Ten Rules for an Operating Geography

The easiest way to design the wrong geography is to start drawing lines.

We now have a founding Mars and twelve jobs that geography will one day need to perform there. The tempting next move is to reach for a globe, divide it into satisfyingly regular shapes and then discover a series of extremely persuasive reasons why the geometry we happened to draw first was obviously inevitable.

Geometry is wonderfully obliging in that respect.

A better sequence is to make the requirements explicit before choosing the architecture.

The brief has ten rules:

Be anchored.
Cover everywhere.
Scale coherently.
Stay put.
Take no sides.
Accept overlays.
Be operationally useful.
Work for people.
Work for machines.
Remain open.

These are not laws of planetary geography. They follow from the founding scenario and the twelve jobs we have chosen to design around. Change the scenario and some of the requirements should change with it.

That is not a weakness in the method.

It is the point of having a design brief.

1–2. Anchor it to Mars, then cover Mars

MR1 — Be anchored.

Any operating geography should begin with the Mars we already know how to measure. Spatial identity should derive deterministically from a defined planetary coordinate reference rather than floating above it as an arbitrary drawing or a database maintained by one organisation.

If two competent implementations receive the same position under the same defined reference, they should be capable of deriving the same spatial identity.

There are circumstances in which a database of hand-maintained boundaries is exactly the right tool. For foundational infrastructure, however, reconstructable geometry has an obvious attraction: the relationship to physical Mars does not disappear when the organisation that first published the map does.

Anchoring also means avoiding pointless reinvention. A civil geography does not need its own prime meridian, planetary shape or coordinate convention simply to demonstrate independence. That would be rather like designing a new railway timetable and beginning by relocating Greenwich.

Use the measurement infrastructure that already works.

Build from there.

MR2 — Cover everywhere.

If the ambition is planetary geography, every surface location should belong somewhere, unambiguously.

No strip of terrain should fall outside because nobody expects settlement there. Base units should not overlap in ways that give one point two equally valid answers. Nor should a new geographic unit have to be invented each time somebody discovers that apparently irrelevant terrain is suddenly useful.

The founding scenario gives this requirement practical weight. Science, transport, autonomous systems, emergency operations and resource surveys can all make empty ground matter.

Exhaustive coverage also avoids an awkward administrative ritual. If structural identity appears only when someone declares an area important, somebody must decide when that threshold has been crossed, assign the new geography, publish it and reconcile it with whatever local systems already exist.

A planet-wide structure can simply be there in advance.

That says nothing about ownership, government or culture.

It just means the map has no civil-reference holes.

3–4. Use several scales, and let references persist

MR3 — Scale coherently.

No single geographic scale can perform the twelve jobs.

A vast region can provide excellent planetary context while being useless for locating an injured technician. A few-metre location can identify the casualty perfectly while giving a human almost no idea where on Mars the incident occurred.

The architecture therefore needs useful relationships between scales.

At one end is the planet. Beneath it may sit broad regions, smaller operating areas, settlement-scale geography and kilometre-scale reference before the eventual move to fine location. We have not yet decided how many levels are required or what they should be called.

That comes later.

For now, the requirement is that moving between scales should feel like travelling through one architecture rather than changing systems every time the map zooms.

Hierarchy is attractive because containment carries context. A small area can belong deterministically to a larger one, which belongs to another above it.

There is an obvious limit. Fifty nested levels may be beautiful software engineering and terrible human geography.

The hierarchy should be deep enough to be useful and shallow enough to remain understandable.

MR4 — Stay put.

A civil reference becomes more valuable as records accumulate against it.

Scientific measurement will improve. Software will change. Settlements will grow. Administrations will reorganise. None of those events should casually cause an established spatial identity to migrate across the surface.

So the requirement is persistence.

Persistent does not mean technically incapable of change. It means that changes affecting spatial meaning should be explicit, versioned and rare enough for an old reference to remain interpretable.

Imagine a maintenance record showing that a buried cable crossed a particular area thirty years ago. Better planetary measurement should improve our understanding of the cable’s position; it should not silently change what the historical area meant.

The same problem appears in statistics, legal records and scientific archives.

Stable reference accumulates value because changing things can continue to be described against it.

Place should be more permanent than politics.

Later we will need to decide what that principle means technically. At this stage it is simply a design requirement: persistence should be deliberate rather than accidental.

5–6. Stay out of politics, and tolerate other people’s lines

MR5 — Take no sides.

A common spatial reference should not itself decide who owns, governs or possesses rights over Mars.

Its base units should therefore avoid encoding sovereignty, property ownership, resource entitlement, jurisdiction or political legitimacy. Future institutions may choose to reuse technical boundaries because they are convenient, and any widely adopted spatial system may influence later behaviour simply by existing.

The narrower design principle is that the geography itself does not make the political decision.

A map sheet is not a country. A coordinate zone has no citizens. A grid line does not become a border because it is straight and impressively documented.

MR6 — Accept overlays.

Neutrality only works if other systems remain free to draw different lines.

