Status: resolved — the convention changed. East now runs against the spin and north is the right-hand-rule thumb. The earlier left-hand rule (backlog SP-0) was a repair to the wrong leg of the frame and has been retired. Applied across the wiki and canon on 2026-08-24.

The reasoning below is kept because it records both the wrong answer and how it was caught.

The answer, in one line#

Residents cannot feel the spin — 0.30 rpm is below every vestibular threshold — but they feel the deflections every day. So name east after the deflections. That gives east = antispinward, and the right-hand rule follows.

Under it, dropped things drift east, rising air drifts west, and heading east makes you lighter, all as on Earth; the habitat turns clockwise looking north, as a planet does; and a habitat aligned with the celestial pole spins prograde with the Solar System instead of retrograde. Facing north, east is still on your right — that invariant is what both conventions were built to protect, and it survives unchanged.

The question as posed#

Humans have a turning bias — they prefer to turn left, consistently with automobile racetracks. Analyse both compass conventions in that light and decide whether the convention needs to change. Either way, the spin direction should be chosen on human biology, not on mathematical convenience.

The premise about human beings is correct and better supported than it was a decade ago. The conclusion about turning bias does not follow — but the instinct that the convention should be re-derived from human biology was right, and this page originally concluded otherwise. What follows is the analysis as it stood, then the correction.

Three distinct things were bundled together in the question, and separating them is most of the work.

Three things, only one of which is free#

1. The sense in which the cylinder physically turns. Not a choice at all. An isolated cylinder has no absolute handedness: reversing its rotation is indistinguishable from turning it end for end and relabelling the caps. Nothing an occupant can measure from inside distinguishes the two. The only thing that could make spin sense meaningful is an external referent — the habitat’s attitude relative to the ecliptic, which is a station-keeping and sky-appearance question and has no human-factors content. There is no biological argument to be made here because there is no decision here.

2. Which end-cap is called north, and which way east runs. These are one choice, not two: fixing either determines the other, because east must stay on a north-facing observer’s right. This is the SP-0 question, and it was settled the wrong way.

3. The circulation sense of loops laid out on the floor. Genuinely free, genuinely a human-factors decision, and the place where turning bias actually belongs. This was never decided, and the review makes it decidable.

On (2): what the two conventions actually cost#

Both conventions are internally consistent and neither is more physical than the other. They differ in which terrestrial intuition they preserve, and the two cannot both be kept — the inversion of “up” inside a shell guarantees it.

Read this table knowing what it does not question: both columns assume east = spinward. That assumption is the thing that was actually wrong, and the next section takes it apart.

The trade is a lived, embodied, hourly relation against a view from outside that almost no one ever occupies. The original SP-0 decision made that trade on the grounds of navigational sanity.

One further observation belongs here, though it was overstated when first written. In a gyrealm the compass is not arbitrary relative to the body. Because the rotation vector lies along the north–south line rather than perpendicular to the floor, which way a person is facing determines which of their own head movements cross-couple with the habitat’s rotation: facing north or south, nodding is the provocative movement; facing east or west, tilting the head toward the shoulder is. On a planet, heading has no such consequence. This raises the stakes on getting the frame right — but it does not by itself select a handedness, because the cross-coupling geometry is symmetric between the two ends of the axis. It amplifies the east-on-your-right argument; it does not stand beside it. See turning bias.

Original verdict: keep the left-hand rule. Turning bias does not bear on this question — a person cannot feel which cap is north — and on the considerations listed above, nothing argued for a change.

Why that was wrong. The table above compares the two conventions with east = spinward held fixed, and east = spinward was never examined. It had been assumed early and every later decision was fitted around it, this one included. Hold it fixed and the only remaining choice is which cap is north, and the left-hand rule genuinely is the better of the two. Release it and a third arrangement appears that is better than either: east against the spin, north by the right hand. It keeps east on the right and recovers everything the left-hand rule had to give up.

