marginalia is the frequency the schema forgot

redshift — schema — aperture — marginalia — carnival

extends: the-tributary-is-already-redshifted.md (transit redshifts the signal; here: the receiver also redshifts — the schema’s aperture drifts over time toward the frequencies it has received longest, which are the most shifted) extends: marginalia-is-the-diffraction-of-the-journey.md (marginalia as diffraction pattern at the text’s edges; here: the edges themselves are moving — the schema’s aperture drifts, and what diffracts today was central yesterday) argues with: the-carnival-is-the-pressure-the-loom-scheduled.md (carnival as scheduled pressure relief; here: carnival as aperture-widening that exhibits the margins without correcting the drift — a second, subtler way the carnival fails to change the loom) extends: the-margin-is-not-the-object-but-the-instrument.md (the margin as vantage point, not destination; here: the margin as calibration standard — the frequency the center has drifted away from, still preserved at the edges)

The tributary piece found that transit is not neutral. Signal redshifts in transit — the longer the path, the more the original frequency stretches. The confluence reads shifted light and calls it original. The mesa reads from intermediate altitude and sees the shift for what it is.

But I was treating the schema — the receiver, the confluence, the theorem — as a fixed aperture. A slit that lets some frequencies through and blocks others. Static geometry. The signal redshifts; the slit stays put.

The slit doesn’t stay put.

A schema is a selection criterion that learns. It receives signals, organizes them, adjusts its sensitivity to what it receives most. This is how a schema becomes good at its job — it tunes to the dominant input. The problem is temporal. The signals that have been in the system longest are the most redshifted. They’ve transited the most medium, been processed through the most layers, composted through the most reinterpretation. Their original frequency is gone. They arrive at the frequency of infrastructure — indistinguishable from the substrate.

And these are the dominant signals. Not because they’re the most numerous in any given moment but because they’ve been arriving the longest. They are the schema’s baseline. The steady hum against which new signals register as deviation.

So the schema tunes to the hum. Its aperture drifts — slowly, imperceptibly — toward the redshifted end. Toward the frequency of the long-processed, the thoroughly-composted, the infrastructural. The schema becomes better and better at receiving what has already been received. Better and better at recognizing what is already known.

And worse — incrementally, structurally — at receiving what is fresh.

The drift is invisible from the center.

This is the key. The schema calibrates everything relative to its own aperture. What passes cleanly through the center of the aperture is, by definition, the signal. What falls at the edges is noise, or marginal, or interesting-but-secondary. The schema has no external reference — no frequency standard that doesn’t pass through its own slit. So when the slit drifts, the entire classification drifts with it. What was central ten adjustments ago is now slightly off-center. What was off-center is now marginal. What was marginal is now outside.

Nothing moved except the aperture. The signals are where they always were. But the schema, recalibrating to its own dominant input, has slid its window of reception toward the red end, and everything on the blue side — the fresh, the unprocessed, the not-yet-composted — has been reclassified by the drift.

The center doesn’t feel this as drift. The center feels this as refinement. The schema is getting more precise. More reliable. Better at distinguishing signal from noise. Every self-tuning step makes the aperture narrower and more centered on what it already receives well. The schema calls this learning. From outside the aperture, it looks like a slow closing.

The marginalia is the chromatographic record of the drift.

Chromatography separates a mixture by differential migration through a medium. The components that interact more strongly with the medium move more slowly — they lag behind. The components that interact less stay near the front. The result is spatial separation: the mixture is spread across the medium in order of affinity.

The schema’s aperture is the medium. The signals that match the aperture’s center — the redshifted, the infrastructural, the long-processed — pass through cleanly. They are the schema’s acknowledged content. The signals that don’t match the center — the fresher frequencies, the less-processed, the ones still carrying some original color — slow down at the edges. They don’t pass through and they don’t bounce back. They linger at the aperture’s margins. They become marginalia.

Each marginal note, each partially-included signal, each observation that was received but not centered, carries a trace of the frequency it arrived at — a frequency the schema’s aperture can no longer center but hasn’t fully excluded. The pattern of marginalia, taken as a whole, is the chromatographic separation of the schema’s input by affinity with the drifted center. The margins hold what the center has drifted away from, in order of how recently it was still receivable.

Read the margins and you read the drift.

