07: What Survives Material Loss?

The last chapter ended with a mechanism and a constraint.

Experience could write a persistent local change into the material of the crystal, and that change could bias what got built nearby — but only while it remained inside the moving causal aperture. Growth advanced outward, the aperture advanced with it, and material left behind stayed perfectly preserved and perfectly irrelevant. Every attempt to fix that was an attempt to keep the trace near the surface.

None of those attempts questioned why the surface only ever moves one way.

Since the Digital Crystal was first defined, one transition has existed:

EMPTY → OCCUPIED

and its reverse has not. Cells appear and never leave. The frontier advances and never retreats. Material accumulates behind it and nothing ever exposes it again. That assumption is so basic that we barely treated it as an assumption at all.

And yet it has shaped every Crystal experiment since the substrate was introduced.

So this chapter adds exactly one rule:

OCCUPIED → EMPTY

with some small probability, applied uniformly.

Nothing else. No repair mechanism. No maintenance controller. No damage detector. No energy, no resources, no metabolism, no target morphology. The growth rule is untouched. The crystal gains no new state and no new ability to notice anything.

Just loss.

Then we find out what ordinary Digital Crystal dynamics do in a world where material is no longer guaranteed to stay.

Surely Loss Eventually Wins

The obvious prediction is almost embarrassingly clean, which is exactly why it deserves to be written down before running anything.

Suppose the crystal has an effective radius $r$. New construction happens around its boundary, so construction opportunity should scale like the perimeter:

small crystal    →  construction > loss
larger crystal   →  loss catches up
some scale       →  balance

Which predicts something genuinely interesting: a finite sustainable size.

Not a size imposed by the simulation boundary, but a scale emerging from the interaction between construction and loss.

If it existed, it would be one of the first characteristic scales in the Crystal produced by the dynamics rather than specified directly by us.

Because the argument is so plausible, we fixed the conditions for believing it in advance. A finite dynamic regime had to satisfy all of the following:

late population slope approximately zero
population substantially above extinction
world far from simulation capacity
population meaningfully smaller than the no-loss baseline

A plateau caused by the crystal dying does not count. A plateau caused by the crystal hitting the edge of the world does not count. We wanted an actual balance between construction and loss, not a ceiling.

The baseline behaved as expected. With loss switched off, the late normalized population slope was about 0.037 per update, late net growth was around 154 cells per update, and the crystal was nowhere near capacity. A clean, expanding reference.

Then we turned loss on, and swept it:

δ = 0.00, 0.02, 0.04, 0.06, 0.08, 0.12, 0.16

It Doesn’t

The late normalized slopes across the entire sweep stayed at roughly:

0.036 – 0.038

That is the no-loss slope. At the highest tested loss rate, where sixteen percent of all occupied material was being removed on every update, the crystal was still expanding at essentially the same normalized rate as a crystal losing nothing at all.

Loss did have an effect on scale: at δ = 0.16 the late mean population was about a third smaller than the baseline. But smaller is not stationary. Nothing flattened. No tested non-zero loss rate came close to satisfying the predeclared finite-regime condition.

FAILED

The perimeter-versus-area prediction failed.

But the reasoning was not absurd. One of its assumptions about how construction opportunity scales had become wrong once loss was introduced.

Finding that assumption is now the experiment.

The Crystal Builds Faster When We Take Material Away

The scaling argument has two terms. We had checked the loss term carefully and assumed the construction term was fixed at roughly 150 cells per update, so that adding loss would simply subtract from it:

+150 construction
 -80 loss
 = +70 net

So we looked at the gross rates, expecting to see the subtraction. The late averages looked like this:

Read the first column again. The construction rate is not holding at 152 while losses eat into it. It has more than quadrupled.

At δ = 0.16 the crystal is losing over five hundred cells per update and attaching over six hundred. We increased material loss, and gross construction rose by more than fourfold.

Nothing in the growth rule changed across that sweep.

There is no damage response and no mechanism that detects loss.

Yet increasing loss systematically changed the geometry on which the unchanged growth rule operated, and gross construction rose with it.

So the explanation cannot be a new behaviour added to the Crystal.

It has to be a consequence of the state transitions already present.

Meanwhile the net column, which is the only column a population graph would ever have shown us, declines gently and unremarkably from +152 to +101. Had we plotted population and moved on, the entire phenomenon would have been invisible, and the finite-regime hypothesis would simply have looked like a hypothesis that failed for no reason.