Settlements will cross base units. Transport corridors will cut diagonally through them. Protected scientific terrain may follow geology. Emergency-service areas may follow travel time. Property boundaries will follow whatever legal history future Mars creates.

Those shapes should be allowed to disagree.

This protects us from a familiar systems-design temptation: once one tidy structure exists, every other structure starts looking as though it would benefit from being tidied into the same one.

Mars will remain inconveniently real.

Terrain will ignore the grid. Engineers will route around obstacles. Settlements will grow in awkward directions. Scientists will protect irregular formations. Governments, eventually, will draw political boundaries according to politics.

The common system should make those structures easier to describe, not require them to obey its geometry.

Matlas should fix the ruler, not what people draw with it.

Once established, that principle should be allowed to stand.

7–8. Make it useful to the civilisation using it

MR7 — Be operationally useful.

A mathematically elegant geography can still be a poor operating system.

The twelve jobs give us a practical standard. A serious civil architecture should support planning, logistics, infrastructure, emergency response, science, aggregation and institutional coordination rather than merely producing an attractive tessellation of the planet.

That does not mean one unit has to perform every job equally well.

It means the architecture should make useful spatial questions easier to answer.

Which area contains this incident? Which larger area contains that one? Which parts of the common geography does a proposed corridor cross? Can records from different organisations be grouped against the same stable units?

A system might perform brilliantly at some abstract mathematical property and poorly at those tasks. It could still be an excellent scientific or computational grid.

It simply would not be the operating geography we are trying to design.

Complexity has to earn its place through use.

MR8 — Work for people.

Humans are awkward geographic users.

We forget long numbers, mishear similar codes, prefer words and patterns, speak locations aloud and routinely describe places relationally. We also move between scales without thinking very hard about it: near the station, north London, southern England, Britain.

A civil geography should accommodate these behaviours rather than treating them as implementation defects.

Its major structure should be understandable enough for people to build a mental model of it. Broad areas should help orientation. Fine location should eventually have human-facing forms that can be spoken, written and recognised without specialised software.

That does not require every internal identifier to become charming dinner conversation.

It requires an interface between machine precision and human use.

Mars will contain computers everywhere.

It will still contain people.

The geography has to respect both facts.

9–10. Give machines certainty, and everyone access

MR9 — Work for machines.

Software has a different set of requirements.

It should not need intuition to determine what a spatial reference means. Given the same inputs and protocol version, independent implementations should derive the same answer. Boundaries need explicit rules. Encodings need canonical forms. Conversions should be reproducible.

Where practical, the core interpretation should also work without a live connection.

That comes directly from the founding scenario. Remote sites, vehicles and emergency systems may occasionally operate without reliable access to central services. Even if communications become superb, geographic identity should not need a proprietary lookup merely to exist.

An API can make a service convenient.

A protocol allows someone else to recreate the answer.

For infrastructure expected to outlive software vendors, organisations and perhaps generations, the distinction matters.

MR10 — Remain open.

The final requirement extends beyond source code.

A foundational Martian geography should be publicly specified, interoperable and independently implementable. Its core rules, constants and conformance behaviour should be open enough for future users to inspect, test and reproduce.

Part of the reason is resilience. Infrastructure embedded in legal records, operational databases and historical archives should not become unintelligible because one website disappears or one company changes its business model.

Part of it is legitimacy. The more foundational a shared layer becomes, the weaker the argument for making its interpretation proprietary.

Open does not mean uncontrolled. A mature standard still needs authoritative versions, change control, conformance tests and eventually stewardship. Otherwise several incompatible implementations can all claim to be open while disagreeing about the actual geography.

The requirement is simpler:

Nobody should need the original designer’s permission merely to work out what the system means.

The requirements fight each other

Written as short commands, the ten requirements look suspiciously compatible.

They are not.

Human legibility may favour a shallow hierarchy with memorable structure, while machine optimisation may prefer deeper recursive systems. Persistence helps historical records but constrains later redesign. Global coverage encourages regularity, yet planetary geometry becomes awkward near the poles. A human interface that works elegantly in one language may become clumsy in another.

That is where design starts.

MR1–MR10 should not be treated as ten independent virtues that any candidate can maximise simultaneously. Some create real trade-offs. Some can be satisfied in several ways. Another system built for climate modelling, geological analysis or autonomous navigation could quite reasonably give equal area, neighbourhood topology or computational efficiency much greater weight.

There is no universal best way to divide a planet because there is no universal answer to best for what?

Our design will be judged against this brief.

Change the brief, change the answer

Imagine a different founding Mars.

If settlement consisted of three giant cities for several centuries, with almost no persistent activity elsewhere, comprehensive planetary coverage might feel unnecessarily ambitious. If one institution really did control every settlement and expected to do so indefinitely, neutrality between organisations would matter less. If continuous network connectivity could genuinely be guaranteed, offline reconstruction might deserve less weight.

Those would not be mistakes.

They would be different requirements produced by different assumptions.

Change the brief, change the answer.

The ten rules therefore do not make Matlas inevitable. They make the later design choices inspectable.

We now have the test.

Before inventing another system, the sensible next move is to ask whether the systems Mars already has can pass it.