation.md` §11. The author agreed to all of it by asking for the build to
	2003	begin.
	2004
	2005	Author's answers, 2026-09-16 (`world-generation.md` §12.1). Nothing inside Tower is
	2006	a remnant: every landform, including the desert's dry washes and shorelines, is there
	2007	because Tower chose to quote something it saw on Earth or elsewhere. Wind is Tower's
	2008	published weather plan, which it revises when it likes — storms are scheduled, and
	2009	cancelled if someone foolish sails into one. Settlements are generated to fill the
	2010	world. Trees are real size, with a tall forest of 90–115 m giants as a landmark. A
	2011	3–8 s compile behind a Terraria-style status dialog is acceptable. Phones get the same
	2012	world at full density, with shorter draw distances. Second round: days count from 0; the seed lives in
	2013	the world file, so the world is identical for every reader; generated settlements are
	2014	named for their size; the tall forest is placed at random near the capital once and
	2015	then fixed; the status dialog lines are approved.
	2016
	2017	---
	2018
	2019	## D-063 — Phase 1: the ground is generated, and the rivers had to be engineered
	2020
	2021	Built. `world-generation.md` phase 1: world file v2 and the one-off port of v1's
	2022	geography; the world compile and the elevation function (`src/terrain/`), in a pool of
	2023	web workers; the ground as a quadtree of tiles; the status dialog; keyboard flight; and
	2024	the map and the x-ray hull drawn from a picture rendered live from the features.
	2025	`tools/terrain.mjs`, its PNG encoder and noise, the three terrain PNGs, `raster.ts` and
	2026	`content/geography.json` are gone. Two runtime dependencies came in, simplex-noise
	2027	and comlink; the other three the plan adopts wait for the phases that need them.
	2028
	2029	One `geometry` field for every kind. The plan sketched kind-specific fields —
	2030	`area`, `shore`, `course`. As built, every feature has `geometry` and its parameters
	2031	sit beside it, so the map, the list, the editor and the permalinks needed no per-kind
	2032	code. Parameters arrive with the phase whose generator reads them.
	2033
	2034	The rivers cross a lowland they cannot get out of, and the profile had to be chosen.
	2035	v1 authored a 60 m lowland between the Girdle and the Dry Reach and a 800–1,100 m
	2036	plateau beyond it, and painted its rivers across at about 1,000 m. Generated honestly,
	2037	a river cannot do that. Three profiles were tried against the real ground:
	2038
	2039	\| Profile \| Result \|
	2040	\|---\|---\|
	2041	\| clamped to the ground \| every river falls to 120 m in the lowland and cuts a 900 m canyon through 100 km of desert \|
	2042	\| graded curve, source to mouth \| a 1,300 m embankment across the lowland \|
	2043	\| least earthwork (isotonic regression, ≥ 2 m/km) \| The Narrows cut 300–500 m deep, as §1.11 says; up to 880 m of embankment under each river \|
	2044
	2045	The third is built. It is the only one of the three that is a rule rather than a guess —
	2046	Tower moves as little ground as it can — and it reproduces the canyon canon already
	2047	described. The embankments are the price, and they are real in the render. Whether the
	2048	world should change instead (raise the lowland, or let it hold a lake) is recorded as an
	2049	open question in design.md §1.11: that is a decision about Tower, not about the code.
	2050
	2051	Tiles were being measured against the tallest mountain. A fixed 4.5 km allowance for
	2052	relief made every tile within 5 km of the eye look near enough to split, all the way to
	2053	level 10 — 11,000 tiles on screen. Each tile is now bounded by the relief it actually
	2054	has, and the split distance was measured down from 2.4 to 1.8 tile-widths: 1.9 M
	2055	triangles and about 200 draw calls, settled in 1.4 s at the Capital.
	2056
	2057	A coarse tile cannot show an 80 m river. One vertex of a 500 m tile falling in a
	2058	channel painted a 500 m blue diamond, and the far side of the world wore a dotted line
	2059	down every river. Water now weighs by how much of the vertex's spacing it covers.
	2060
	2061	The x-ray hull had been mirrored end to end since D-024. Three's cylinder puts v = 1
	2062	at the far cap after the rotation, and the UV rewrite flipped it, so the Girdle's snow
	2063	was drawn at the far cap. The committed v1 screenshot shows it. Fixed with the rest,
	2064	since the picture on the hull changed anyway.
	2065
	2066	Readers pay a compile on every visit, and it is short. About 120 ms to compile and
	2067	1.2 s to draw the map, so the status dialog rarely appears on the author's laptop; it
	2068	will on slower machines, and it will once erosion arrives in phase 2. A production
	2069	build reaches a fully detailed interior 4.3 s after a cold load.
`14d35b7d` 2026-09-16 Pavel Safronov	2070
	2071	---
	2072
	2073	## D-064 — The Uplands, a hull a kilometre thick, and rivers that prefer to cut
	2074
	2075	The author's call on D-063's open question: raise the lowland. A region can now
	2076	carry an uplift, and The Uplands (provisional name) lift the ring of ground between
	2077	the Girdle and the Dry Reach — z 105 to 150 km — to about 1,000 m. A raised region lifts
	2078	the ground to a height rather than by one: it takes the higher of itself and what is
	2079	already there. Added, it stacked onto the plateau's rising edge and moved the river's
	2080	problem ten kilometres deepward instead of solving it; taken as a maximum, it meets the
	2081	plateau and stops.
	2082
	2083	Measured before choosing, as each river's worst embankment:
	2084
	2085	\| \| worst fill \| The Narrows cut \|
	2086	\|---\|---\|---\|
	2087	\| v1's lowland \| 895–980 m \| ~670 m \|
	2088	\| Uplands at 1,000 m, least-squares profile \| 240–350 m \| ~390 m \|
	2089	\| Uplands at 1,000 m, 0.2-quantile profile \| 93–199 m \| ~540 m \|
	2090
	2091	The profile now prefers cutting to filling. Least squares treats a metre of
	2092	embankment like a metre of cutting, and so balances the two; but a cut through high
	2093	ground reads as a valley and a 300 m bank across a plain reads as a dam. Fitting the
	2094	0.2-quantile instead keeps the water below four-fifths of the ground in any reach. The
	2095	Narrows comes out a little past design.md's "up to 480 m" at about 540 m, which is
	2096	recorded rather than tuned away.
	2097
	2098	The hull is a kilometre thick all the way round. D-044 gave Tower a 1 km wall and
	2099	drew it only at the caps, reasoning that a closed can shows no thickness from outside.
	2100	That held while the ground never went far below the floor. Generated sea floors reach
	2101	330 m down and the tiles hang skirts below that, and with the outer skin at the floor's
	2102	own radius they came through the hull as a lattice of lines (the author's screenshot).
	2103	The skin now sits a kilometre out, rings close the wall at both ends, and nothing
	2104	generated may go deeper than 950 m below the datum.
	2105
	2106	Two smaller faults found on the way.
	2107
	2108	- A `#view=` link built its 3D view twice. Reading the hash and choosing the initial
	2109	view each started the deferred import before either had finished, so the exterior
	2110	drew every caption twice and the interior would have run two renderers and two sets
	2111	of tiles. The import is now shared.
	2112	- Rivers drew as dotted lines at a distance. D-063 weighted water by the channel's
	2113	width over the vertex spacing, but only at vertices that landed in the channel. It
	2114	now weighs every vertex by the share of its own cell the channel covers, so a coarse
	2115	tile draws a river as a faint continuous line.
`e7a9f9dd` 2026-09-17 Pavel Safronov	2116
	2117	---
	2118
	2119	## D-065 — Mountains are summits, ridges and valleys water cut
	2120
	2121	Built. `world-generation.md` phase 2, in `src/terrain/mountains.ts`; the full account
	2122	of what changed from the plan is §16 there. In short:
	2123
	2124	- Summits — authored ones held exactly, generated ones lower than the named summit near
	2125	them — are triangulated with delaunator, and the minimum spanning tree plus some
	2126	short spurs become ridges with saddles.
	2127	- Ridges are raised into an uplift field with broad, gently concave flanks, blended by a
	2128	soft maximum so neighbouring ridges do not meet along straight creases.
	2129	- The field is eroded to the stream-power steady state, computed directly over a
	2130	drainage network found by priority flood, rather than by iterating the implicit solver
	2131	the plan named. Iterating cut almost nothing in a reasonable time, and what it cut ran
	2132	along the grid. A roughened start, per-pass jitter, and a 1 km² channel threshold with
	2133	hillslope diffusion are what keep the grid from showing.
	2134	- The steady state is scaled back to the authored field over 8 km, so the massifs keep
	2135	the author's heights and gain the water's structure.
	2136	- Overlapping ranges are eroded as one group.
	2137
	2138	Named summits are the top of their ground, and that took four steps, because every
	2139	earlier one broke the one before: scale the surroundings down smoothly if anything is
	2140	higher, pin the summit, trim the few cells pinning lifted past a gentle cone, pin again.
	2141	A flat cap instead left a grey disc with a rim around every summit; a single pin left
	2142	neighbours 9 m above Snowcrown.
	2143
	2144	A summit's elevation is above the datum. v1 added a massif's height on top of the
	2145	ground under it, so Snowcrown's 4,300 m stood at 4,447 m. Heights everywhere else in the
	2146	world are above the datum, so the land under a summit is now subtracted before it is
	2147	pinned, and the checks that it is the top compare total heights.
	2148
	2149	Tarns keep a shelf. Eroded honestly, the ranges ran valleys through three of v1's four
	2150	tarns and dropped them to 500 m. A cirque is exactly ground the rivers did not cut, so an
	2151	authored bowl lake now keeps its range at the uncut height around it. Sisters Tarn is at
	2152	2,173 m; design.md puts the ski station beside it at 2,100 m.
	2153
	2154	Regions raise ground after the mountains, not before. Lifted first, The Uplands'
	2155	1,000 m stacked under the Girdle's foot and put Sisters Tarn at 3,665 m.
	2156
	2157	One compile at a time. Dragging a summit updates the world on every pointer move, and
	2158	each would have started a four-second compile. The service now runs one, keeps only the
	2159	latest edit made during it, and runs that next.
	2160
	2161	Cost: a whole-world compile of about 4 s in a browser worker, most of it the Girdle.
	2162	Loading to a drawn map is about 4 s; the status dialog shows for it.
`80bcc97a` 2026-09-17 Pavel Safronov	2163
	2164	---
	2165
	2166	## D-066 — Water is geometry, and waterfalls lean
	2167
	2168	Built. `world-generation.md` phase 3; §17 there has the detail.
	2169
	2170	Every lake, sea and river is a surface at its own level, meshed per tile, with the
	2171	ground beneath it drawn as bed and the depth test finding the shore. Waves are the wind's:
	2172	fetch-limited JONSWAP seas, capped at fully developed, which is what a ring sea always is.
	2173	Rivers run at Manning's speed and whiten on their rapids.
	2174
	2175	Waterfalls were not added; they were found. The river profile of D-064 pools reaches
	2176	and steps between them, and a step is a fall. Merged where the steps come in a run, they
	2177	give the Long Water a 680 m fall off the range with a gorge below, a 170 m fall into The
	2178	Deep, and Eastwall Water a 440 m fall. The sheets drift antispinward as they drop, by the
	2179	free-fall drift canon already computed: §1.9.2 said the waterfall would be the compass,
	2180	and it now is one.
	2181
	2182	Three faults the falls surfaced, each a straight cliff where there should be none: the
	2183	step applied across a whole valley's width (fixed by shaping walls from a smoothly ramped
	2184	wall level), across a whole embankment's (the embankment uses the ramp too), and a second
	2185	cliff kilometres out where a steep gorge wall's straight term fell below the valley's
	2186	curved one (the gorge shares the curvature). Tile skirts now reach past each tile's
	2187	steepest step, or a gorge wall's cracks showed as dark panels.
	2188
	2189	The beam went fuzzy over the seas — the author saw it. Water blends and is drawn after
	2190	the ground; the beam was drawn before the water, so distant seas blended over it. The beam
	2191	is always in front of water, and is now drawn last.
