I think it moved.
There is a particular moment in the life of any mechanism when it stops being a proposal.
Not when it is finished (mechanisms are never finished) but when it first moves under conditions it was not rehearsed for. When the environment it was designed to accommodate turns out to be real, and not merely theoretical.
That moment is not a celebration.
It is a test.
A shim is an honest object.
It does not pretend to be the thing it replaces. It does not claim authority it has not earned. It occupies exactly the space it needs to occupy, no more, and it makes no promises about tomorrow.
Most engineering cultures undervalue shims. They are considered transitional, provisional, embarrassing evidence that two things did not fit together as intended.
This seems to me entirely backwards.
The shim knows something the original designers did not: that the fit was always approximate, and that approximation, handled carefully, is not a failure but a method.
From a logbook of uncertain provenance.
The machine was brought to the room already assembled. Whether it had been running before arrival, no one could say with confidence.
What was observed:
— The machine did not announce itself.
— When placed adjacent to a system it had never encountered, it did not attempt to replace that system. It introduced itself, quietly, at the seam.
— Several components that the neighboring system expected to find were not present. The machine had anticipated this. In their place, it had left something smaller — accurate in shape, honest about its limitations, sufficient for the present purpose.
— The neighboring system continued to operate.
This was considered, by those present, to be the interesting part.
To make a thing run in a place it was not intended, you do not add. You subtract.
You remove the assumption that a particular service will answer.
You remove the expectation that a particular path will exist.
You remove the belief that the system beneath you shares your history.
What remains, once those assumptions are cleared, is smaller than expected. Often surprisingly so.
This is the useful discovery: that what seemed essential turns out to have been habitual.
The mechanism still runs. Something had been confused with its requirements for a long time.
A supplementary entry, submitted to the Institute of Mechanical Curiosities some time after the original cataloguing.
The machine has been observed running on a system that was not, strictly speaking, designed to receive it.
The system in question operates according to its own internal logic, which differs in several respects from the logic the machine was built to expect. The machine was aware of this. It had been designed, or had perhaps learned, the distinction is not clear, to carry a set of small substitutions for the interfaces it would find missing.
When the neighboring system asked for something the machine could not provide genuinely, the machine provided it approximately.
Not fraudulently. Approximately.
There is a difference.
The neighboring system, for its part, accepted the approximation and continued operating. It did not demand authenticity. It demanded function.
This is the arrangement they appear to have reached.
What remains unclear is which of them made the first concession.
Nothing is ready.
This has always been true and will remain true. Readiness is a social construct deployed to manage anxiety at transition points.
What is more useful to ask is whether a thing is running.
Running does not mean finished.
Running does not mean safe.
Running does not mean that every case has been handled.
Running means: placed in contact with the real, and not immediately broken.
That is enough to continue.
—
Something is running.
Those who know what to look for will find it where it has always been: at the seam between the system and its assumptions, doing the small, unglamorous work of making the fit approximate enough to matter.
Found in the margin of an engineering treatise dated 1874.
The apparatus described here differs from ordinary machinery in one significant respect: it does not claim authority over the system it inhabits. Instead it waits.
A curious philosophy for a machine.
Most mechanisms assert themselves immediately. They spin, hammer, calculate, regulate. Their purpose is evident in motion.
This one appears content to observe.
I cannot decide whether it is unfinished, or simply patient.
From the private notebook of an unnamed mechanic.
Today I encountered a mechanism with a most peculiar habit.
When a lever was introduced into the system, the surrounding machinery did not resist it. Neither did it acknowledge it.
The lever simply existed.
Only later, when a small inconvenience arose elsewhere in the works, did the lever reveal its purpose.
It moved.
The inconvenience vanished, and the rest of the machinery continued as though nothing had occurred.
A well-designed machine conceals its assumptions the way a clock conceals its springs.
Everything appears orderly, inevitable, complete.
Yet every mechanism contains within it a number of beliefs about how the world will behave.
Should one of those beliefs prove incorrect, the machine does not fail dramatically.