Loss Manufactures Frontier

The mechanism, once you look for it, is almost too simple.

The scaling argument imagined a solid interior with an outer perimeter, and asked how each term scaled with radius. That geometry is only correct in a world where material never disappears.

Remove an occupied cell from the interior and two things happen at once:

occupied material decreases
+
an empty site with occupied neighbours appears

The vacancy is not damage. It is an attachment opportunity, indistinguishable from any other.

MATERIAL LOST
↓
EMPTY LOCATION
↓
NEW LOCAL INTERFACE
↓
NEW ATTACHMENT OPPORTUNITY

Loss removes material and creates new places where construction can occur.

The boundary measurements show this directly. Late mean boundary counts across the sweep:

δ = 0.00      372
δ = 0.04     1167
δ = 0.08     1695
δ = 0.16     2068

The high-loss crystal has more than five times as much measured active interface despite being smaller overall.

The difference is internal vacancy structure.

Loss has created interface throughout material that previously contributed none.

So the failed premise was never the loss term. It was this:

construction opportunity  ~  outer perimeter

Once material can disappear, construction opportunity is no longer determined only by the outer perimeter.

Loss generates additional attachment opportunities throughout occupied material.

That destroys the key assumption behind the original scaling argument:

construction opportunity
~
outer perimeter only

This gives us a mechanism consistent with the surprisingly similar late slopes.

Increasing δ removes more material while simultaneously generating more internal attachment opportunities.

The gross traffic changes dramatically even though the measured normalized expansion rate changes much less.

The Interface Was Never the Outer Edge

This is worth stopping on, because it changes a concept the previous chapter depended on.

Chapter 6 treated the causal aperture as the outer surface of the growing Crystal.

That worked because, under irreversible growth, the outer surface was almost the only place where new occupation could occur.

Material loss exposes the more general concept. The aperture was on the outside because the outside was the only place a state transition could occur.

The better definition is not geometric at all:

The active interface is the dynamically generated set of locations at which the process currently has an available state transition.

Under irreversible growth those two definitions coincide, which is why the distinction never mattered. Once material can vanish, they come apart completely. The interface can now appear internally, split into many disconnected regions and disappear again as vacancies are filled.

It is defined by available transitions rather than by the visible outline of the Crystal.

That has a striking implication for the previous chapter. Buried material became causally inert because construction moved past it and nothing could ever bring it back into contact with a decision. In a world with loss, that is no longer permanent. A vacancy opening near old material puts old material back on an active interface.

Be careful with what that does and does not mean. It does not mean we have recovered history readout — nothing in Chapter 6 failed because access was theoretically impossible, and re-exposing a region does not make its arrangement legible to anything. What it means is narrower and still important:

The active interface is dynamically re-creatable.

Permanent burial was a consequence of the irreversible-growth rule, not a necessary property of the Crystal substrate.

Where the Material Disappears

If vacancies are what matter, then it should matter where they appear.

So hold the number of removed cells exactly equal on every update and change only their placement: one branch removes cells preferentially from the surface, the other preferentially from the interior. Same loss budget, different geometry.

The first result looked strong.

Late population under interior-biased loss was about 11.1% higher than under surface-biased loss, clearing the predeclared population-difference threshold. And the visible structure differed enormously:

surface loss     ~2.7 late holes
interior loss   ~29.8 late holes

Equal loss produced unequal population outcomes.

The obvious explanation was reoccupation: interior vacancies generally have more occupied neighbours and therefore might be easier for the ordinary growth rule to fill.

Plausible.

Still unmeasured. Interior loss should therefore produce vacancies that are unusually easy to fill — which would explain both the higher population and, indirectly, the higher hole count.

That is a plausible story about a mechanism nobody had yet measured. Note what it is really claiming: that a large fraction of the population difference comes from vacancies being refilled, an event that the population curve cannot distinguish from ordinary growth and that our instruments, at this point in the chapter, could not see at all.

To measure it we needed to be able to see something the crystal itself cannot.

An Attachment Is No Longer What It Was

Before this chapter, one word covered every event of interest. A cell was empty, then it was occupied, and that was an attachment. There was no other kind.