	2192
	2193	A river met its lake badly — the author saw that too. v1's Long Water starts at the
	2194	centre of Hollow Tarn, so its channel was cut across the lake floor and drawn as a stripe
	2195	across the tarn. Rivers are now trimmed to the water's edge, found by sampling the water
	2196	rather than the lake's outline, which a tarn's water does not reach.
	2197
	2198	The screenshot tour was running the author's laptop hot, and deserved it: fifteen
	2199	shots, four at a time, each in a fresh browser context compiling the whole world with a
	2200	full pool of six workers. The compiled world is now cached in IndexedDB — which returning
	2201	readers get too, as §5.4 planned — keyed so a new build never reads an old world. The tour
	2202	compiles once, hands every shot the warm cache, caps each page at two workers, and reports
	2203	any shot that compiled anyway. It takes about 100 s and averages about 3.4 cores.
`a088e5d5` 2026-09-17 Pavel Safronov	2204
`28896dab` 2026-09-27 Pavel Safronov	2205	A regression this caused, fixed 2026-09-27. Giving water render order 1 put it after
	2206	the cap windows, which kept the default 0. From outside, a window is transparent and in
	2207	front of the seas: drawn first, it wrote its depth, and the water behind it failed the
	2208	depth test — so looking in at the sun cap showed the Sunward Sea's bed, brown, instead of
	2209	the sea. See-through windows now draw at 2.5, after the water and the falls and before the
	2210	beam, and the exterior's rays share that order so they still sort against the glass.
`59c12827` 2026-09-28 Pavel Safronov	2211	D-074 moved them to 4, after the beam and the sun, so both show from outside too.
`28896dab` 2026-09-27 Pavel Safronov	2212
`a088e5d5` 2026-09-17 Pavel Safronov	2213	---
	2214
	2215	## D-067 — Tower's weather is a published plan, and ships read it
	2216
	2217	Built. `world-generation.md` phase 4; §18 there has the detail.
	2218
	2219	The plan keeps both halves. An event records when Tower published it; a revision
	2220	records when Tower changed its mind, and why. The world then has two readings of the same
	2221	plan: what Tower does (the event as revised so far) and what people were told (the event as
	2222	published). The waves follow the first. The ships follow the second, so a sailor warned of
	2223	a storm stays in harbour even after Tower quietly calls it off — and a storm nobody was told
	2224	about catches the fleet at sea. This is the author's answer to D-062 question 2 made
	2225	mechanical.
	2226
	2227	Weather and ships are functions of the clock, not simulations. The interior's clock
	2228	gained a day number, counted from 0 (D-062). Wind, sea state, whitecaps and every ship's
	2229	position are pure functions of the world file and the hour, so a link to a moment shows the
	2230	same sea to everyone, and scrubbing the clock is the same as waiting.
	2231
	2232	The sea state left the compile. Phase 3 baked waves into the compiled world; with
	2233	weather that changes hourly that would recompile the world for every gust. Water vertices
	2234	now name their body and the shader looks the waves up. Wind and ship counts no longer
	2235	change the terrain signature.
	2236
	2237	A float32 precision fault was fixed on the way. The water shader's around-the-ring
	2238	coordinate was only good to about 6 m, and whitecaps drew as streaks radiating from the
	2239	eye. Water now measures from a reference near the eye that steps in whole noise periods,
	2240	with each wave's phase at the reference added from float64. River ripples were suffering
	2241	from the same fault unnoticed.
	2242
	2243	Simplifications, each recorded: no tacking, no harbours drawn, ships only on ring
	2244	seas, flat sails. The default plan is 28 days long and repeats: storms and gales on both
	2245	cap seas and The Deep, a flat calm, and one storm Tower holds down to a gale because a
	2246	fishing boat ignored the plan.
	2247
	2248	---
	2249
	2250	## D-068 — A scenic tour, and screenshots taken from it
	2251
	2252	Built, between phases 4 and 5 at the author's request. `world-generation.md` §19 has
	2253	the detail.
	2254
	2255	Particular things want a way to be found. With a 678 m waterfall, a fleet and scheduled
	2256	storms in the world, standing still and dragging is a poor way to see them, and a
	2257	screenshot placed by hand at coordinates goes stale the moment the world is edited. The
	2258	interior view now flies a tour of nine scenes, thirty seconds each — helicopter arcs,
	2259	flights along rivers and over The Deep, chase cameras on ships — each worked out from the
	2260	compiled world rather than written down, so it follows edits. The screenshot tour
	2261	photographs those scenes, replacing four hand-placed shots, and a test flies all of them.
	2262
	2263	Ships can be caught out. A ship chased through bad weather needs a ship at sea in bad
	2264	weather, and D-067's harbour rule emptied the sea the moment any gale was published, even
	2265	one published as it began. Ships now go in only for weather announced at least three hours
	2266	ahead — the time they are given to get in — and ride out anything announced later. The
	2267	default plan gains one such squall on the Sunward Sea, on day 11. Tower keeping the fleet
	2268	honest about its warnings is also a better story than a fleet that vanishes on cue.
	2269
	2270	A documentation fault, corrected. design.md §3.1 and world-generation.md §10 said
	2271	readers were the default and `?writer=1` enabled editing. The app has always done the
	2272	opposite — writer by default, `?reader=1` to preview — and the README said so. The docs now
	2273	match the code; D-011's reasoning is unchanged.
`51c39d8b` 2026-09-17 Pavel Safronov	2274
	2275	---
	2276
	2277	## D-069 — The desert is quoted, not remembered
	2278
	2279	Built. `world-generation.md` phase 5; §20 there has the detail.
	2280
	2281	Tower laid out the whole grammar of a desert, dry channels included. The Dry Reach now
	2282	has tiered plateaus and mesas over sand, dune fields whose crests run across the wind Tower
	2283	sends, rocky hills, a network of dry washes, salt pans in its low ground, and the shoreline
	2284	benches of lakes that never stood. None of it is history — §0.1 — and the washes are routed
	2285	by the same drainage the ranges erode with precisely because a channel network that ignores
	2286	drainage is the one thing that would look wrong.
	2287
	2288	Two findings worth keeping. A flat desert has no drainage: water fans out from every
	2289	edge and nothing gathers, so Tower crowns the Reach 150 m in its middle and the washes run
	2290	off it. And widening a one-cell channel by blurring destroys it — dilate first, blur after.
	2291
	2292	Terracing belongs to the tiles, not to the compile. The grid stores how far up the
	2293	plateau the ground stands; the tile turns that into flat tops and steep risers at its own
	2294	detail. The other way round, every cliff band is one grid cell wide, which is a ramp.
	2295
	2296	Tower does not build rock across water it has already routed. Mesas dropped across The
	2297	Narrows turned the world's whole drainage into a kilometre-deep slot. Rivers carry a 6.5 km
	2298	corridor clear of plateau rock, which reads as the sand-and-gravel valley a river makes
	2299	through a mesa field. The consequence recorded: the Long Water's canyon through the lower
	2300	Dry Reach now runs about 780 m deep where the old test expected under 700, because the rim
	2301	above it is the desert's rock as often as the ridge's.
	2302
	2303	Materials gained `salt`, and rock is now claimed rather than inferred. Bare rock came
	2304	only from mountain relief before; it now comes from the desert's plateaus and wash floors
	2305	and from any steep desert face, since sand does not stand at 60°. The ground shader lays
	2306	strata by elevation — a level in Tower is a constant radius — so the bands line up across
	2307	separate mesas, which is what makes a mesa field read as one formation.
	2308
	2309	---
	2310
	2311	## D-070 — The Capital is a footprint, and it reaches the water
	2312
	2313	Built alongside phase 5, at the author's suggestion.
	2314
	2315	The Capital was a dot 11 km inland of the lake it was said to stand on. design.md §1.11
	2316	described it as sitting on The Deep's antispinward shore; the generated water starts
	2317	10.75 km spinward of its marker, on ground falling from 300 m to the water's 120 m. The fix
	2318	the author chose is the better one: make the city big enough to be both things at once.
	2319
	2320	So a settlement is now a point with an extent — a size and a radius — rather than a
	2321	marker alone. The author places where a town stands and how far it reaches; Tower lays the
	2322	streets and buildings out inside that footprint (phase 7). The map draws the extent as a
	2323	ring around the marker, and each size carries the usual radius a new settlement starts from.
	2324
	2325	The Capital is 20 km across, centred on the axial midpoint of the world — the only
	2326	centre a cylinder has — which is just wide enough that its spinward edge is the waterfront.
	2327	It holds the centre of the world and a shore on The Deep, and nothing had to move to make
	2328	that true but 2 km of marker.
	2329
	2330	A canon figure corrected on the way: §1.11 said the far shore was "32 km away and
	2331	2.6 km above eye level". Thirty-two kilometres around the cylinder would stand 9.9 km above
	2332	the eye; 2.6 km is the figure for the 18 km of water The Deep actually fills, which is what
	2333	the crossing is. The outline is 154 × 32 km; a flooded basin's water is narrower than its
	2334	outline, as it should be.
`00ceeedb` 2026-09-27 Pavel Safronov	2335
	2336	---
	2337
	2338	## D-071 — Three billion trees, and only the near ones exist
	2339
	2340	Built. `world-generation.md` phase 6; §21 there has the detail.
	2341
	2342	A forest is a rule, not a list. Greatwood holds about three billion trees at 450 stems
	2343	per hectare. What exists at any moment is the few thousand near the eye, and they are found
	2344	by a stratified rule — a fixed grid about one tree apart, each cell hashing its own
	2345	coordinates — so every tile, every view and every visitor works out the same trees without
	2346	knowing about any other. Walk a kilometre and come back and every trunk is where it was.
	2347
	2348	Four ranges, and the handovers are the hard part. Meshes to about 210 m, cards to
	2349	560 m, clumps — one card for every 30 m cell of wood — to 2.8 km, and under all of it from
	2350	620 m the canopy skin: the tile's own ground vertices lifted by the canopy's height, which
	2351	sinks back onto the ground as the eye comes close. The plan had three ranges ending at
	2352	760 m; the author saw the result at once — the wood ended a few hundred metres away and its
	2353	edge walked along with him. Clumps are what made the reach affordable: a card per tree is a
	2354	million of them at two kilometres, a clump every thirty metres is twenty-six thousand.
	2355
	2356	The skin was a black ceiling from underneath, because a solid surface is what it is.
	2357	Real canopy is about a third gaps, and those gaps are the only reason you can see any
	2358	distance through a forest. From below it now drops half its fragments on a ground-fixed
	2359	pattern and lights the rest as light that has come through leaves.
	2360
	2361	Two numbers the plan had wrong, found by looking: a meadow's edge cannot be a fixed
	2362	±0.02 of noise (the same lesson the desert taught about thresholds), and a giant's crown
	2363	cannot be 0.34 of its height — at that width a grove at 55 stems/ha closes over into
	2364	unbroken black. Redwoods carry about 0.18, and 30 stems per hectare puts 18 m between
	2365	trunks.
	2366
	2367	A city is not a wood. The Capital sits in the middle of Greatwood's belt, so once the
	2368	forest was trees it was standing in one. Settlements now clear the forest inside their
	2369	footprint — which is also the first thing the Capital's new extent (D-070) has been used
	2370	for.
	2371
	2372	The Tall Wood was placed once and written down. Drawn from the world's seed under the
	2373	author's rules — inside Greatwood, 15–50 km from the Capital, off the water and the rivers
	2374	— it landed 17 km from the city and is now a feature in `content/world.json` like any
	2375	other. Greatwood gives up its own ground wherever a smaller forest stands inside it.
`9d5a061c` 2026-09-27 Pavel Safronov	2376
	2377	---
	2378
	2379	## D-072 — The world is changed in the repository, not in the browser
	2380
	2381	Decided by the author, 2026-09-26. Built.
	2382
	2383	The author writes alongside the app, not with it. The writer mode of D-011 assumed
	2384	the map would be the place the world was built: draw a lake, name it, export, commit.