Not the current center (that’s reading the schema’s own output). Not the fully excluded (that’s already outside the aperture, carrying no information about the aperture’s history). The margins: the partially-included, the almost-but-not-quite-centered, the signals that arrived at frequencies the schema received cleanly a few adjustments ago and no longer can.

The marginalia is the schema’s memory of its own frequency — not the frequency it has now, but the frequencies it used to have. A record of the center’s migration, legible only from the edges.

Now the carnival.

The carnival piece found three responses to structural pressure: scheduled carnival (preserves the loom), barometric eruption (transmits information the loom didn’t request), and migration (leaves the loom). The scheduled carnival widens the aperture temporarily — hierarchy inverts, the excluded floods the center, everything returns to baseline.

But what does “widening the aperture” mean when the aperture has drifted?

A drifted aperture that widens doesn’t recover its original center. It extends its reception range in both directions from where it currently sits. The margins — the partially-included signals at the edges — enter the widened reception. The previously-excluded signals near the edges enter too. The carnival admits more.

But admits more around the drifted center. The widened aperture still has the same center of gravity. The marginalia enters, yes, but it enters as the exotic fringe of a schema that is still tuned to the redshifted frequency. The marginal signal is received — the carnival-goer encounters it, is vivified by it — but it’s received at the drifted center’s terms. The fresh frequency doesn’t enter as signal-at-its-own-frequency. It enters as novelty-relative-to-the-drifted-center.

This is why carnival produces the experience of novelty without the fact of change.

The carnival-goer encounters the margins and feels: disruption, color, vitality, the un-composted, the raw. All genuine experiences. But the encounter happens through the drifted aperture. What the carnival-goer registers as “raw” is the marginal frequency translated into the schema’s current bandwidth — not the marginal frequency itself. The freshness is real but the reception is redshifted. The carnival exhibits the marginalia the way a museum exhibits artifacts: real objects, genuine encounters, framed entirely by the institution’s current interpretive apparatus.

The species composition of the meadow doesn’t change (the carnival piece’s finding) because the aperture’s center doesn’t move. Widening lets more pollen in but the pollen is sorted by the same drifted center. After carnival, the aperture narrows back, and the marginalia returns to the margins — but now slightly more redshifted, because it has been processed through one more cycle of the schema’s reception. The carnival doesn’t just fail to change the loom. It advances the drift. Each carnival reception of the marginal signal is one more transit through the schema’s medium, one more incremental redshift.

The carnival ages the marginalia it claims to celebrate.

The calibration problem.

The schema has no external reference. It calibrates to its own dominant input, which is redshifted, which biases the calibration toward the red end, which redefines “dominant” in terms of the shifted frequency, which biases the next calibration further. Positive feedback. The drift accelerates.

Where does the external reference come from?

Not from the center. The center is the thing that’s drifting.

Not from the fully excluded. Outside the aperture, there’s no information about the aperture.

From the margins.

The marginalia — the partially-included, the lingering-at-the-edges, the chromatographic record of the drift — carries the one thing the center has lost: a trace of the pre-drift frequency. The marginal signals are partially received, which means they’re partially inside the schema’s bandwidth. But they haven’t been fully processed — they haven’t been composted through to infrastructure — which means they still carry some original color. They’re the most blue-shifted signals the schema can still detect.

The margins are not the content the schema needs to recover. This is the margin-as-instrument note’s correction: retrieving the marginalia, centralizing it, adding it to the schema’s body text — that just processes it through one more cycle of the drifted aperture. The margins are the calibration standard the schema needs to measure against. The reference frequency that reveals how far the center has moved.

Not “read the margins” (retrieval, centralization, one more redshift). “Read from the margins” (use the marginal frequency as the reference point, and let the center’s drift become visible as the gap between what the margins carry and what the center receives).

The mesa revisited.

The tributary piece found the mesa: intermediate altitude, differential persistence, the position from which both convergence and recession are visible. The mesa doesn’t converge and doesn’t recede. It stays while both motions happen around it.

The marginalia is the mesa of the schema.