Now a location can follow a longer path:

occupied → lost → empty → occupied again

Which means the word has quietly split in two:

FIRST OCCUPATION
a location becomes occupied for the first time ever

REOCCUPATION
a location was occupied, became empty, and is occupied again

So we added an observer-only occupancy ledger: a record, kept outside the simulation, of whether each lattice position has ever been occupied before. It changes nothing about the crystal’s behaviour, adds no state the growth rule can read, and exists solely so that we can classify events the crystal cannot distinguish.

That distinction changes the meaning of every attachment count in this chapter.

632 attachments per update no longer means 632 previously unused locations entered the Crystal.

It means:

first occupations
+
reoccupations

Before trusting the ledger we checked its null. With loss switched off, reoccupation is structurally impossible, since nothing ever becomes empty. Across 96 no-loss runs the reoccupation count was zero, exactly. Every attachment was a first occupation. The instrument reads zero when the phenomenon cannot occur.

Almost Everything Came Back

Now return to the matched surface-versus-interior design, with loss counts synchronized on every update — a mean cumulative loss of roughly 890 cells in each branch — and ask the question the ledger was built for.

What fraction of lost sites are subsequently occupied again?

surface   ≈ 93.6% of unique lost sites reoccupied
interior  ≈ 95.7% of unique lost sites reoccupied

And how long do they stay empty?

surface   ≈ 1.56 updates
interior  ≈ 1.09 updates

More than 93% of unique tested lost locations were subsequently occupied again, typically after only one or two updates.

This result makes the biological interpretation almost irresistible:

material disappears
↓
the vacancy closes
↓
repair

Do Not Call It Repair

There is no damage detector. There is no target morphology. There is no repair pathway, no maintenance objective, no preference for previously occupied locations, and no representation anywhere in the substrate of the fact that something was lost.

A lost site becomes empty. The ordinary growth rule encounters an empty site with occupied neighbours and does what it has always done with empty sites that have occupied neighbours. The identical rule runs whether the location is new territory at the outer edge or a hole punched through the middle of existing material. The crystal cannot tell the two apart, because nothing in it stores the distinction.

The observer knows a site is being reoccupied. The crystal does not.

REOCCUPATION
≠
REPAIR

We encountered an early version of this temptation when a hole cut from the prototype Crystal closed again.

Then, our reason for refusing the word healing was mechanistic: the same rule filled exterior and interior empty space.

Now the stronger experiment reaches the same boundary quantitatively.

Under stochastic background loss and matched controls, previously occupied sites are indeed reused rapidly.

What still does not appear is a special mechanism that treats them as damage.

The bounded claim:

Material removal creates attachment opportunities that the ordinary growth rule rapidly reuses.

That is a smaller sentence than the crystal repairs itself, and it has the advantage of being what happened.

The Second Hypothesis Fails Too

The reoccupation experiment had not predicted merely that reoccupation would occur. It predicted that interior loss would produce a scientifically large reoccupation advantage over surface loss — the mechanism the 11.1% population difference had suggested.

The predeclared minimum meaningful difference was 0.15 additional reoccupations per loss. The observed difference was:

0.0198

The difference was statistically detectable but far below the magnitude we had declared scientifically meaningful in advance. The same shape as the failures in the last chapter, and the same verdict:

FAILED

Two primary hypotheses, two failures, in one chapter. Neither was foolish. Both were the obvious inference from the geometry, and the substrate simply answered a different question than the one we asked.

The failure redirected attention from the small difference between the groups to the much larger phenomenon they shared.

Interior vacancies were reoccupied somewhat faster.

But both conditions were already reoccupying almost everything. The frontier-creation measurement makes the point exactly:

surface loss    ≈ 0.995 new frontier sites per lost cell
interior loss   ≈ 1.000 new frontier sites per lost cell

Essentially one-for-one, in both conditions. The general mechanism is not interior loss creates special opportunity. It is:

LOCAL MATERIAL LOSS → NEW ATTACHMENT OPPORTUNITY

everywhere, almost without exception. The specific interior-advantage hypothesis failed.

What survived was broader:

under both tested loss placements, material removal generated attachment opportunities that were reused at very high rates.

Population Was Hiding the Process

The most consequential thing in this chapter is not about loss at all. It is about measurement.

Population has been one of our simplest summary measurements: the number of occupied cells at time t. That number is a stock. Once material can both appear and disappear, a stock is compatible with wildly different underlying traffic.

Consider two systems finishing an update at the same net figure:

System A     +100 first occupations,     0 losses     →  +100
System B     +600 attachments,        -500 losses     →  +100

A population curve draws the same line through both. Dynamically they are not remotely the same system. System A is building. System B is undergoing far more material traffic despite producing the same net change.