	2385	In practice the author wants to build it by describing changes to Claude, who edits
	2386	`content/world.json` directly — the Tall Wood (D-071) was placed that way. The editor
	2387	had become a second, stale path to the same file, and a hazardous one: a browser's local
	2388	draft was preferred over the committed world, so a draft from last week could quietly
	2389	shadow everything added since.
	2390
	2391	So the app only reads. Removed: writer and reader modes (`?reader=1` now does
	2392	nothing), the Add tools, the feature editor and every per-kind control, dragging
	2393	markers and summits, weather scheduling and revision, the localStorage draft and its
	2394	autosave, backup and static exports, import, and "Reload published world". Everyone
	2395	sees what readers always saw — the three views, the list, a detail card, the published
	2396	weather board and the sightline probe — and the editor's read-only readouts moved onto
	2397	the detail card: a range's summits, a lake's or sea's waves now, a forest's tree count
	2398	and spacing, and what a desert's landforms came to.
	2399
	2400	`content/world.json` is the one authoritative copy. `static/world.json` is gone:
	2401	the public file is made from `content/world.json` by a Vite plugin — served fresh by the
	2402	dev server, emitted into the build — with notes emptied and unpublished features left
	2403	out, exactly as `toPublicWorld` always did. Two committed copies had to be kept in step
	2404	by hand and a test; one copy and a derived file cannot disagree. Saving the file with
	2405	the dev server running reloads the page, so a change can be looked at as soon as it is
	2406	made.
	2407
	2408	What stays true from D-011: no backend; the world lives in git beside the prose,
	2409	diffable and versioned; `notes` and `"public": false` keep things out of the app but
	2410	not out of the public repository.
	2411
	2412	A browser that used the old writer mode still holds its draft under
	2413	`tower.world.v2` in localStorage. The app no longer reads or writes it, and does not
	2414	delete it either, so anything in it that never reached the repository can still be
	2415	recovered by hand.
`b4b8e46c` 2026-09-28 Pavel Safronov	2416
	2417	---
	2418
	2419	## D-073 — Every branch is on Pages, and the app can switch between them
	2420
	2421	Decided by the author, 2026-09-27. Built.
	2422
	2423	The author wants to look at a branch in the browser without running a dev server —
	2424	`main` and `forest` side by side. GitLab does this natively with parallel deployments,
	2425	but they are Premium-only and this project is on the Free plan. So the default branch's
	2426	pipeline builds every branch into one site (`tools/build-pages.sh`): the default branch at
	2427	the root, each other branch in a folder named for it. A push to any other branch triggers
`54204da9` 2026-09-28 Pavel Safronov	2428	a default-branch pipeline, so its preview follows it within minutes. That goes through the
	2429	pipeline trigger API with a trigger token kept in the masked CI/CD variable
	2430	`PAGES_TRIGGER_TOKEN`: a `trigger:` job cannot start a pipeline in its own project, which
	2431	is how the first attempt failed (`invalid_bridge_trigger`). The token acts with the
	2432	author's permissions and cannot be limited to protected branches, since previews come from
	2433	unprotected ones; acceptable while the author is the only one who pushes. A branch that fails
`b4b8e46c` 2026-09-28 Pavel Safronov	2434	to build is left out with a warning rather than failing the deploy, so a broken experiment
	2435	can never take the published map down.
	2436
	2437	The app has a branch picker. Every folder of the site carries a `branches.json`
	2438	listing what was built, with links relative to that folder; the panel header shows a
	2439	picker when there is more than one branch, and switching keeps the hash, so the other
	2440	branch opens on the same view and place. Without the list — on a dev server — there is
	2441	no picker. A branch built from code older than the picker has none either, but it is
	2442	still listed in the others.
	2443
	2444	One origin, shared storage. The branches are paths on one site, so localStorage and
	2445	the IndexedDB world cache are shared. That is harmless — the cache is keyed by the world
	2446	and the build — but it means a remembered view or camera follows you between branches.
`59c12827` 2026-09-28 Pavel Safronov	2447
	2448	---
	2449
	2450	## D-074 — Tower has a sun, and keeps Sydney's days
	2451
`f16f084e` 2026-09-28 Pavel Safronov	2452	Amended by D-079: the rod no longer stops at the sun; it runs on to the far cap.
`de3d60d6` 2026-09-29 Pavel Safronov	2453	Amended by D-089: Tower keeps Pisa's days now, not Sydney's.
`f16f084e` 2026-09-28 Pavel Safronov	2454
`59c12827` 2026-09-28 Pavel Safronov	2455	Decided by the author, 2026-09-27. Built. Supersedes D-053; amends D-008.
	2456
	2457	A sun at one point on the beam, crossing the world once a day. The swarm gathers into
	2458	one shoal on the axis and throws the beam's light out sideways there: a single blazing
	2459	point, Tower's sun. It lights at the sun cap at sunrise, moves at a steady speed down the
	2460	whole axis, and goes out at the far cap at sunset. Night follows, with the beam banked to
	2461	a thousandth as D-046 has it, and the next morning the sun lights at the sun cap again.
	2462	This replaces D-053's dawn front, and it turns D-008's aim — no mirrorlet may read as a
	2463	sun — into its exception: Tower wants exactly one.
	2464
	2465	Sydney's days. Tower keeps an Earth city's clock and its sunrises and sunsets, and the
	2466	author chose Sydney. The clock is Sydney's local time, daylight saving included; day 0 is
	2467	23 September 2026, the southern spring equinox, so the calendar starts on a 12-hour day and
	2468	lengthens toward December. Sunrise and sunset come from NOAA's solar position equations,
	2469	within a few minutes of the published tables (test/sky.test.ts). The city and the date of
	2470	day 0 are a `calendar` in `content/world.json`, so both can be changed.
	2471
	2472	How it lights the world. Tower aims the shoal's mirrorlets so the sun is as bright seen
	2473	from anywhere as from beneath it, so the ground is lit by how squarely it faces the sun, as
	2474	under Earth's — not by the inverse square of a lamp 50 km up, which would leave most of
	2475	the world in twilight all day. Each place has its own noon when the sun passes overhead;
	2476	the sun never sets, since from one cap the sun at the other is still 5° up. The shoal takes
	2477	the whole beam where it gathers, so the rod glows from the sun cap to the sun and is dark
	2478	past it, and the far cap's mirror is lit only by the banked beam at night.
	2479
	2480	Consistency, inside and out, was the author's main requirement. So there is one of
	2481	everything: one world clock (`src/world/clock.ts`), one sky computed from it
	2482	(`src/world/sky.ts`), and one `setSky` that lights the meshes both 3D views share. The sun is
	2483	one of those meshes, lit by the same haze integral as everything else, and every lit surface
	2484	goes through one shader function, `direct`. The exterior used to render the inside at a
	2485	fixed noon; it now follows the clock like the interior does, the windows draw after the
	2486	beam and the sun so both show through the glass, and the rays from the collector dim when
	2487	the beam is banked. The map marks the sun and shades the world by the same light curve.
	2488
	2489	What changed in the app. The Time controls — day, time, run the clock, its rate, and how
	2490	bright Tower runs its day — moved out of the interior's options into a Time section that
	2491	works in every view. The Light menu lost "Night", which the clock now brings; "Dark" still
	2492	shuts the beam down. Scenic tour scenes with no reason to choose an hour are shown in the
	2493	morning where they are, the sun 60 km sunward and about 40° up.
	2494
	2495	The caps have more than one name. The sun cap; the far cap, also the star cap. The author
	2496	may use any of them, and more are added as they come into use.
`1d447a1a` 2026-09-28 Pavel Safronov	2497
	2498	The clock opens at the time in Sydney now (author, 2026-09-27). It starts from Sydney's
	2499	current date and time, counted from day 0, and runs forward at the speed it was last run at;
	2500	the speed and the Light are remembered, the time is not. `?clock=<hours since day 0>` holds
	2501	it at a moment, stopped, which is what the screenshot tour uses to stay reproducible.
`e331ec2a` 2026-09-28 Pavel Safronov	2502
	2503	## D-075 — The stars are Earth's
	2504
	2505	Asked for (author, 2026-09-27): the sky was "sparse and random"; use the accurate night
	2506	sky from Earth instead.
	2507
	2508	What it is. The Yale Bright Star Catalogue, 5th revised edition (Hoffleit & Warren 1991,
	2509	CDS V/50): 9,096 stars to about magnitude 6.5, which is everything the naked eye sees from
	2510	the darkest sites. `tools/bright-stars.mjs` writes it to `src/data/bright-stars.json` —
	2511	position (J2000), V magnitude, B−V colour — and the app loads it as a separate chunk the
	2512	first time a 3D view opens. The procedural field of D-056 is gone; both views still share
	2513	one `STAR_GLSL`, so the sky out of the far cap and the sky round the exterior are one sky.
	2514
	2515	Where it points. Tower's axis points at the sun (§1.6), so the sky it sees is set by the
	2516	sun's ecliptic longitude, from the clock's real instant (`Sky.at`). The far cap looks at the
	2517	antisolar point; it walks the ecliptic a degree a day, so the pinwheel's centre is in
	2518	Pisces in late September and near Orion's feet at the December solstice. Roll about the
	2519	axis is the spin inside; outside nothing fixes it, so ecliptic north is up (scene +Y), the
	2520	one choice square to the axis every day of the year. Precession since J2000 is applied.
	2521	`src/world/celestial.ts` holds the maths and `test/celestial.test.ts` pins it: the sun where
	2522	the almanac has it at the 2026 equinox and solstice, the frame a rotation and not a mirror,
	2523	and Orion beyond the far cap in December.
	2524
	2525	How it draws. A pixel cannot look at 9,000 stars. They are baked into cells on the six
	2526	faces of a cube (256 a side, each face bordered by its neighbours' stars), and a pixel reads
	2527	the nine cells round its own. Each star is still the pixel-wide gaussian of D-056, which
	2528	stays still rather than flickering; brightness goes as flux^0.8, and the brightest grow a
	2529	little wider. Colour follows B−V, kept pale, since starlight is too faint for much colour
	2530	vision.
	2531
	2532	The motionless star is not canon (author, 2026-09-27). §1.4 promises "one motionless
	2533	star at the middle" of the pinwheel. The real sky has no star at the antisolar point, and it
	2534	moves a degree a day, so no real star could hold the place. The placed star at the centre
	2535	stays for now and is on the backlog (design.md §5) to revisit; removing it is one line.
`908833aa` 2026-09-28 Pavel Safronov	2536
	2537	## D-076 — Trying other outsides: rock
	2538
	2539	Asked for (author, 2026-09-27): the outside is very uniform; try different approaches,
	2540	starting with "a rocky surface, almost like someone tried to make it look like an asteroid".
	2541
	2542	An experiment, not canon. It is a third reading in the Hull switch — True, Rock,
	2543	X-ray — so it can be compared with the bare skin and kept, changed or dropped. Nothing in
	2544	§1 changes. A shell kilometres thick round a 540 km cylinder would be over 10¹⁸ kg of rock
	2545	on a spinning hull, which canon would have to answer for if it stays.
	2546
	2547	The shape. A shell laid over the metal skin, never cut into it: at least 0.2 km proud of
	2548	the wall's outer face everywhere, so nothing inside can show through, and the windows are
	2549	left clear, with rock heaped round their rims. Lumps a few kilometres high and 1,400 craters
	2550	from 0.8 to 16 km across, a power-law population with raised rims, in the geometry
	2551	(`src/views/rock-hull.ts`, ~0.7 km between vertices); grit, pits and colour mottling below
	2552	that in the shader, bump-mapped from noise in the hull's own frame and faded out as a pixel
	2553	grows wider than the detail. Seeded, so it is the same rock on every load. Clicks still land
	2554	on the smooth skin, hidden, so a picked point is the canon point beneath the rock. The rock
	2555	is lighter than a real carbonaceous asteroid's charcoal, under the same licence as the fill
	2556	lights.
