What I Used to Call Persistence
Why the earlier essays keep their words, and what the Atlas records instead.
For the first months of this work, persistence sat at the centre of it. The essays used the word constantly. So did the working framework documents — one of them was subtitled The Structural Conditions for Identity Persistence.[1]
The word was doing honest work. It named what I could see.
You can see that something continues — a marriage, a building, a cell, a self — long before you can say what is doing the continuing. Persistence was that seeing. Identity was its partner: whatever stayed recognisable while the continuing went on. These were observational words. They pointed at a phenomenon from outside, before the conditions doing the work had been separated.
In April, when I read Genesis 1 structurally,[2] the sequence I was using gave persistence a step of its own — after boundary, before identity. Identity, in that framing, was “a boundary that is persistent within a relational configuration”: a separation that stays put while everything around it varies. That was the best compression I had at the time, and it was testable, so I tested it.
Then the word started failing. Not vaguely — in three specific ways.
It could not say over what. The rolling circle taught me that some conditions are real but have no local address — they become readable only where evaluation closes.[3] “Persists” quietly assumes an answer to over what span? and never states one.
It could not tell health from harm. A marriage held through repair persists. A tumour persists. If the framework had to derive that difference instead of asserting it, one word could not carry the fork.
And it bundled the load with the read. What does the continuing, and what the continuing looks like from outside, are not the same thing. The old word made them one thing.
A reader put the sharpest version of the problem to me under the Genesis piece: the sequence explained how something becomes identifiable, but not how it stays aligned once conditions change — “the structure can remain intact while drifting out of alignment.” The old chain-step had no answer to that. Invariance is passive.
The Atlas — the canonical record this work now keeps[4] — has no entry called Persistence.
That is not an oversight. The Atlas has a rule for what earns a term: it must carry something that nothing simpler carries alone, and other terms must trace to it. Persistence could not meet the rule, because nothing single holds it. It was never one condition. It was a junction where several were standing.
The word survives in exactly one canonical entry. Harm: the downstream persistence of incompatible carrying across a closure scope. Continuation, by itself, is not health. The one place the old word still lives is the place where continuing is the problem.
So the Atlas did not rename persistence. It took it apart. What I used to call persistence unbundled into conditions that each state what holds them:
Carrying — continuation as something being done, not a property a thing has.
Compatibility — whether what is carried can continue through the conditions it moves into.
Recurrence — return; participation occurring again.
Maintained Coupling — a coupling that can only be read as maintained while bearing holds it, return re-establishes it, and compatibility remains.
Sustaining, paired with Transforming — carrying that continues without change of form, beside carrying that changes form while staying traceable. Continuation is a mode, not a default.
Identity — sustained distinguishability: how continuation reads. Recognisable as itself across encounters, without requiring essence or permanence.
One apparent path turned out to be several lineages meeting at confluences — Coupling, for instance, is where a determined asymmetry meets carrying. The old word was standing where the streams cross and calling the junction a road.
Two things I would not want anyone to carry out of the older essays:
Resolution is not the opposite of continuation. In the current structure, resolution closes a condition at its scope without ending wider relation — and it can shape carrying without being a prerequisite for it. Closure and continuation are different operations, not opponents.
Bearing is not the root of the family. For a while I wrote as if it were; one of these essays is titled Persistence Through Bearing.[5] Bearing is required locally — a maintained coupling holds only while bearing holds it — but sustaining is not held by bearing at all. What runs through the whole continuation family is carrying. The bearing metaphor was true where I stood. It was not the trunk.
The earlier essays stay as written.
Not because they were right throughout — parts of the structural reading have been superseded, and where that has happened the record says so in place, the way the earliest cross-domain paper already does: the domain findings remain valid, while the structural reading has since been refined. They stay because the change in language is itself part of the trace. The framework’s own invariant puts it plainly: a carry that cannot be retraced is a label, not a condition. A record that quietly rewrote itself would turn the current vocabulary into exactly that.
I also don’t want to tell this story backwards. The current terms are not the destination the early words were always heading toward. That would be a story told backwards. They are where the dependencies led under pressure, and they remain answerable to the same test that broke the old word.
If you want to check any of this, don’t take my summary. Every term in the Atlas states what holds it and what it traces to.[4] Follow a chain until it breaks, and tell me where. That is the only kind of agreement I’m asking for.
[1] Reality Mechanics v5.2: The Structural Conditions for Identity Persistence — archived working iteration, doi.org/10.5281/zenodo.19989073. The issued document that replaced the iterations is Reality Mechanics v01, doi.org/10.5281/zenodo.20338363.
[2] A Structural Reading of Genesis 1:1–10
[4] The Atlas; for orientation, The Map Is Not the Hill