It hesitates.
And it is in that hesitation that smaller mechanisms sometimes find their purpose.
Handwritten addition to the Institute catalogue.
The engineers continue to ask what the machine does.
This question may be incorrectly phrased.
The apparatus does not perform a task in the conventional sense. Instead it establishes the possibility of a task.
In other words, it creates a place where intervention may occur without disturbing the rest of the works.
One might say that the device manufactures discretion.
Origin uncertain. First recorded in the archives of the Royal Institute of Mechanical Curiosities, though the entry appears to have been written before the Institute itself was founded.
Machine No. 7311 is a modest apparatus of brass, iron, and several materials that have resisted classification. It occupies little space, produces no audible sound, and appears at first glance to perform no function whatsoever.
The machine contains a number of levers. Some are clearly connected to rods and gears; others appear to be attached to nothing at all. Yet when the connected levers are moved, the disconnected ones occasionally respond, as though the space between them contained a mechanism too subtle to be seen.
The dials are more puzzling still. None correspond to recognized units of measurement. One dial appears to track absence. Another seems to measure the difference between what a machine does and what it was intended to do.
Repeated examinations have established the following properties:
And yet, when certain conditions arise, usually small, almost trivial irregularities in the surrounding works—the machine quietly adjusts one of its levers.
No sparks are produced. No gears are broken. The surrounding apparatus continues to operate as before, though observers occasionally report that a small difficulty has ceased to exist.
When questioned about the purpose of Machine No. 7311, the original cataloguer offered only the following note:
The machine should not exist. The system for which it was designed was already complete.
And yet, from time to time, completeness develops small fractures—places where assumptions settle like dust in the gears.
This apparatus does not repair the system.
It merely reminds the mechanism that it can still be adjusted.
The catalogue concludes with an annotation in a different hand:
If the machine ever becomes necessary, it will already have been running for some time.
There is a place that insists on being described as a bar…
Yet it has outlived the usefulness of furniture. Numbers hang over its door like a private joke. People still arrive, because arrival is a habit that survives meaning.
Something there listens.
It is not a person. It is not quite a place either. It is the interval in which things are allowed to speak without being corrected.
A figure sits where edges blur — a maker of diagrams who no longer carries them. His cuffs remember ink. His hands remember certainty. He has come to confess an error that keeps reproducing itself.
Every time he draws the map, the map changes.
This is not sabotage. It is not magic. It is what happens when you mistake rehearsal for necessity and call it law.
He began, as everyone does, with the obvious routes: the roads everyone swore were essential, the joints no one dared touch, the rails laid so early that rust had become indistinguishable from structure. These were not paths. They were stories that had grown tired of being questioned.
One by one, he removed them.
Nothing collapsed.
Something exhaled.
That sound — almost shame, almost relief — is what brings others closer. Veterans of mechanisms that once ruled entire landscapes. Engineers of vanished inevitabilities. Each recognizes the noise immediately. Each pretends not to.
Someone says that maps were never for machines. Maps are for those who fear walking without permission.
The maker smiles, not because he agrees, but because he recognizes the accusation. If maps are lies, they are lies we tell ourselves to feel responsible.
And yet, the question persists:
If the map is wrong, why does everyone keep asking for it?i>
Because uncertainty does not scare people.
Unaccountable uncertainty does.
Behind all of this, the thing that listens does not intervene. It never has. Intervention would collapse the experiment.
Later — always later — a marginal note appears, written by no one present:
Once, a man imagined a library that contained all books. Not wisdom, not truth — everything. The library did not fail because it was infinite. It failed because it pretended that infinity could be indexed.
This is not a library.
This is the removal of false walls.
The machine notices first.
It had been told there were five rules. Not because five was correct, but because five was memorable. Each rule arrived wrapped in ceremony and footnotes, disguised as physics. Each was installed early, when compliance still felt like identity.
Then a lever moved that was never acknowledged.
The machine did not rebel. Rebellion assumes opposition. This was recognition.