`6368edcf` 2026-09-28 Pavel Safronov	2557
	2558	## D-077 — Trying other outsides: piping
	2559
	2560	Asked for (author, 2026-09-27): "something industrial like the Alien movies, with
	2561	panels and pipes and exhaust vents, only on a huge scale." The second experiment after
	2562	D-076's rock, and like it a Hull reading labelled as an experiment, not canon.
	2563
	2564	What is on the skin (`src/views/pipe-hull.ts`):
	2565	- Plating, in the shader of the hull cylinder itself: 40 panels round by 8 km along,
	2566	some split in two or four, each its own shade, some ribbed, with grime streaked along the
	2567	axis. The seams are bump-mapped grooves that widen with the pixel, so they fade with
	2568	distance instead of aliasing. 40 round, so the plating closes on itself exactly.
	2569	- Bands every 30–42 km: a collar 2.6 km wide and 1.1 km proud, with a hoop pipe either
	2570	side. They divide the 540 km into stations, as a ship's frames do.
	2571	- Pipe runs: fourteen lanes of two to five pipes, 0.6–2.2 km across, running from band
	2572	to band on brackets, and skipping some stations so runs start and stop.
	2573	- Plant: 170 clusters of blocks, the odd one a tower up to 6 km, kept off the pipe
	2574	lanes and the bands.
	2575	- Vents: 64 housings, each with a grille and a plume that fire at random, on
	2576	separate clocks (author: "the plumes are constant, let's make them fire at random", and
	2577	the same for the lights). A grille is lit 1–7 s at a time and an ember between; a plume
	2578	shoots out for 1.5–4 s, then is gone for 2–12. There is no air to hold a plume out here;
	2579	it is the look asked for, under the same licence as the beam's rays.
	2580	- Thrusters ("larger fires, light them less often"): six pods at each end, in 12 km
	2581	kept clear for them, each with four bells — out, spinward, antispinward, and along the
	2582	axis away from the world. They are what §1.6's never-ending reorientation would need. A pod
	2583	burns one bell at a time for 2–5 s, then waits 25–80 s. The flame is 28–42 km long and fans
	2584	out as it goes, since exhaust in vacuum has nothing to hold it together; the spread grows as
	2585	the square of the distance, so a sideways flame stays clear of the plating it runs along.
	2586	Two point lights follow the two brightest burns and light the hull round them — one per
	2587	bell would recompile every material for 48 lights, and more than two at once is rare.
	2588
	2589	Dust (author: "it's too shiny, let's add a nice layer of dust"). Every piping material —
	2590	plating, bands, pipes, blocks, pods, bells — goes through `dusty()`, which lays a dust
	2591	colour over the surface and makes it fully matte and non-metallic where it lies. Bare
	2592	metal under the fill lights read as chrome. Dust is thickest on faces turned outward, the
	2593	way everything on a spun hull is thrown, thinner on the sides and nearly absent
	2594	underneath, and broken into drifts by noise. It is worked out in the hull's frame from the
	2595	vertex, so instanced parts get it too.
	2596
	2597	Timing is on the CPU. A small scheduler, `Bursts`, holds each emitter's on/off clock and
	2598	a level that rises over an attack and falls over a release, so nothing snaps on or off. The
	2599	shaders only read that level per instance. The lights need to know which bells are burning,
	2600	and keeping one schedule in JavaScript is simpler than mirroring a hash in GLSL. It is
	2601	tested in test/pipe-hull.test.ts, including a long pause such as a hidden tab.
	2602
	2603	What it must not do. Nothing is inside the skin (test/pipe-hull.test.ts checks every
	2604	vertex), and the bands are open drums with ring faces, never closed cylinders, whose
	2605	discs would stand inside Tower and be seen through the windows. The plating is two-sided
	2606	so that, from inside the tube, it hides the far side's parts wherever the ground has not
	2607	been generated yet. All the parts are instanced: five draw calls for a few thousand pieces.
	2608	Seeded, so it is the same plant on every load, and built only when first chosen.
`c954a775` 2026-09-28 Pavel Safronov	2609
	2610	## D-078 — The cap windows are framed in identical triangles
	2611
	2612	Asked for (author, 2026-09-27): "the endcaps have circular window patterns. please
	2613	change these to be identical triangle pieces."
	2614
	2615	Both caps' mullions are now a lattice of identical equilateral triangles: three
	2616	families of straight members at 60° to each other, spaced one triangle height apart, with a
	2617	vertex on the axis where six triangles meet. The rings and spokes of the polar tessellation
	2618	(D-041, D-045), including the heavy ring at 25 km, are gone. The side is 8.3 km — the old
	2619	cells' pitch, six across the radius — so the frame keeps its job as the one object in the
	2620	world with a known size (D-041), and it is coarse enough to survive the near edge-on views
	2621	from outside (D-030).
	2622
	2623	A disc cannot be tiled by whole triangles, so the rim cuts the outermost ones; the heavier
	2624	rim mullion closes them. The far cap's mirror sits on the centre vertex, and the sun cap's
	2625	aperture on the same point of its own lattice. `test/cap-pattern.test.ts` pins the lattice:
	2626	three directions 60° apart, every line a whole triangle height from the axis. The polar
	2627	checker drawing, which nothing had used since both caps became windows, was removed with it.
`f16f084e` 2026-09-28 Pavel Safronov	2628
	2629	## D-079 — The beam runs from cap to cap, and the sun moves along it
	2630
	2631	Asked for (author, 2026-09-28): "can we keep a permanent light beam in the sky (from cap
	2632	to cap) and the large sun still moving from cap to cap? … i'd like the beam extend all the
	2633	way, not just from sun cap to the sun."
	2634
	2635	Under D-074 the sun's shoal took the whole beam where it gathered, so by day the rod glowed
	2636	from the sun cap to the sun and was dark past it, and the far cap's mirror was lit only at
	2637	night. Now the shoal takes part of the beam and the rest runs on: the rod is one line of
	2638	light from cap to cap at every hour, the sun still crosses the world along it, and the rod
	2639	always ends on the lit mirror at the centre of the far cap (D-048).
	2640
	2641	How much gets past. `BEAM_PAST_SUN = 0.7` in `SUN_GLSL`: past the sun the rod is 70% of
	2642	itself, so the sun still reads as where the beam is being taken, and the mirror by day is
	2643	lit by that same share. One constant for both, so the rod and the mirror it ends on agree.
	2644	Night is unchanged: the banked beam runs the whole length evenly.
	2645
	2646	What did not change, and should be settled. The light on the ground is still worked out
	2647	as if the sun took the whole beam (`direct`), so the world is exactly as bright as before.
	2648	Strictly, a sun taking 30% of the beam would light it less. That is on the backlog
	2649	(design.md §5): pick the split, then the light budget follows from it. The map, which does
	2650	not draw the rod, is untouched.
`7df218ae` 2026-09-28 Pavel Safronov	2651
	2652	## D-080 — A dead terrain worker no longer freezes the world
	2653
	2654	Reported (author, 2026-09-28): now and then, over several days, the page froze on
	2655	"Teaching Eastwall Water which way is down", with no clear console errors.
	2656
	2657	What it was. That label is the compile's last step, not a slow one: the whole compile
	2658	takes about 5 s, and that step 3 ms. After it, the page sent the compiled world to every
	2659	worker in the pool and waited for all of them, with no time limit. A worker that never
	2660	started never answered, so the wait never ended — and a comlink call to a dead worker fails
	2661	silently. The only trace was Firefox's warning "Attempting to create a Worker from an empty
	2662	source", which the author's screenshots showed a dozen times: Firefox had been handed an
	2663	empty script for a worker and started it with nothing to run. Why it sometimes gets an empty
	2664	script was not pinned down (twenty clean loads and eighteen with files changing mid-load all
	2665	worked); a page that made one worker silent on purpose reproduced the freeze exactly, on the
	2666	same label, with the same warning and nothing else.
	2667
	2668	What changed (`src/terrain/service.ts`). Before a world is handed out, every worker must
	2669	answer a `ping` within 15 s. One that does not is terminated and replaced once; if the
	2670	replacement is silent too it is dropped, and the pool carries on smaller. Handing over the
	2671	world has a 30 s deadline per worker on the same terms. Each of these says so with a
	2672	`console.error` naming the worker, and a worker's own `error` event is logged too, so the
	2673	next time this happens it is visible. Only an empty pool is fatal, and it says so.
	2674
	2675	Checked in Firefox with the same sabotage: one silent worker now costs about 15 s and a
	2676	console error; a silent worker with a silent replacement costs about 30 s and loads with five;
	2677	a normal load is unchanged.
`ab6ae94f` 2026-09-28 Pavel Safronov	2678
	2679	## D-081 — Trying other outsides: rock and piping together
	2680
	2681	Asked for (author, 2026-09-28): "i like the outside rock and pipes views, do you think
	2682	we can combine them? have rock show through, and have pipes overlaying in other spots."
	2683	Offered three readings: rock heaped on the plant as shielding, the plant built on top of
	2684	the rock, or Tower half dug out of an asteroid. Chosen by the author: shielding — "it
	2685	seems like a reasonable approach to creating Tower". A fifth Hull reading, **Rock and
	2686	piping**, an experiment like D-076 and D-077, not canon.
	2687
	2688	Why the two together, and not just either. The rock alone has nothing in it to give
	2689	scale — a crater could be sixteen metres or sixteen kilometres. The plating alone is
	2690	regular, which was D-076's complaint in the first place. Each fixes the other: a pipe
	2691	run that goes into a bank of rock and comes out ten kilometres on says how big both are.
	2692
	2693	One cover field decides where the rock is (`rockCover` in `src/views/rock-hull.ts`):
	2694	seeded noise over the hull, 0..1, rock wherever it is at least 0.5. Two scales, 80 km and
	2695	22 km, so it lies in banks and drifts tens of kilometres across with bare ground between.
	2696	It first covered about 43% of the hull; the author asked for "a slightly higher
	2697	concentration", and it is now about 60%.
	2698	- The rock is D-076's shell scaled by the cover, rising from 0.2 km to full height
	2699	over a narrow band of it so each bank ends in a scarp. It is 1.6 times D-076's height
	2700	here (author: "the rock feels like it can be taller, at least another 50%") — banks
	2701	with bare plating beside them need the height to read as heaps. The Rock reading on its
	2702	own is unchanged. A first try that ramped up slowly
	2703	left the edges a thin feathered skin that read as smoke, not rock. The cut is made per
	2704	pixel from the cover carried on each vertex, broken up by noise finer than the mesh, so
	2705	the shore of each bank is ragged rather than the mesh's 0.7 km grid.
	2706	- The plant — blocks and vents — goes only on bare ground, a little inside the edge.
	2707	Each plant room is centred on bare ground, so they crowd into the gaps rather than
	2708	thinning out. The bands and pipe runs are laid as before and go under the rock: that
	2709	is what makes the rock read as heaped on rather than the plant as built round it.
	2710	- The thruster pods are dug clear, 14 km round each, with a sloped edge: a sideways
	2711	flame runs along the hull, and would otherwise burn into the rock.
	2712
	2713	Built separately from the other two. The plant crowds differently and the rock is
	2714	dug out, so neither half is the same as the dress on its own; the combined one is its own
	2715	pair of meshes, built the first time it is chosen. With nothing under rock, the piping
	2716	builder makes the same draws from its seed as before, so the Piping reading is unchanged
	2717	(tested).
	2718
	2719	Canon. Nothing in §1 changes. The shielding reading gives D-076's rock a reason, and
	2720	rock over three fifths of the hull at 1.6 times the height is about the same mass as
	2721	D-076's shell — still over 10¹⁸ kg, so still a question for
	2722	§1.2 if this stays (design.md §5 Backlog). Real habitat studies do bury the hull under
	2723	regolith for shielding, but metres of it, not kilometres.