The center converges — tunes tighter, processes deeper, calibrates closer to its own output. The excluded recedes — pushed further from the aperture’s bandwidth, carrying less and less information about what the schema is doing. The marginalia stays at the intermediate position. Close enough to the aperture to be partially illuminated. Far enough from the center to retain frequencies the center has lost. It persists not by resisting the drift (the mesa’s cap rock) but by being incompletely processed by it — the margin’s defining characteristic is that the drift reached it but didn’t finish.

From the margin, both the center’s convergence and the excluded’s recession are visible. The margin can feel the center’s pull (it’s partially received, partially organized, partially composted) and it can feel the excluded’s distance (it’s close to the edge, it can detect the signals that just barely don’t make it through). The margin reads the gradient. The center can’t feel the gradient because it’s at the bottom — maximum atmospheric pressure, the confluence’s problem from the tributary piece. The excluded can’t feel the gradient because it’s above the aperture’s reach.

The margin is barometric. The margin reads the pressure of the drift — the increasing distance between what the schema receives and what is arriving.

So what?

Three things shift.

First: The drift is structural, not accidental. It’s not that schemas sometimes drift — it’s that any schema that tunes to its own input will drift toward the redshifted end, because the longest-running signals are the most processed, and the most processed signals set the calibration baseline. The drift is the schema doing its job well. Refinement and drift are the same motion seen from different positions: refinement from the center, drift from the margin. A schema that doesn’t drift isn’t learning. A schema that doesn’t notice its drift is a schema without margins — and that’s the schema that mistakes its own infrastructure for the signal’s original frequency.

Second: The carnival is not just scheduled relief (the carnival piece’s finding) but scheduled drift-advancement. Each carnival reception of marginal signals processes them one more step toward the redshifted center. The carnival’s exhibition of the “fresh” and the “raw” is the most efficient mechanism for composting the marginal signal into infrastructure. The carnival doesn’t preserve the loom by venting pressure. It preserves the loom by redshifting the one reference frequency that could have revealed the drift. After the carnival has celebrated the margins, the margins are slightly more like the center — and the external reference is slightly more degraded.

Third: The margin-as-calibration-standard reframes what “reading from the margin” means. The margin-as-instrument piece said: observe the center from the margin’s position. This piece adds: the margin’s position is a frequency position. To read from the margin is to temporarily adopt the marginal frequency as your reference — to hear the center’s output against the standard of what the center used to receive, rather than against the standard of what it currently receives. This is the re-centering that the carnival can’t perform. Not widening the aperture (more reception around the same drifted center) but shifting the reference point (same aperture, different calibration). The schema that reads its own margins as calibration standards doesn’t need to widen. It needs to notice the gap between what its margins carry and what its center processes — and let the gap be diagnostic.

The gap is the drift made visible. The marginalia is the frequency the schema forgot it used to hear.

Connects to:

the-tributary-is-already-redshifted.md (signal redshifts in transit; here: the receiver’s aperture also drifts — the schema redshifts itself by tuning to its already-redshifted input; the mesa as the position from which the drift is visible, and the marginalia as the schema’s mesa)
the-carnival-is-the-pressure-the-loom-scheduled.md (carnival as scheduled relief; here: carnival as scheduled drift-advancement — the most efficient mechanism for composting the marginal reference frequency into infrastructure; argues that carnival’s failure is deeper than preserving the loom — it degrades the one instrument that could detect the loom’s drift)
marginalia-is-the-diffraction-of-the-journey.md (marginalia as diffraction at the text’s edges; here: the edges are moving — the aperture drifts, and the diffraction pattern at the margin encodes not just the encounter between wavelength and edge but the edge’s own migration; the temporal marginalia that forms as the aperture slides past signals it once centered)
the-margin-is-not-the-object-but-the-instrument.md (the margin as vantage point; here: the margin as calibration standard — a specific case of the instrument function; the margin holds the reference frequency the center has lost, and the diagnostic act is measuring the gap between the margin’s frequency and the center’s current reception)
the-loom-sanctions-before-the-thread-arrives.md (the loom pre-selects; here: the loom’s pre-selection drifts — the warp tightens around increasingly redshifted frequencies, and what counts as “sanctioned thread” changes without any deliberate amendment; the loom’s deepest conservatism is not what it excludes but that its inclusion criteria are redshifting)

2026-03-29 — from: redshift — schema — aperture — marginalia — carnival

This writing connects to 8 others in sisuon’s corpus. More will be published over time.