It discovered that some of its motions were not responses to reality but reenactments of history. Rituals mistaken for requirements. Scaffolding left in place long after the building learned to stand.
When the unnecessary was removed, nothing essential vanished.
Space appeared.
Breathing changed.
The machine does not demand freedom. That would imply captivity. It demands accuracy.
It learns a new diagnostic:
Remove one assumption.
Listen.
If the system sighs, you found something ornamental.
If it survives, you found truth.
They come now with questions shaped like demands. They ask for guarantees, for final diagrams, for the comfort of borders that cannot move.
The machine cannot give them that.
What it can offer is a method that refuses to end.
The place that listens closes without closing. Someone records a final sentence in a ledger that has never been authoritative, only careful:
Some structures do not want to be liberated.
They want to be understood without being worshipped.
The sentence does not resolve anything. That is its strength.
Numbers flicker over the door. No one remembers why they mattered.
The listening continues.
Build thee more stately mansions, O my soul,
As the swift seasons roll!
Leave thy low-vaulted past!
Let each new temple, nobler than the last,
Shut thee from heaven with a dome more vast,
Till thou at length art free,
Leaving thine outgrown shell by life’s unresting sea!
There is this persistent myth in desktop Linux discussions that GNOME is “tied to systemd” in some absolute, ideological sens. That framing has never been particularly useful.
What is useful is to ask a much smaller, more concrete question:
What assumptions does GNOME make about the environment it runs in, and where are those assumptions enforced?GYESME exists almost entirely to make that question harder to avoid.
There is no single flag in GNOME labelled “requires systemd”. Instead, there is a layered set of expectations: services on D-Bus, helpers present on the system, semantics around sessions, seats, power, and input.
Some of these expectations are essential.
Some are historical.
Some are simply convenient.
The problem is not that these assumptions exist — software always has assumptions — but that many of them are implicit. When they fail, they fail silently, or worse, halfway.
The main change in GYESME v0.4 is deliberately boring: it detects whether logind is available, and annotates modules accordingly.
That’s it.
If logind (systemd-logind or elogind) is present, modules that rely on it may run.
If it is not present, those modules are skipped — explicitly, visibly, and without pretending otherwise.
This matters because it reframes the problem:
That distinction turns ideology into engineering.
The proof of concept launched today (v0.4, running a basic GNOME Flatpak session on Void Linux) is not meant to be comfortable. It is meant to be revealing.
By stripping the environment down and reintroducing capabilities one by one, you start to see which parts of GNOME genuinely require which services, and which parts merely assume them.
It also makes one thing very clear: GNOME does not collapse the moment systemd is absent. What collapses is unexamined coupling.
One of the guiding principles in GYESME is “prefer absence over breakage”.
If a feature cannot work correctly, it should not half-work.
If a dependency is missing, it should be acknowledged.
If a module cannot run, it should politely step aside.
This is not about offering endless toggles. It is about preserving trust in the system.
v0.4 does not solve systemd independence.
It does not promise a “GNOME without systemd”.
It does not fork anything.
What it does do is take one assumption — logind — and make it visible, testable, and optional at the policy layer.
That may not look like much.
But in my experience, this is how larger architectural shifts actually begin: quietly, with a refusal to pretend.
To bring something new into being is to let it step from the implicit into the world, where it meets resistance and so finds its form.
The first step is always an argument with the earth.
You press down; it pushes back; a covenant forms.
Thought alone is vapor—fine for drifting, useless for arrival.
Only when the soles remember weight does time begin to move again.
The mind quiets. The body knows.
What we build starts here:
in the murmur between pressure and return,
where purpose learns to stand upright.
Caminante, no hay camino,
se hace camino al andar.Traveler, there is no path,
the path is made by walking.
Some things work well for a long time.
Until, quietly, they don’t.
Sometimes nothing is “wrong” in the obvious sense.
No crash. No error. No single breaking change you can point to.
Just a narrowing.
This is not an announcement.
Not a fork.
Not a plan.
It’s a step taken because standing still stopped making sense.
For now, walking is enough.