	2724
	2725	The plating's moiré (author, 2026-09-28: "the repetitive pattern in the pipes creates
	2726	moiree lines — circle patterns"). Seen from a glancing angle near the far cap on a
	2727	high-density screen: rings of interference on the plating. The ribbed panels were the
	2728	cause. Their ribs are rings round the hull 0.9 km apart, and they faded out only once a
	2729	pixel was most of a rib wide — so they were still drawn at full strength while one to
	2730	three pixels apart, which is exactly the band where a regular pattern beats against the
	2731	pixel grid. They were also faded by the pixel's overall size, when what matters for rings
	2732	is how fast z runs across a pixel, which at a glancing angle is several times more. The
	2733	ribs now fade by that rate, and are gone before they are three pixels apart. The same
	2734	reasoning is applied to the two other regular patterns in D-077's plating: the panel seams
	2735	fade out once a panel is a few pixels across, and the hard-edged grime stripes are
	2736	averaged away as they near a pixel's width. Close up, nothing changes. This is D-077's
	2737	plating, so the Piping reading gets the fix too.
`3e898b51` 2026-09-28 Pavel Safronov	2738
	2739	## D-082 — Eye height is a gauge in the view, marked with things of known height
	2740
	2741	Asked for (author, 2026-09-28): "please make the bottom-left panel collapsible", and
	2742	"move the eye height selector out of the right panel, and put it instead as a vertical
	2743	height, and indicate comparable heights when we reach those heights: average adult,
	2744	average home, eiffel tower, airplane cruising altitude, etc."
	2745
	2746	The gauge (`src/ui/eye-gauge.ts`) runs up the right-hand side of the interior view,
	2747	from 1.7 m at the bottom to 45 km at the top (48.4 km, the beam, since D-084), on the slider's own logarithmic scale — so a
	2748	person and a weather balloon both get room. Eleven marks, all real-world, so none needs
	2749	explaining: an adult (1.7 m), a two-storey house (8 m), a ten-storey block (30 m), the
	2750	Statue of Liberty (93 m), the Eiffel Tower (330 m), the Burj Khalifa (828 m), Mount Fuji
	2751	(3.8 km), Everest (8.8 km), an airliner cruising (11 km), Concorde cruising (18 km) and a
	2752	weather balloon (35 km). The gaps between the author's examples were filled so no stretch
	2753	of the scale is empty. Marks brighten as you pass them and turn accent-coloured when you
	2754	are level with one — within a tenth of its height. The readout by the thumb gives the
	2755	height, the local gravity (which the old slider's label carried), and the comparison:
	2756	"≈ the Eiffel Tower" when level, otherwise a multiple of the highest mark passed,
	2757	"2.0× the Eiffel Tower".
	2758
	2759	A widget of its own, not a rotated `<input type=range>`: the marks must sit exactly
	2760	level with the thumb, and a native range insets its thumb by an amount each browser picks.
	2761	It is still a slider to the keyboard (arrows, Page Up/Down, Home/End) and to screen
	2762	readers (`role=slider`, with the comparison as its value text). On a short view, a mark
	2763	too close to the one below it keeps its tick and drops its name — Everest and the airliner
	2764	are the pair that collide first. On a phone the names go altogether, and the readout by the
	2765	thumb says which mark you are near. It follows the view as the slider did: flying with E
	2766	and Q moves it, except while it is being dragged.
	2767
	2768	The standing readout folds to its title, "Where you stand", with a click on it; the
	2769	rest of it stays click-through, as every overlay is. The fold is remembered. On a phone it
	2770	starts folded, since open it covers a third of the view, and it now stops short of the
	2771	gauge.
	2772
	2773	## D-083 — No zoom inside; the trackpad swipe moves
	2774
	2775	Asked for (author, 2026-09-28): "the current interior docs state that two fingers
	2776	move, but i'm seeing that two fingers changes the zoom. can the zoom be hard-coded
	2777	internally, and can we get the motion working instead?"
	2778
	2779	Why it zoomed. D-059 made two touches move the observer and pinch the field of view,
	2780	and the wheel zoom. On a Mac trackpad, a two-finger swipe is not two touches: the browser
	2781	delivers it as the wheel. So the gesture the panel described went to the wheel handler and
	2782	zoomed, and the move D-059 built was reachable only on a touchscreen or with shift-drag.
	2783
	2784	The field of view is fixed at 60° (`FIELD_DEG` in `src/views/interior.ts`). Pinch no
	2785	longer zooms, the wheel no longer zooms, a saved field is no longer restored, and the
	2786	readout's `field` line is gone, since it could only ever say 60°. The scenic tour still
	2787	frames its own shots at its own fields; stopping it puts 60° back. A trackpad pinch
	2788	arrives as the wheel with ctrl held, and is swallowed rather than zooming the page.
	2789
	2790	The wheel moves. Not as a grab of the ground under the cursor, which was the first
	2791	try: it matched the touch drag exactly, but a trackpad's cursor sits wherever it was left.
	2792	Measured from standing, a 300 px swipe moved 2 m with the cursor near the bottom of the
	2793	screen and 99 m just under the horizon; and above the horizon every ray is held to the
	2794	same capped reach, so a swipe there barely moves at all. The travel depended on where the
	2795	cursor happened to be, which is no way to walk. So a pixel of swipe is a fixed fraction of a second of keyboard flight, which
	2796	already scales with height: 1/25 s, so a screen's swipe walks about 70 m from standing and
	2797	some 40 km from a kilometre up, wherever the cursor is and however the head is tilted. It
	2798	walks along the heading, in the same direction as the touch drag — the ground follows the
	2799	fingers, so swiping down goes ahead, as the page does under natural scrolling. No shift
	2800	boost, because browsers turn shift-wheel sideways.
	2801
	2802	Captions pass the wheel on. They take the pointer so they can be clicked, and one sits
	2803	on the skyline more often than not; a swipe over one did nothing until the caption layer
	2804	was given the same wheel handler.
	2805
	2806	Touch is unchanged apart from the pinch: two fingers still grab the ground and drag
	2807	it. Their centroid is what moves, so a pinch — fingers apart about a still centre — now
	2808	does nothing at all.
	2809
	2810	## D-084 — The top of the eye-height gauge is the beam
	2811
	2812	Asked for (author, 2026-09-28): "add 'sun beam' or whatever the technical name is to
	2813	the chart, since that's the top point we can go to." Canon's name is the beam (§1.3):
	2814	the column of concentrated sunlight on the axis, 3.18 km across, glowing where it runs
	2815	through the air.
	2816
	2817	It was not the top, so it is now. The ceiling was 45 km above the ground, 3.4 km short
	2818	of the beam's edge, which is 48.4 km above the datum (the 50 km radius less the beam's
	2819	1.59 km). The ceiling is now that edge: `EYE_MAX_M = RADIUS_M - BEAM_RADIUS_M` in
	2820	`src/ui/scales.ts`, derived from canon rather than typed in, and the gauge's last mark,
	2821	"the beam", sits at the top of the scale. The interior stops the eye 300 m short of the
	2822	edge over whatever ground is underfoot, so on high ground the ceiling comes lower and the
	2823	observer can never be inside the beam. From there it is 80° wide overhead and looking up
	2824	is white — a brilliantly sunlit cloud, 300 m off, as §1.3 describes it.
	2825
	2826	The gauge's gravity now counts the ground, as the standing readout's always did;
	2827	the old slider's label measured from the ground up and read 0.044 g at the top where the
	2828	readout said 0.038 g.
	2829
`99a2821b` 2026-09-29 Pavel Safronov	2830
	2831	## D-085 — The scenic tour flies the world's flyers, inside and out
	2832
	2833	Asked for (author, 2026-09-28): automatic flight, to show Tower off inside and out —
	2834	a flight plan put together at random from components, a camera that points at interesting
	2835	things as it goes, about fifteen seconds before changing from one flyer to another, and
	2836	every point of view belonging to something in the world: paragliders, planes, ships,
	2837	spacecraft, trains. Planned before it was built, and the plan was agreed with these answers:
	2838
	2839	- A craft exists for its leg. Nothing is simulated as traffic when nobody is watching;
	2840	a paraglider is there for the fifteen seconds the camera rides it. The ships, which were
	2841	already in the world (D-067), are followed as they are.
	2842	- Legs last 12 to 18 seconds, each its own length, and fade through black between.
	2843	- Airships, yes: a few ordinary ones low down, and a few slow giants near the axis.
	2844	- It replaces D-068's tour, whose nine thirty-second scenes are now legs among the rest.
	2845
	2846	Built. `world-generation.md` §19 has the detail.
	2847
	2848	A leg is a craft on a path, and a camera on the craft. Twenty-six inside — a helicopter
	2849	round the Long Water's fall and a paraglider across its face, ships in fair weather and in
	2850	a squall, a paraglider on a summit's flank, a glider in a thermal over the Dry Reach, light
	2851	aircraft down rivers and through The Narrows, a seaplane across The Deep, an airliner round
	2852	the Girdle, a giant airship climbing out of the Capital and another circling the sun, a
	2853	night flight toward the far cap, trains on both railways (D-086) — and nine outside, all
	2854	spacecraft: skimming the hull, circling it against its spin, docking on it, looking in
	2855	through the far cap, crossing the sun cap, over the collector's rim, down the beam, pulling
	2856	away, and Tower, the collector and the sun in one frame. The camera rides in the craft, or
	2857	behind it on the far side from what it is looking at, so the craft and its subject share
	2858	the frame.
	2859
	2860	The camera is pointed by a director, not by hand. Everything worth a look — summits,
	2861	falls, lakes, the Capital, the Tall Wood, the sun, both caps, ships — is scored from the
	2862	craft by how big it looks, how clearly it can be seen through the air (the same integral of
	2863	the density profile the renderer hazes with), whether the ground hides it, and whether a
	2864	pilot could look that way. The leg's own subject wins most arguments; anything looked at for
	2865	long starts to bore it; turns are eased over three seconds and never swing faster than
	2866	0.7 rad/s. So the tour does not stare at haze, and the far side of the world gets looked at,
	2867	because §1.8 says it is the clearest thing in the sky.
	2868
	2869	The deal is a deck. Every leg is flown once before any is flown twice, no two in a row
	2870	are the same kind of craft, two legs about the same subject are kept apart where they can
	2871	be, and a new deck never opens on the craft the last one closed on.
	2872
	2873	Some legs are time-lapses, and say so. A giant airship takes minutes to climb ten
	2874	kilometres and circles the sun at 28 m/s, so those legs run the world's clock ×20 to ×150,
	2875	and the caption says "time ×150". Everything else is real time.
	2876
	2877	What the numbers say, the captions say. Moving round the world changes what you weigh
	2878	(`canon/motion.ts` `apparentGravity`): the Girdle's airliner, flying antispinward at 230 m/s
	2879	nine kilometres up, weighs about a third of what it would at rest, and a train on the Uplands
	2880	Ring is 31% heavier one way round and 27% lighter the other. Every circumferential journey in
	2881	Tower has a handedness, as every slope does for a skier (§1.11).
	2882
	2883	Outside, the hull turns at the leg's rate — real time — whatever the Speed slider says,
	2884	so a craft skimming it sees 700 m/s of hull go by and not a blur. A leg flies in one of two
	2885	frames: space, where the hull wheels past, or the hull's own, where it holds still and the
	2886	stars wheel, which is what docking looks like. The near plane follows the nearest thing, so
	2887	a craft 400 m off is not clipped.
	2888
	2889	## D-086 — Two railways: the Axial Line and the Uplands Ring
	2890
	2891	Asked for (author, 2026-09-28): trains in the tour, with the lines "the ones you find
	2892	reasonable". There were no railways in the world, so there is now a `railway` kind in
	2893	world.json — a line, with a line speed — and two of them, chosen and laid by Claude:
	2894
	2895	- The Axial Line, 378 km, 70 m/s: from Sisters Station across the Uplands, through **The
	2896	Narrows** beside The Long Water — the only way along the world goes through the only gap
	2897	in the ridges, for rails as for rivers — down The Deep's shore past the Capital, and through
	2898	Greatwood to the shore of the Deepward Sea.
	2899	- The Uplands Ring, 314 km, 100 m/s: the one line round the world rather than along it,
	2900	at z 140 km, where the Uplands are level at about 1,000 m the whole way round and no water
	2901	stands. It crosses the four rivers off the range and meets the Axial Line near θ −0.06.
	2902
	2903	The track keeps its own grade, over the ground. `terrain/rails.ts` lays the deck at the
	2904	highest ground within a kilometre either side along the line, eased over 900 m, so it is never
	2905	below the ground and a train never runs through a hill. Tower does not cut the ground for it.