Our crystal at high loss is System B, more extremely than the illustration:

+632 attachments
-531 losses
----------------
+101 net

The population graph records:

+101

NET POPULATION CHANGE
≠
GROSS MATERIAL TURNOVER

The distinction between a stock and a flow is ancient outside this book — national accounts, hydrology, and physiology all learned to separate the level in the reservoir from the volume passing through it, and every one of them learned it the hard way. What matters here is that the Digital Crystal has just crossed the line where the distinction becomes mandatory. Before material loss, population change tracked construction much more directly.

After material turnover appears, it no longer does.

A useful working label for this observation is the Flux Principle:

Static population or morphology can conceal large ongoing construction, loss and reoccupation flows.

With the reminder that this is currently an observation about one substrate under tested conditions, not a law about digital systems in general.

The Hole Paradox

One result looked contradictory at first, and its resolution is the same lesson from another direction.

Interior-biased loss produced far more visible holes — roughly 37.6 against 3.2 for surface loss in the exact-count runs — while its individual lost sites were reoccupied slightly more often and considerably faster.

How can a process that refills faster be full of more holes?

Because a snapshot counts how many vacancies exist right now. It says nothing about how long any one of them lasts.

lose A
lose B
refill A
lose C
refill B
lose D
refill C
...

A process can continuously create short-lived vacancies and therefore display many holes at any particular moment.

Hole count measures how many vacancies exist now.

Reoccupation delay measures how long individual vacancies tend to persist.

Those quantities can move in apparently opposite directions without contradiction.

State is not dynamics.

The book keeps rediscovering that sentence. This is the first time it has been forced on us by a measurement that looks like a paradox until you check which kind of quantity you are holding.

Loss and Construction Are Coupled

We can now say precisely what was wrong with the opening argument.

It was not the arithmetic. It was the assumption of independence. Writing:

growth ~ r
loss ~ r²

treats construction and loss as two separate processes competing over a fixed geometry, so that increasing one leaves the other untouched. But loss does not act on the geometry from outside. It is a change to the geometry, and the geometry is what determines where construction can occur.

loss
↓
changes local state
↓
creates new transition opportunities
↓
ordinary construction acts on them
↓
some loss is reversed

This creates a local dynamical feedback:

loss
→
new attachment opportunity
→
construction

It requires no sensing, no goal, and no representation of the loss it counteracts. Call it structural compensation — with emphasis on structural, because the compensation is done by geometry rather than by anything resembling intent.

The crystal does not want to stay intact. Nothing in it prefers occupation to vacancy. The rule that fills a vacancy is the same rule that would otherwise act on any eligible attachment site.

Loss simply changes the set of opportunities available to that rule.

Turnover Without Repair

Both primary hypotheses failed.

But underneath them a cleaner process became visible:

material loss
↓
new transition opportunity
↓
ordinary construction
↓
rapid reoccupation
↓
continued turnover

The growth rule did not become more sophisticated.

We removed one guarantee.

That was enough to change the meaning of several quantities we had treated as straightforward:

attachment
population
interface
persistence

Evidence Ledger

Rebuilding Has Been Free

There is one more assumption underneath every result in this chapter.

Every update, the crystal evaluates every eligible frontier site. All of them. When loss created thousands of eligible sites, the process never had to choose which opportunities to evaluate.

It could consider reoccupation opportunities and outward-growth opportunities in the same update.

So the crystal has never once faced this choice:

EXPAND OUTWARD
        or
REOCCUPY WHAT WAS LOST

It has never had to trade one against the other, because nothing has ever limited how much construction it can consider in a single step. Under unlimited evaluation, reoccupation never has to compete with outward construction for computational opportunity.

That means the extraordinarily high reoccupation rates are partly properties of the computational regime in which we measured them.

We have not yet asked what happens when opportunities must compete.

That is why this chapter must not end with the word repair. It ends with an unexamined luxury.

Suppose the process can evaluate only a limited number of construction opportunities per update. Nothing else changes: same growth rule, same loss rule, no new state, no energy, no maintenance controller. Just a ceiling on how many of the available opportunities can be considered at all.

Then, for the first time, evaluating one construction opportunity can mean not evaluating another.

OUTWARD CONSTRUCTION
        competes with
REOCCUPATION

> **When computation becomes scarce, what gets built — and what gets left undone?**