The deciding argument is physiological after all, but not the one above. Coriopolis turns at 0.30 rpm — below every vestibular threshold. Nobody can feel which way the habitat spins. What residents feel, daily and unmistakably, is the deflection package: water drifting as it falls, weight changing by several percent as they travel. Naming east after the direction of spin anchors the compass to the one circumferential fact that is imperceptible, and inverts every fact that is not.

Spinward wins one row, and it is the row nobody can perceive in either place — no human has ever felt Earth’s rotation carrying them east either. Antispinward wins the five that a resident meets before breakfast.

Corroboration. The unreconciled source notes use east = antispinward throughout: right hand to north, curled fingers pointing west, clockwise looking north, a cascade curving east, and winds “against the rotation, from the west.” Canon had logged three of those as errors. They were not errors — the original material was self-consistent, and the wiki was the deviation. This was noticed after the decision, not used to make it.

On (3): where turning bias does decide something#

The physics is genuinely neutral between left and right, and this is not obvious. Because the rotation vector lies flat in the local horizontal plane, Coriolis acceleration converts horizontal motion into vertical force and vertical motion into circumferential force. It never converts horizontal motion into horizontal force. Nothing anywhere in a gyrealm pushes a walker sideways. The whole interior is in the condition of Earth’s equator.

So a left-turning circuit and a right-turning circuit are mechanically identical, the choice is unforced, and it should be made on the occupants — which is exactly the principle the question asks for.

The evidence supports counterclockwise, with caveats worth keeping:

• Strong. Symmetric-enclosure locomotion trials show better than four in five walkers circulating counterclockwise, replicated across cultures and group sizes, and unmoved by handedness, ocular dominance, or sex. Circle-running studies agree in sign. Pseudoneglect gives a plausible mechanism in right-hemisphere spatial attention.
• Weak. Racetracks are a worse argument than they look. Athletics is counterclockwise by rule, adopted after a clockwise standard was abandoned — real, but a governing-body record, and self-reinforcing thereafter. Horse racing runs clockwise across much of Europe and Asia. Motor sport is split: ovals counterclockwise, road circuits mostly clockwise. The premise is sound; this particular support for it is not.
• Countervailing. The retail “invariant right” points the other way, and larger surveys find that split near even. The honest reading is that the bias is small and overridden by any environmental asymmetry at all — a door, a sightline, a sign. It governs only where the space gives it nothing to work against, which is precisely the case in a purpose-built habitat concourse.

Proposed: gyrealm loops — concourses, ring roads, transit loops, tracks, processional routes — circulate counterclockwise as seen from above, meaning from the axis looking down at the floor. In compass terms that carries a traveller north, then west, then south, then east, through a sequence of left turns, identically to a terrestrial athletics track.

This is written into the coordinate system and turning bias, is unaffected by the convention change above, and is one edit from removal if it is not wanted. It is a soft convention, not a physical fact: an individual habitat could run its loops the other way, and the cost would be a mildly worse-liked concourse, not a broken one.

What the two conventions have in common#

Worth recording, because it is the reason the review closes cleanly rather than reopening everything downstream:

Both conventions leave the deflection table untouched. Falling water still drifts west, eastbound motion is still heavy, axial motion is still undeflected. Those facts are stated in terms of spinward and antispinward, which are defined by the rotation itself and not by the labelling. Nothing in the geography moves under either convention, which is what made SP-0 a pure relabelling and what makes this review cheap.

One point of contact is worth flagging for whoever settles SP-3: the direction the artificial dawn sweeps is also a free human-factors choice, also unforced by physics, and also currently open. It is a different question from this one — it concerns apparent motion of the sky rather than circulation on the ground — but it belongs to the same family, and the principle argued here (where geometry leaves a choice free, resolve it toward what people can actually perceive and already know) would settle it in favour of the spinward sweep — which, under the corrected convention, is the one that makes the sun rise in the east, and is also what the scene prompt already specifies.