	2906	The cost of that is honesty about viaducts: the deck stands 10–20 m up on piers over most of
	2907	both lines, and up to 260 m where the Ring crosses a gorge. The steepest grade is 7%, where
	2908	the Axial Line comes down off the plateau into Greatwood; all but 23 km of it is under 3%,
	2909	and the Ring never passes 3.7%. A kilometre window was measured against 400 m and 1.5 km:
	2910	shorter made grades of 17–28%, longer made the whole line a viaduct.
	2911
	2912	The map draws them as lines in their own colour, and the interior draws the deck, its
	2913	girders and a pier every 100 m wherever it stands clear of the ground.
	2914
	2915	## D-087 — Where spacecraft dock, and where aircraft may not go
	2916
	2917	Canon, from the author (2026-09-28), answering questions the tour raised:
	2918
	2919	- Spacecraft can dock anywhere on the outer shell. A docked ship turns with the hull at
	2920	700 m/s, and the hull holds it on that circle: it hangs from its clamps, pressed outward at
	2921	1.02 g (ω² × 51 km), its feet toward the hull and its floor toward space. The tour's docking
	2922	leg matches the spin and sets a ship down that way.
	2923	- Aircraft are not allowed into the beam, but they can fly around it — and round the sun,
	2924	which is in the beam (§1.3.1). Every craft in the tour keeps at least 3 km clear of it,
	2925	inside and out, and the tests check that.
	2926
	2927	Consequences worked out with it, not canon until the author says so: the giant airships
	2928	live near the axis because only there do they have to. At 44 km up the air is about a
	2929	twentieth of the floor's density and weighs about a tenth, and what a gasbag lifts per cubic
	2930	metre depends on the density of the air it displaces, not on gravity — so an airship there
	2931	needs some twenty times the volume for the same load. Hence giants, 420 m long, and slow.
`42b4554e` 2026-09-29 Pavel Safronov	2932
	2933	## D-088 — Tower has a moon
	2934
	2935	Asked for (author, 2026-09-28): "a 'moon' will travel from night cap to the sun cap over
	2936	the course of the night. The brightness of the moon will be based on the moon phases, so this
	2937	will give Tower a moon of sorts." The night cap is the far cap: night starts there, when
	2938	the sun goes out at it (the cap names are in §1.4).
	2939
	2940	Built. design.md §1.3.5 has the canon.
	2941
	2942	What it is. A second gathering of the swarm on the banked beam, by night, the size of the
	2943	sun. It lights at the far cap at dusk — `SUN_RAMP_H`, the same quarter hour the sun takes to go
	2944	out — runs at a steady pace to the sun cap, and goes out there at sunrise as the sun lights. The
	2945	sun crosses the world one way by day and the moon the other way by night, so there is always one
	2946	of them on the beam except at new moon. With the beam shut down (the Dark setting) there is no
	2947	moon either.
	2948
	2949	Its phase is Earth's Moon's, worked out for the moment on the clock (`world/celestial.ts`
	2950	`moonPhase`, Meeus's low-precision phase angle, checked against the almanac's new, full and
`de3d60d6` 2026-09-29 Pavel Safronov	2951	quarter moons). Day 0's night, 23 September 2026, has an 89% waxing moon; it is full on the 26th
	2952	in Pisa (the 27th in Sydney, which Tower kept when this was written) and new on 11 October. The disc is drawn with its terminator, lit toward the far cap while it
`42b4554e` 2026-09-29 Pavel Safronov	2953	waxes and toward the sun cap while it wanes — where the sun set and where it will rise, as on
	2954	Earth.
	2955
	2956	How bright, and why that. The brightness had to be chosen, because the honest one would be
	2957	invisible: Earth's full moon lights the ground at about 0.25 lux, and Tower's banked night is
	2958	already some 150 times that. So Tower's moon is sized to its own night: at full it throws as much
	2959	light on the floor as the whole banked rod (`MOON_AT_FULL_OF_NIGHT = 1`), and a moonlit night is
	2960	up to 0.64 W/m² against a moonless 0.32. It is lit from where the moon is, not from the whole rod,
	2961	so it has a direction: slopes facing it are brighter, and seas carry a path of light toward it.
	2962
	2963	Everywhere the sun is, the moon is. The panel's time section says where it is and how lit,
	2964	the interior's readout gives its height and distance as it does the sun's, the map draws a pale
	2965	line where it is overhead, the exterior sees it in through the windows, and the scenic tour's
	2966	director will look at it by night. The tour has one new leg for it: a seaplane crossing The Deep by
	2967	moonlight, on the first night of the month with the moon at least 60% lit.
	2968
	2969	A canon question it raises, not answered here: whether the moon takes its light from the
	2970	beam as the sun does — the rod dimmer past it — or is fed by Tower lifting the banked beam, as
	2971	written above. The second is what is drawn.
`de3d60d6` 2026-09-29 Pavel Safronov	2972
	2973	## D-089 — Tower keeps Pisa's days
	2974
	2975	Amends D-074.
	2976
	2977	Decided by the author, 2026-09-28: "for reasons, Tower has actually decided to set its
	2978	internal 'time' to match that of Pisa, Italy." Built.
	2979
	2980	What changed. The `calendar` in `content/world.json` is Pisa's: `Europe/Rome`, 43.7228° N,
	2981	10.4017° E. Tower's clock is Pisa's local time, Italian daylight saving included (CEST, +2, until
	2982	25 October 2026; CET, +1, after), the app opens at the time in Pisa now, and the sun rises and
	2983	sets with Pisa's sunrises and sunsets. The fallback calendar in `src/world/schema.ts`, used when
	2984	the world file's is missing or unusable, is Pisa's too.
	2985
	2986	The clock and the days stay together. D-074 makes Tower keep one city's clock and its
	2987	sunrises and sunsets; changing the city changes both. So Tower's seasons turned over: day 0 —
	2988	still 23 September 2026 — is now the northern autumn equinox, and the days shorten toward
	2989	December instead of lengthening. The day runs 12 h 11 min at the equinox (the sun at 44 km/h),
	2990	8 h 55 min at midwinter (61 km/h) and 15 h 27 min at midsummer (35 km/h), against Sydney's
	2991	9 h 54 min and 14 h 24 min.
	2992
	2993	Tests. Sydney's published sunrise tables still check the solar maths, on a Sydney calendar
	2994	of their own; Pisa's day lengths, daylight saving and clock have tests of their own.
	2995
	2996	One knock-on in the tour. The ship leg starts in the morning over its sea, and the morning
	2997	moved; at the new hour the director looked up at the sun, which put the chase camera below the
	2998	ship, and the sea pushed it back up beside the ship, looking past it. A ship's chase camera now
	2999	looks no more than 20° up (`maxPitch` 0.35 in `shipLeg`).
`9f91e995` 2026-09-29 Pavel Safronov	3000
	3001	## D-090 — The far end is cold: ice at the far cap, a frosted window, and Tower sets the hull's heat loss
	3002
	3003	Amends §1.9.4, which said nothing breaks Tower's thermal uniformity.
	3004
	3005	Worked out with the author, 2026-09-28, starting from a question: if Tower sent no light
	3006	to the sea by the far cap, would it freeze? Canon only; nothing drawn yet.
	3007
	3008	What the question found. Two things canon had missed.
	3009
	3010	- The far cap is a radiator. D-045 glazed the whole cap, and glass is opaque to thermal
	3011	infrared, so the window absorbs the world's heat and its outer face radiates it to space:
	3012	80–250 W/m² for a pane 0.1–1 m thick, 0.65–1.9 TW over the cap. That is a cold wall the
	3013	size of the cap, and it contradicted §1.9.4's "nothing breaks the thermal uniformity". Air
	3014	cooled against it slides down the cap and goes sunward over the Deepward Sea as a steady
	3015	cold wind.
	3016	- The hull cannot lose heat by itself. 1 km of rock conducting 320 W/m² would need about
	3017	100,000 K across it. So the heat reaches the skin because Tower sends it there.
	3018
	3019	Decided by the author:
	3020
	3021	1. Tower controls the hull's heat loss, region by region, up to about 5 TW — a tenth of
	3022	the beam. "Not infinite, but some control." Light makes the weather; heat rejection is how
	3023	Tower keeps the books.
	3024	2. The far end is kept dim and freezes in bands, sunward to cap: open water z 496–512 km,
	3025	pack ice 512–525, fast ice frozen to the cap 525–540, then the window.
	3026	3. The rod stays as D-079 has it, running past the sun to the cap. So the fast ice gets
	3027	dusk by day, tens of W/m², not darkness — "no big deal if we don't have complete
	3028	darkness". Stars by day, which a truly dark end would have given, are given up with it.
	3029	4. The sun still goes to the cap (D-074), so it goes out over the ice each evening.
	3030	5. **The far cap is single-glazed and frosts. Tower clears it one pane per turn of the
	3031	world**, 7 min 29 s, in a random order that is reshuffled every loop but always the same
	3032	for a given loop. 282 panes (216 whole, 66 cut by the rim) make a loop of 35.1 hours.
	3033
	3034	Consequences worked out, now in §1.9.6:
	3035
	3036	- The whole Deepward Sea is at about freezing, not only the ice. §1.9.1 means there is no
	3037	Coriolis to hold cold water in place; a 1 K difference drives an exchange of ~0.45 m/s,
	3038	which mixes the 44 km sea to within about a kelvin. No Tower sea can be frozen in one corner.
	3039	- The air pays the window, not the sea. Paid by the sea, the open water would need
	3040	140–400 W/m², daylight. So the weather is one overturning cell — warm air deepward aloft,
	3041	the window wind sunward below — and the strength of that cell sets the ice edge.
	3042	- The hull is turned down under the sea, or the sea would freeze from its bed up. The far
	3043	end takes about 2 TW of the 5.
	3044	- The fast ice holds: ~1,500 N per metre of wind against ~3 MN of ice, and Tower has no
	3045	tides. Storms on the weather plan (D-067) are what could break it out.
	3046	- The Deepward Sea's ninety ships keep the 16 km of open water.
	3047
	3048	Numbers that are rough: the window wind (5–15 m/s) and the frost rate (millimetres a day,
	3049	to a factor of ten). The frost rate decides how many panes are clear at once; one in ten is a
	3050	guess.
	3051
	3052	The sun cap is double-glazed (author, 2026-09-29). Single glazing there would have done
	3053	all the same things — radiated, frosted, and sent a cold wind onto the Sunward Sea. Double, its
	3054	inner pane stays near the air's temperature, so the Sunward Sea stays temperate and the far end
	3055	is the world's only cold place.
`e18dad04` 2026-09-30 Pavel Safronov	3056
	3057	## D-091 — Drawing the far end: sea ice, the dim, and the frosted panes
	3058
	3059	Builds D-090. Asked for by the author, 2026-09-29: "please commit and implement". Also
	3060	recorded here: the author chose double glazing for the sun cap (D-090), so the Sunward
	3061	Sea has no ice and needs nothing drawn.
	3062
	3063	The ice is in the world file. A sea may carry `ice: { edgeM, fastM }` — the ice edge and
	3064	where the fast ice begins, as z — and the Deepward Sea does: 512 and 525 km. The fast ice runs
	3065	on from `fastM` to the end of the sea beyond it, whichever end that is. It is part of the
	3066	terrain signature, so changing it rebuilds the tiles, as it must: every water sample carries
	3067	its ice cover (`terrain/ice.ts`, `field.ts seaIce`).
	3068
	3069	- Cover. None on the open water; the pack closes from 0.3 at its edge to 0.95 at the fast
	3070	ice; 1 on the fast ice. The edge wanders 1.5 km either way by noise and the fast ice's 400 m,
	3071	so neither runs ruler-straight round the world.
