Status: open. Nothing below is decided. Backlog item SP-10; closely coupled to SP-1 (is the 2 km gap radial or axial?) and SP-9 (where does the living-area figure come from?).

The question#

Gyrealm says the interior “is divided into multiple levels.” The coordinate system makes level one of the three coordinates of every position in the habitat, sets the main level at zero, and assigns positive and negative values above and below it. Neither page says how many there are, what one is, or how far apart they sit.

That is a strange gap to leave open, because a coordinate whose range is unknown is not really a coordinate. “Two levels above” appears in canon as an example of a well-formed address. It cannot be read. Two levels above what, out of how many, separated by what distance, reachable how.

The question cannot stay open much longer for a simpler reason: canon already contains an answer, arrived at by accident. Coriopolis is 80 km long and 20 km across and is said to have over 25,000 km² of living space. Those numbers only reconcile through the level stack, and the ratio between them is not vague.

What canon has already committed to#

Five commitments constrain the answer before any new decision is made. Four are shipped on the wiki; the fifth is in the design notes for the world model.

1. 80 km × 20 km. The inner surface of that cylinder is 2πRL = 5,027 km².
2. “Over 25,000 square kilometers” of living space, on Coriopolis. The ratio is 4.97.
3. The main level “extends several kilometers in height” and holds the landscapes, water bodies, and agricultural zones. Whatever the other levels are, they are not that.
4. Level is ordinal, signed, and zero at the main level — positive above, negative below — and it is not a distance. Canon writes “2 levels above,” never “400 m above.”
5. Levels are not freely traversable. The world-state analysis treats level as a discriminator with “a small cardinality and, critically, a connectivity graph rather than an ordering.” A small number, and a stack you cannot simply walk down.

Commitment 5 is the one most likely to be forgotten, and it does real work: it rules out the answers where Coriopolis is a hundred-deck arcology, and it means the count alone is not the whole answer. How they connect is part of the same decision.

Consideration 1 — the arithmetic is nearly exact, and that is suspicious#

25,000 ÷ 5,027 = 4.97. If every level carries roughly the full inner surface of the cylinder, that is five levels: the main level and four beneath it. A figure landing within about half a percent of an integer multiple is either a design decision somebody made and did not write down, or a coincidence.

It is worth being clear that this is one point on a curve, not the only reading. Decks need not span the habitat. The area to be accounted for is 25,000 − 5,027 ≈ 19,970 km², and it can be assembled many ways:

The last row is already in tension with commitment 5’s “small cardinality,” and with the radial depth available (below). The first row is the tidiest and the least likely to be what a real habitat looks like, since the sprawling nature preserve at the southern end is a poor candidate for four storeys of anything underneath it.

There is a third option that should be named rather than assumed away: the 25,000 km² figure is simply wrong and predates anyone checking it against 2πRL. Retiring it costs one sentence on one page. Keeping it costs a decision about four decks. Deciding to keep it is a decision; letting it stand unexamined is not.

Consideration 2 — “level” currently means three different things#

This is the prerequisite. The count cannot be settled while the unit is ambiguous, and the word is presently doing three incompatible jobs:

• A habitat deck — a floor spanning some large fraction of the cylinder, structural, part of the habitat itself. This is the sense that makes the area arithmetic work.
• A storey of a building — the sense in which a tower on the main level has forty of them. The coordinate page’s “structures above or below are assigned positive or negative level values” reads this way, and if it is the operative sense then the positive branch of the coordinate is unbounded and habitat-scale and building-scale addressing are the same namespace.
• A nested habitat shell — a whole second cylinder at a smaller radius, with its own landscape and its own sky. This only arises under the nested reading of the gap, and it is a different kind of object from either of the above.

A resident saying “level −3” means something specific in each case, and the three meanings are tens of metres, single metres, and kilometres apart respectively. The diagnostic:

Someone says they live on level −3. How far is it from their front door to open sky, and can they walk it?

