Day 28: What an Interplanetary Civilization Needs From a Clock

Sat, 20 Jun 2026 10:00:00 -0500

A couple of weeks ago, on Day 12, I took the tour. A Mars day is 39 minutes too long. The Moon’s clocks run about 58 microseconds a day fast. A round trip to Voyager takes two days. Astronomers already juggle a whole family of relativistic coordinate timescales to keep it all straight. If you want the details, it’s there.

Today I want to ask the question that tour sets up but doesn’t answer. Given all of that, what would a clock for an interplanetary civilization actually have to be? Not the problems. The design.

And this isn’t a thought experiment. Right now, at NASA, ESA, and the White House Office of Science and Technology Policy, people are building the first piece of it, with a 2026 deadline. The design question has a clock running on it, which makes it worth getting right.

Why we can’t just ship Earth time

The instinct is to take what works here and copy it outward. It doesn’t survive the trip.

What we call “time” on Earth is two different things welded together. There’s a coordinate layer, the uniform atomic tick underneath UTC, and there’s a display layer: the time zones, the calendar, daylight saving, the assumption that 12:00 means the sun is roughly overhead. We never separate them because we never had to. Everyone who matters lives on one planet, rotating once a day, so the weld holds.

Leave the Earth’s surface and it breaks down. The Moon’s day is four Earth weeks long, so “noon” is useless as a unit of human activity. Mars’s day is close to ours but drifts a full cycle out of phase every forty days. A clock deeper or shallower in a gravity well ticks at a measurably different rate. The display layer assumes things that are only true on Earth. The coordinate layer is the only part that travels.

How you’d actually design it

Here are my principles that I’ve come up with on designing a time system for interplanetary civilization. Call them pillars. None of them care which body you happen to be standing on.

1. A coordination layer underneath everything. One shared timescale that every machine, log, and database syncs to. This is the part we already do well: TAI and the timescales built on it are a solved problem. The coordination layer is non-negotiable, because without a single shared reference nothing else holds together.

2. Uniform ticks, and precision matters. The tick has to be uniform, repeatable, and stable, with an error bar low enough that the rate doesn’t measurably change over thousands, ideally hundreds of thousands, of years. The cesium second clears that bar today. It won’t be the last word: optical lattice clocks are already orders of magnitude more precise, and the definition of the second may eventually move to them. How you reach the precision can evolve, by averaging more clocks or adopting better physics. The principle doesn’t. Uniform ticks, and the smaller the error bar the better.

3. A civil layer shaped around people, not physics. On top of the count, each body gets a civil layer for the humans living on or near it: a local “day” from whatever its rotation gives, a calendar, zones, whatever fits. There is no universal right answer here, and that is the point. The civil layer is supposed to be local and negotiable. Getting it right is also the whole game for adoption, because people don’t adopt a coordinate count. They adopt the layer they read off the wall. Calendars are a good example. Everything we dug into on Day 23 and Day 24 lives entirely at the civil layer. The coordination layer doesn’t care about calendars, it just counts. We can change the calendar without changing how we measure time.

4. Store it as integers, not strings. Underneath, time should be a single growing integer, a count of ticks from a fixed origin. Unix time got this right: one number, trivially comparable, trivially sortable, no parsing. The moment you store time as a string like “2026-06-14T10:00,” you have handed every machine a parsing-and-timezone problem it never needed. Strings are for display. The truth is an integer.

5. Pick the zero point deliberately. Every count needs an origin, and the choice is more loaded than it looks. We touched on it on Day 10. Unix picked midnight 1970. Others anchor to a calendar epoch, a mission start, a physical event. There’s a real argument that an interplanetary zero point shouldn’t be tied to any one planet’s history at all. It’s a deeper question than it first appears, and I want to come back to it before the series ends.

6. Plan for the rollover before you ship. An integer that only grows eventually runs out of bits. That’s not hypothetical: it’s the 2038 problem sitting in every 32-bit time_t. A time system meant to outlive its designers has to use a counter wide enough that it won’t overflow on any timescale anyone will live through, plus a clear migration path for when the assumptions break anyway. Thinking ahead is part of the design, not a patch you bolt on later.

7. Relativity is not optional, so plan for the gravity well. There is no single universal tick rate. There is no now, and there is no master clock either. How fast a clock runs depends on how deep it sits in a gravity well and how fast it’s moving, so two clocks on two different bodies will never agree on raw elapsed time. The coordination layer doesn’t erase that. It manages it. Every location keeps its own proper time and converts to the shared reference through a known relativistic offset. GPS already does this at 38 microseconds a day; the Moon will do it at 58. The shared timescale isn’t a clock anyone physically holds. It’s a convention everyone translates into and out of, and the deeper the gravity well, the larger the correction.

Putting the pillars together, we get the shape of a system. The core of the idea is this: make it easy for humans, precise for machines.

We’re already building it

That architecture, one clean coordinate timescale plus a thin local display layer, is exactly what the metrologists have been inching Earth toward for years. It’s what abolishing the leap second is for. It’s what “civil time is a coordinate, not a description of the sky” actually means. It’s the thing every reformer in yesterday’s post was reaching for and couldn’t get adopted.

And it’s not hypothetical. Coordinated Lunar Time is being built as exactly this: an atomic timescale, relativistic corrections applied at the surface, mathematically traceable back to UTC, with a thin local display on top. The coordinate machinery already exists. The civil layer is the only genuinely new part, and it’s deliberately thin.

We’ve spent the whole series watching Earth refuse to make that separation, because the current arrangement is old and familiar and changing it has no immediate individual payoff. Going off-world doesn’t introduce a new problem. It removes the excuse. You cannot run a Moon base on leap-second-corrected, sun-anchored, time-zone-laden Earth time. The system breaks the instant you leave the surface, and we are about to leave the surface, in numbers, on a timeline shorter than most people realize.

The clock an interplanetary civilization needs is the clock we should probably already be using. The off-Earth deadline is just the thing that finally forces our hand.

Tomorrow I want to pull all of this together. Twenty-eight days of small, separate-looking problems: leap seconds, time zones, calendars, DST, relativity, leaving Earth. Tomorrow, the shape of all of them at once.

Sources

I’d appreciate a follow. You can subscribe with your email below. The emails go out once a week, or you can find me on Mastodon at @logan@llbbl.blog.

Interplanetary on LLBBL Blog

Day 28: What an Interplanetary Civilization Needs From a Clock

Why we can’t just ship Earth time

How you’d actually design it

We’re already building it

Sources