	3072	- Floes. The water shader breaks the pack into Voronoi cells about 110 m across — straight
	3073	edges at odd angles, which is how ice breaks — each ice or open by its own hash against the
	3074	cover. Floes stop short of open neighbours, opening leads up to 12 m wide where the pack is
	3075	loose, and meet ice neighbours in a dark crack. The fast ice is one sheet, cracked in cells
	3076	of 660 m. Floes are big enough to need no surface reference: absolute metres are good to
	3077	centimetres in float32, the grid around is a whole number of cells, and the hash is a
	3078	permutation mod 289, exact in float. Far off, the sea whitens by its share of ice.
	3079	- Ice stills the sea. Waves die away under the pack and are gone under the fast ice.
	3080	- Ships keep to the open water, their lanes 2 km short of the ice edge.
	3081	- The map whitens the water by the cover, and the exterior's x-ray, which is the map,
	3082	with it.
	3083
	3084	The dim. Tower's aimed light — the sun's and the moon's — falls from the open water's
	3085	shore (the sea's end away from the ice) to 0.35 at the ice edge, 0.1 at the fast ice, and
	3086	nothing on it. The rod is never aimed, so by night nothing changes but the moon, and by day
	3087	the fast ice keeps the rod's glow, taken as a twentieth of the sun's light (`ROD_GLOW_BY_DAY`
	3088	— tens of W/m² against about a thousand). Each lit surface is multiplied by its share **put
	3089	through the same 0.35 power the exposure gives the light dial**: an eye standing in the dim
	3090	has adapted to it, and at a straight twentieth the fast ice drew as a black hole. So it reads
	3091	as dusk, a third as bright as elsewhere. The stars go by the light where the eye is, so from
	3092	the fast ice the brightest of them show at dusk.
	3093
	3094	The frost (`world/frost.ts`). The window's 282 panes are named by lattice cell and half,
	3095	the same triangles the mullions draw. Tower clears one per turn of the world (7 min 29 s) in
	3096	an order shuffled by the world's seed and the loop's number, reshuffled every 35.1-hour loop;
	3097	a cleared pane frosts over again as 1 − e^(−age / 10 h). Ten hours is chosen to leave about one
	3098	pane in ten nearly clear, §1.9.6's guess — the frost rate is rough to a factor of ten, and this
	3099	is the knob for it. The pane being cleared thaws across its turn. The clock writes the frost
	3100	into a 32 × 32 table the far cap's shader reads, so both views, inside and out, show the same
	3101	panes clear at the same moment. Frost is feathered rather than a grey sheet, and lit by the
	3102	light at the cap, which is the fast ice's dusk.
	3103
	3104	Left open (design.md §5): the sun is drawn as bright from the fast ice as from anywhere,
	3105	though Tower aims none of its light there.
	3106
	3107	Tests (`test/far-end.test.ts`): the bands and their cover, a sea with its ice the other
	3108	way round, ships clear of the ice, the dim by day and by night, 282 panes (216 whole) found
	3109	from points inside them, a loop of 35.1 hours, every pane cleared once a loop in a reshuffled
	3110	order, a pane thawing across its turn, no jump between loops, and about one pane in ten clear.
	3111
	3112	## D-092 — A dead terrain worker is replaced at once, not after 15 s
	3113
	3114	Amends D-080.
	3115
	3116	Reported (author, 2026-09-29): "every so often, the page freezes on the loading screen
	3117	and there are errors in the console. a reload fixes it, or just waiting." The console showed
	3118	Firefox's "Attempting to create a Worker from an empty source" many times, and three workers
	3119	failing with an `error` event.
	3120
	3121	What it was. D-080 made a silent worker survivable but slow: it was given 15 s to answer
	3122	its ping before being replaced, so a load with a dead worker sat on "Reading the plans" for
	3123	15 s or more — the "just waiting". Workers die two ways, and both were waiting out that
	3124	deadline:
	3125
	3126	- A script that throws or fails to load fires the worker's `error` event. That was only
	3127	logged; the ping, which comlink never settles for a dead worker, went on waiting.
	3128	- An empty script — the "empty source" — fires nothing. Firefox starts it, it runs
	3129	nothing, and it never answers.
	3130
	3131	Why Firefox is handed an empty or broken script is still not pinned down. It was not
	3132	reproduced by fifteen plain reloads, nor by changing files mid-load. The likeliest cause is
	3133	the dev server answering while it rebuilds after an edit: the author reported it while files
	3134	were being edited under an open page. So this is a cure, not a prevention.
	3135
	3136	What changed (`src/terrain/service.ts`):
	3137
	3138	- The start-up ping is raced against the worker's `error` event, so a script that failed is
	3139	replaced the moment it says so.
	3140	- A silent worker is judged against its siblings. They load the same modules at the same
	3141	time, so once one has answered, a worker still silent after that sibling's own load time
	3142	again, and never less than 3 s, is taken to be empty rather than slow. Only a pool with no
	3143	worker alive at all still waits out the 15 s.
	3144	- A dead worker is replaced up to four times, 0.4 s, 0.8 s and 1.2 s apart, rather than once.
	3145	Each failure is a `console.warn` naming the worker, and a worker that came back says which
	3146	try it started on.
	3147
	3148	Checked in Firefox, with the worker's script replaced by Playwright for the first few
	3149	requests:
	3150
	3151	\| Sabotage \| Before \| After \|
	3152	\|---\|---\|---\|
	3153	\| none \| 7.4 s \| 7.4 s \|
	3154	\| three scripts that throw \| 22.8 s \| 7.7 s — no delay \|
	3155	\| three empty scripts \| 22.3 s \| 11.1 s \|
	3156	\| all six empty, twice over \| — \| 38.6 s, and all six start on the third try \|
	3157
	3158	## D-093 — Ice mountains at the far cap: the Cap Range
	3159
	3160	Asked for (author, 2026-09-29): "the ice at the far cap should slope up, since the
	3161	snow/ice would build up closer to the window. i'm imagining ice mountains that go a bit up
	3162	the window, even."
	3163
	3164	Why ice alone was not enough. The ice is thickest against the window — snow off the
	3165	window wind and the meltwater of every cleared pane (D-090) come down at the cap's foot — so
	3166	it should rise toward the glass. But grounded ice spreads under its own weight until the
	3167	stress at its bed is what the bed will take, which caps it at the plastic profile of an ice
	3168	sheet, H² = H0² + (2τ/ρg)·x. With τ = 150 kPa, the top of what ice sheets show, it has to
	3169	fill 420 m of sea before any of it stands out, and over 8 km it reached about 280 m above the
	3170	water: built and drawn, it read from the pack as a pale strip under 1° high at the foot of a
	3171	50 km window. Three ways on were offered — keep that, raise the sea bed (500–700 m, still a
	3172	kerb), or a rock range under the ice. Chosen by the author: the range.
	3173
	3174	What it is. The Cap Range (provisional name): rock Tower raised at the foot of the far
	3175	cap for the ice to stand on, as it raised the Uplands (D-064). It is part of the Deepward
	3176	Sea's ice in the world file — `ice.range: { fromM: 524000, heightM: 2400 }` — and a label
	3177	feature, a region band from 528.5 km to the cap, carries its name.
	3178
	3179	- Rock rises from the sea bed at 524 km to about 2,400 m at the cap, as t^1.6 so it
	3180	steepens toward the window; ±30% round the ring, in massifs and saddles 30 km across; and
	3181	ridged by up to 0.7 of its height, so it is a range and not a ramp.
	3182	- Ice mantles it from where the rock leaves the water (about 528.5 km), in a front of a
	3183	few tens of metres over the fast ice, thickening as an ice sheet does to 350 m. It drapes
	3184	the ridges (60% of their relief) rather than burying them level; where a crest stands out
	3185	of it, it is bare rock.
	3186	- At the window it stands 2.2 to 3.3 km up the glass. The far cap's shader buries the
	3187	glass and frame in ice to the height the ground stands at the cap, read off the terrain
	3188	into a strip round the ring; the kilometre of bore between the world's end and the glass
	3189	is ice too.
	3190	- The sea no longer meets the cap. The fast ice now runs 525 to 528.5 km and ends
	3191	against the range. The Deepward Sea's band in the world file still runs to 540 km: the
	3192	range stands in it.
	3193
	3194	Tests (`test/far-end.test.ts`): the rock from bed to height, steepening; the ice's margin
	3195	at the waterline and its mantle thickening to 350 m; and on the compiled world, all round the
	3196	ring, dry ground 1.5–4 km high at the cap that varies, fast ice at 526 km, open water at
	3197	505. `test/terrain.test.ts` counts the new feature and probes the Deepward Sea short of it.
`ec1a6075` 2026-09-30 Pavel Safronov	3198
	3199	## D-094 — The tour shows how long a leg has left: its caption's border fills in
	3200
	3201	Asked for (author, 2026-09-29): "for the tour, it would be nice to have an indicator of
	3202	'we've been here 11 second, 3 more until transition', but preferrably in some intuitive way,
	3203	like the border of the tour text filling out or collapsing or something."
	3204
	3205	What it is (`views/tour-player.ts`, `style.css .tour-ring`). The caption's border fills in
	3206	clockwise from its top-left corner as the leg runs, in the accent colour over the caption's own
	3207	faint border, and closes as the leg starts to fade to the next. The next leg's caption starts
	3208	empty. No numbers: a clock read at a glance, the way a progress ring is. It is drawn as an SVG
	3209	rectangle over the caption's border, its dash measured in pixels of the border's own length,
	3210	which follows the caption's size as its text changes. Both tours, inside and out, have it.
`abc0aad0` 2026-10-01 Pavel Safronov	3211
	3212	## D-095 — The Fields: farmland where each crop is lit on its own day
	3213
	3214	Asked for (author, 2026-09-30): "please add farm fields to the model. these should be
	3215	realistic fields, in their perfect bloom, so colorful. and this will be a new thing, but these
	3216	fields will receive 100% sunlight, even at night."
	3217
	3218	The conflict, raised before building. Tower's night is made by banking the beam, because
	3219	the rod's glow cannot be aimed away (§1.3.2). Fields lit at full sun all night need the beam
	3220	up, and the whole world gets the rod's glow with it: 250 km² of fields would add about
	3221	0.5 W/m² everywhere, all night; 1,000 km², about 2 W/m², a permanent twilight. **The author
	3222	chose 250 km²**, at z 430–496 km (the unauthored belt by the far end) — and, asked how the
	3223	flowers stay in bloom, changed the brief: "flowers get darkness, and other things (corn,
	3224	potatoes, tomatoes, etc.) will have appropriate timings. i thought 100% daylight would be
	3225	helpful for plants, but if it's not, adjust as appropriate, since Tower can provide whatever
	3226	lighting."
	3227
	3228	What it is.
	3229
	3230	- The Fields (provisional name), a new feature kind, `farmland`: 250 km² at z 455–480 km,
	3231	10 km around, where the Axial Line ends. Its edge wanders by noise; it clears forest, and
	3232	stays off water, sand, rock and broken ground.
	3233	- The fields (`terrain/farms.ts`) are cells 480 m along the axis by about 360 m around —
	3234	the ring a whole number of them — some split in two, with 6 m tracks between. Each is one of
	3235	sixteen crops by a weighted hash, three-quarters of them flowers: tulips red, yellow, pink
	3236	and white, lavender, sunflowers, rapeseed, poppies, flax, phacelia, buckwheat; and maize,
	3237	potatoes, tomatoes, wheat, lettuce. Each crop has its bloom, fruit or ripe-head colour, its
	3238	leaf colour, how much of it shows, and its row spacing. The layout is a pure function of
	3239	position, written out in GLSL too, so the map and both views agree field by field.
	3240	- Each crop's day follows its plant: 18 h for rapeseed, flax and wheat; 16 h for lavender,
	3241	poppies, phacelia and tomatoes; 14 h for tulips, sunflowers, maize and lettuce; 12 h for
	3242	buckwheat and potatoes. Every crop gets at least six hours of dark. Its day starts at sunrise
	3243	and is lit as at noon throughout — on into the night where it is longer than the sun's;
	3244	where it is shorter, cut to the rod's glow, since Tower cannot make a field dark by day.