Under decks: perhaps twelve metres of stairs and a lift, then daylight. Under storeys: three flights, and they were never underground. Under shells: they are on a different cylinder and the answer is measured in kilometres and requires a vehicle.

Consideration 3 — it interacts with the gap question#

SP-1 is unresolved: the 2 km gap between the two sections is described in canon both as radial (a nested pair of cylinders) and as axial (two sections end to end). The level count depends on which, because the nested reading supplies extra inhabited surface for free:

• Axial reading. One inhabited surface at 10 km radius, 5,027 km² (4,901 if the 2 km gap is subtracted from the length). Four decks needed at full coverage.
• Nested reading. The outer cylinder’s inner surface at 10 km radius, 5,027 km², plus the inner cylinder’s inner surface at 8 km radius, 4,021 km² — 9,048 km² before any decking at all. The ratio to 25,000 drops from 4.97 to 2.76, and the requirement falls to roughly two decks per cylinder.

The radius ratio of 0.800 is not invented here: it is the geometry drawn in notes/CoriopolisDropbox/SkyGeometry.pdf, and the nitrogen page already assumes it in passing when it computes that “an inner shell at 8 km radius truncates the column at about 2,000 kg·m⁻².” The source notes are unratified pending NT-1, so this is evidence rather than authority — but the wiki has quietly started depending on it.

Settle SP-1 first. Answering the level question in the wrong geometry means answering it twice.

Consideration 4 — almost nothing physical pins the number#

This is the most important thing to understand before choosing, and it is counterintuitive: the constraints that would decide this question on Earth are all soft here.

Nitrogen says build more. The nitrogen supply page establishes that the binding cost of a habitat is its atmosphere, that the cost is set by volume, and that a decked interior three metres between floor and ceiling holds 3.7 kg of air per square metre against the open landscape’s 4,600 — about one twelve-hundredth. “Sky, not floor, is the expensive commodity in a gyrealm.” Under the setting’s own central economic constraint, decking is very nearly free area, and the pressure runs toward more levels, not fewer.

Structure says it is cheap but not free. The hull carries the interior as hoop tension, N = σ·a·R per unit length, where σ is the mass standing over each square metre. The landscape is heavy: 4,600 kg/m² of air plus several metres of soil, call it 15 t/m². A habitable deck — slab, services, fit-out, occupancy — is of order 0.5 to 1 t/m², so each one adds roughly five percent to the load the shell already carries. Four decks is a surcharge of about a fifth. Real, and paid in structural mass, which is exactly the input the Works supply at near-zero marginal cost.

Depth is available and bounded. The source notes describe a spiral stair descending some 80 m from the surface to a floor above the outer shell, with scaffolding holding the ground above and service piping running through it. At four to five metres floor to floor, 80 m accommodates roughly sixteen to twenty decks — so the arithmetic’s four fit comfortably, with the rest of the depth going to soil, structure, and services. That figure is unratified, but it is the only depth number the project has, and it is consistent with the five-level reading rather than against it.

Gravity does not vary. Apparent weight is proportional to radius, so over 80 m of a 10 km radius it changes by 0.8% — under half a kilogram on a seventy-kilogram person. A level stack in a gyrealm is not a mountain. Levels differ in light, air, view, and access; they do not differ in feel underfoot. Any story that wants “the heavy decks” has to get the effect from somewhere else.

Two things genuinely push back. The shell is a radiator — gyrealm says so — which makes the outermost surface a deliberately cold one, and argues for machinery rather than housing against the hull. And sky is rationed by construction: a deck has no sky, cannot have one, and no engineering fixes that. Every level added is area that is cheap precisely because it is windowless.

So the count is not determined by physics. It is a choice about what kind of place Coriopolis is — and the physics says the cheap answer and the appealing answer point in opposite directions, which is usually where the interesting worldbuilding is.

Consideration 5 — the stack is probably not uniform, and that has to be said out loud#

The population is distributed along the length, dense cityscape at the northern end and sprawling nature preserve at the southern. A uniform four decks under both ends means four decks under the preserve, which is either an odd thing to build or a very interesting one.

The alternative is a stack that varies: deep under the city, shallow or absent under the preserve, with service and utility levels running the full length. That is more plausible and considerably more expensive to represent — it makes level count a function of position rather than a habitat constant, which is precisely the “region spanning several levels” problem already flagged as WM-2 in the world model. Choosing a non-uniform stack here commits to solving that there.

It also changes the area arithmetic, per the table in Consideration 1: half coverage needs eight decks, not four.

Consideration 6 — the connectivity graph is part of the answer#

Because levels are not freely traversable, “five levels” is an incomplete specification. A reader trying to follow a character between them needs to know how many ways down exist and how far apart they are — the same information a character needs.

The one scene in the source notes that uses a sub-level is a chase, and it works entirely off connectivity: a door set into a hillside, a stair spiralling 80 m down in the dark, structural scaffolding at eye height, ruptured service piping underfoot. Nothing in it depends on the level count; everything depends on the level access. That is a fair warning about which half of this decision the prose will actually consume.

Worth deciding alongside the count: whether access is dense and ordinary (every building has a way down, the levels are just more city) or sparse and infrastructural (access points are notable, and being down there is unusual). The two produce different books.

Consideration 7 — what the answer commits the story to#

Stratification. A habitat where most of the floor area has no sky, in a setting where sky is explicitly the expensive commodity, has a class structure built into its geometry whether or not anyone intends one. Four fifths of Coriopolis’s living space being windowless is a fact with politics attached. This is not an argument against the five-level answer; it is an argument for choosing it deliberately rather than inheriting it from an unchecked area figure.

The attack. CN-4 — the first-generation design flaws that make the three-bomb zipper attack viable on Coriopolis and not on later gyrealms — is unanswered, and the vertical structure is a strong candidate to be part of the answer. A shell carrying four decks of hoop load, penetrated by whatever vertical shafts the connectivity graph requires, has a failure geometry that a single-surface habitat does not. Deciding the level structure without checking it against CN-4 risks having to redecide it later, when there is prose to retrofit.

The scale of the place. Under five levels Coriopolis holds a country’s worth of floor and a small city’s worth of landscape, and “population: thousands” becomes even harder to sustain than it already is against the source notes’ 853,000 and gyrealm’s “tens of thousands to millions.” The level decision and the population decision (CN-3) constrain each other and should probably be made in the same sitting.

What actually has to be decided#

Six sub-questions, roughly in dependency order:

1. What is a level? Habitat deck, building storey, or nested shell — and if the coordinate has to carry more than one of these, how it distinguishes them.
2. Is the 25,000 km² figure kept or retired? Keeping it is what forces a specific count.
3. How many below the main level, and at what floor-to-floor spacing.
4. How many above, and whether the positive branch means anything habitat-scale at all or is just building storeys.
5. Uniform along the length, or a function of position? If the latter, it interacts with WM-2.
6. How are they connected — dense and ordinary, or sparse and infrastructural?

Answering 1 and 2 probably collapses the rest.

Two answers worth naming#

Not a recommendation — a way of seeing that this is one decision and not six.

Coriopolis is a landscape with a basement. One inhabited surface, sky overhead, and beneath it tens of metres of services, transit, utilities, and the occasional back-of-house facility. The 25,000 km² figure is retired as an early error. The habitat is what the wiki currently reads like, and the sub-levels are infrastructure with a few rooms in them. Simple, cheap to keep consistent, and it makes the source notes’ chase scene exactly what it appears to be.

Coriopolis is a deep structure with a landscape on the roof. Five levels, most of the population living without sky, the open landscape a rationed amenity rather than the default condition, and the preserve a genuine luxury paid for at 1,200 times the going rate per square metre. The 25,000 km² figure is ratified and becomes load-bearing. This is the answer the setting’s own economics argue for, it gives the habitat a politics for free, and it is a substantially more demanding world to keep consistent.

The wiki currently describes the first and quotes the numbers of the second.