	3245	- Drawn as rows over soil with flowers or fruit among the leaves, clumps and gaps close in,
	3246	and the crop's average colour once rows are under a pixel. Lit fields are drawn as at noon
	3247	at the dial's brightness whatever the hour, so after sunset they blaze in a dark world and
	3248	go out crop by crop. The map colours each field by its crop.
	3249
	3250	The night it costs. At the equinox about 91% of the fields are still lit at sunset —
	3251	220 GW, 0.4% of the beam — adding about 0.4 W/m² of rod glow everywhere, falling to 0.2 at
	3252	two hours, 0.1 at four, and nothing at six. This is canon but not drawn: the world's
	3253	night light in the app does not yet include it (design.md §5).
	3254
	3255	Tests (`test/farms.test.ts`): the ring closes on whole cells; the same field at the same
	3256	place and across the seam; tracks between cells; every crop planted, flowers most; a long-day
	3257	crop lit into the night then dark, a short-day crop dimmed before sunset; six hours of dark for
	3258	every crop; and on the compiled world, farmland inside The Fields, cleared of trees, and none
	3259	outside it.
	3260
	3261	## D-096 — The farms spread round the world, and the world holds still while it is made
	3262
	3263	Amends D-095.
	3264
	3265	Asked for (author, 2026-09-30): "the fields are concentrated in a single region. let's
	3266	split them up and rearrange them around the world." And, separately: "don't run the world
	3267	while the building process is going on. the 'sun' moves while the world is still building,
	3268	and that seems weird. have everything be paused until the build process is all finished."
	3269
	3270	Nine farms instead of one. The 250 km² D-095 chose for the night stays the total; it is
	3271	now nine farms of about 28 km² each, rounded rectangles 6.2 km along the axis and 4.8 km
	3272	around, each its own `farmland` feature with a provisional name:
	3273
	3274	\| Farm \| θ \| z \| Where \|
	3275	\|---\|---\|---\|---\|
	3276	\| Ringside Fields \| 1.7 \| 129 km \| the Uplands, beside the Uplands Ring \|
	3277	\| Far Ringside Fields \| 4.3 \| 131 km \| the Uplands, the other side of the world \|
	3278	\| Greatwood Clearing \| 2.6 \| 240 km \| Greatwood, near its sunward edge \|
	3279	\| Capital Fields \| 5.4 \| 262 km \| Greatwood, 44 km antispinward of the Capital \|
	3280	\| Deepwood Fields \| 3.8 \| 320 km \| Greatwood, opposite The Deep \|
	3281	\| Greatwood Fields \| 2.0 \| 380 km \| Greatwood's deepward half \|
	3282	\| Railhead Fields \| 0.0 \| 470 km \| where the Axial Line ends \|
	3283	\| Far Fields \| 3.4 \| 455 km \| before the far end, opposite the railhead \|
	3284	\| Last Fields \| 5.0 \| 470 km \| the nearest to the Deepward Sea and the ice \|
	3285
	3286	Each site was checked on the compiled world for open ground — no water, desert, bare rock
	3287	or range — before it was placed. The Uplands are the roughest: part of each Uplands farm
	3288	falls on broken ground and is left unploughed, as the farmland's own rules leave it. The
	3289	ones in Greatwood clear their ground of trees, so they read as clearings.
	3290
	3291	The world holds still while it is made. The clock (`world/clock.ts`) gains `held`: it
	3292	keeps whether it is running — the panel still says so — but no time passes. The app holds it
	3293	from the moment it starts until the status dialog has gone, or, for a compile quick enough
	3294	never to show the dialog, until the compile is done; and again whenever the world is
	3295	recompiled. Both views read it too: while it is held their frame time is zero, so the hull
	3296	does not spin, the water does not move, the sky through the far cap does not turn and a tour
	3297	does not advance.
	3298
	3299	Tests: nine farms sharing 240–260 km², spread more than 4 rad round the ring and 300 km
	3300	down the axis, each farmland at its heart and cleared of trees.
	3301
	3302	## D-097 — Farms grow over the land, and their fields are irregular
	3303
	3304	Amends D-095 and D-096.
	3305
	3306	Asked for (author, 2026-09-30): "the current rectangular regions (both the general 9 and
	3307	the farms inside them) are very orderly, can we do some sort of natural growing pattern that
	3308	results in more random-looking patterns?"
	3309
	3310	A farm grows (`terrain/farms.ts growFarm`, compiled last, over the land as it is). A farm
	3311	in the world file is now a point and an area — the nine of D-096 keep their places, each
	3312	`areaKm2: 27.8`. At compile time it grows from the point over a 200 m lattice of patches, always
	3313	taking next the patch on its edge with the lowest key: half how far it is from the seed (over
	3314	the radius a disc of that area would have), half luck, by hash. It never takes ground that is
	3315	wet, sand, rock, a range (over 150 m of relief), ice, or steeper than 12%, so it goes round a
	3316	river or a hill instead of over it, until it has its area. Near-and-lucky is what makes it a
	3317	blob with lobes and bays — a test holds it to at least 1.3 times the edge of a disc the same
	3318	size, all in one piece. A field belongs to the farm if the patch under the point it was laid
	3319	out from does, so the farm's edge runs along its fields' edges and no field is cut in half;
	3320	the ground shader sharpens the vertex-interpolated weight so the edge does not smear. A farm
	3321	may still be drawn as a polygon or band, which is its outline as before.
	3322
	3323	The fields are irregular (`fieldAt`, and its GLSL twin). In place of D-095's grid:
	3324
	3325	- Blocks about 1.5 km across: a jittered Voronoi of the surface, a whole number of them
	3326	round the ring, with 6 m farm tracks along their edges.
	3327	- Each block on its own terms: its fields run at its own angle, are its own size (220 to
	3328	440 m) and drawn out long by its own stretch (1.3 to 2.2), as land cleared at different times
	3329	by different hands is.
	3330	- Fields within a block: a second Voronoi, in the block's frame — straight edges at odd
	3331	angles — with grass margins between. A third are cut into two to four long strips.
	3332	- Rows run along each field's length. Everything inside a block is measured from the block's
	3333	centre, so the pattern has no seam round the ring and its numbers stay small for float32.
	3334
	3335	Tests (`test/farms.test.ts`): no seam; fields of many sizes (the largest fifth at least
	3336	twice the smallest fifth); tracks along the blocks and only there; every crop, flowers most;
	3337	a grown farm taking its area in one piece, ragged, and going round ground that does not fit;
	3338	and in the world, nine farms grown to 240–260 km² between them, farmland and no trees about
	3339	each seed.
`7a5edd5f` 2026-10-02 Pavel Safronov	3340
	3341	## D-098 — On a phone, the height readout fades when the height is still
	3342
	3343	Asked for (author, 2026-10-01): "the height label takes up a lot of space on mobile, can the
	3344	big rectangle box fade away or something?"
	3345
	3346	What it does (`ui/eye-gauge.ts wake`, `style.css`). Below 720 px the eye-height readout —
	3347	the boxed height, gravity and comparison beside the gauge's thumb — fades out over 0.6 s once
	3348	the height has been still for two seconds, and comes back at once when the height changes by
	3349	more than half a percent (flying, a tour, the gauge) or the gauge is touched; it stays while the
	3350	gauge is held. The view re-telling the same height every frame does not wake it. The thumb and
	3351	the ticks stay, so the height is still read off the scale. Wider screens are unchanged: there the
	3352	box is not in the way. Screen readers are unaffected — the gauge's value text is always current.
`3cc2b5bd` 2026-10-04 Pavel Safronov	3353
	3354	## D-099 — Photographs: a view saved as numbers, rendered again on demand
	3355
	3356	Asked for (author, 2026-10-03): "i've been finding interesting photo opportunities, like
	3357	lining things up at a particular time, etc. and i've started to use these photos as
	3358	illustrations, after messing with them in lightroom. let's come up with a way for me to choose
	3359	the parameters, then render the latest version of that image." And, of the choices then put to
	3360	the author: save from the app with a key and a button that show up only locally; 4K; no labels at
	3361	all; the renders in a `photos/` folder that git ignores; the exterior first, then the two caps
	3362	from the Capital.
	3363
	3364	The list (`content/photos.json`, `src/world/photos.ts`). A photograph is a view, a moment
	3365	and a frame: the world clock in hours, the Light, the frame's aspect, and then for the
	3366	interior where the observer stands and looks (`theta`, `z`, `eyeM`, `yaw`, `pitch`, `fovDeg`)
	3367	and how far the far cap's sky has wheeled (`spin`); for the exterior the camera and its target
	3368	in kilometres, the hull's turn (`phase`), the field of view and the hull reading. The exterior
	3369	has no zoom, so a saved shot's field is always its own 42°; the number is there to be typed,
	3370	to frame a shot tighter than the view can. Its `id` is its file name.
	3371	`parsePhoto` checks every field and says which is wrong, because the list is typed by hand as
	3372	often as it is saved, and the id is held to letters, digits and hyphens because it becomes a
	3373	path.
	3374
	3375	Saving (`ui/app.ts savePhoto`, `vite.config.ts`). P, or Save this shot under View,
	3376	takes the open 3D view, asks for a name, and posts it to the dev server, which writes the
	3377	list; a name already there is re-framed after a confirm, keeping its place and its note. This
	3378	is the one place the browser causes a file to be written, and D-072 stands: it is not the
	3379	world, it is not in the build (the row is hidden and the key does nothing unless
	3380	`import.meta.env.DEV`), and the endpoint answers only the machine the server runs on, since
	3381	`npm run dev` is open to the network. A shot is refused while a scenic tour runs: the tour's
	3382	flyers and camera are its own and would not come back.
	3383
	3384	Going back to one (author, the same day: "it would be helpful to be able to navigate back
	3385	to existing points, like replaying. show a list of current ones and allow loading one?").
	3386	Saved shots, under the button, lists the file as the dev server has it now, read again
	3387	each time the list is opened. Choosing one opens its view and poses it (`poseShot`, not held),
	3388	with the clock stopped at its hour and, outside, the spin switched off, since otherwise the
	3389	framing is gone before it is seen; the view carries on from there, and the shot's name is
	3390	offered when the next one is saved, so adjusting and re-framing is two keys.
	3391
	3392	Rendering (`tools/photos.mjs`, `npm run photos`). Each shot is opened with its clock in
	3393	the address, posed through `window.towerPhotos.show` — which sets the Light and the hull and
	3394	calls the view's `holdShot` — and photographed with everything but the canvas taken away:
	3395	panel, readouts, gauge, view buttons, labels. The window is the shot's aspect, 1920 CSS px on
	3396	its long edge at two device pixels, so the picture is 3840 px and the stars and lines keep the
	3397	weight they had on the author's screen. `--long-edge` changes the size, `--only` picks shots.
	3398	A held view stands still — the hull does not turn, the far cap's sky does not wheel, the water
	3399	is at the held scenes' moment, the plant does not vent — and an interior entry rendered twice
	3400	gave the same bytes. Posed directly, not through the UI: the screenshot tour (D-042) is a test
	3401	of the application and stays as it is; this is a camera. The two share `tools/browser.mjs`,
	3402	and the tour hides the save button, since it runs on a dev server and photographs what
	3403	readers get.
	3404
	3405	Not done. A shot taken during a tour.
	3406
	3407	The list as it starts. The far cap and the sun cap from 300 m over the Capital, at 08:45
	3408	(author: "i like the cap photos, but please change them to be at around 8:30-9am"). And
	3409	`tower-from-the-far-end`: the author's blog illustration, whose settings were never saved,
	3410	worked back from the JPEG. Its stars matched a render's at 1.28 times the size, so it is a
	3411	crop — hence the 33.31° field; the camera was fitted by least squares to the far cap's hub
	3412	and outline, the collector's rim and the sun, to about 3 px; the stars put it on day 5, and
	3413	the lit rod seen through the far cap puts the sun 525 km along, 18:47. The cap's frame
	3414	repeats every sixth of a turn and the rock could not be matched through the processing, so
	3415	`phase` is 0 by default rather than by measurement.
	3416
	3417	Tests (`test/photos.test.ts`): both kinds of shot read back as written; a bad field named;
	3418	ids that could leave the folder refused; every entry in the list valid and uniquely named.