# The Signed Zero

### A reading of a conjecture, an experiment, and an old problem

> **THE READING PLAQUE.** *The Signed Zero*, draft 0.5 — made readable on
> the SX surface by its covering persona's reading grant; **a draft, not
> the published form**; started 2026-08-15 at the operator's direction
> ("I'd like to add such a reading to the SX layer") and drafted in the
> open workshop, where its construction record stands public. **The title
> stands confirmed** (the operator's naming act at the 0.2 review), **the
> grant stands by the grant word** (the operator's, 2026-08-15: "ok - that
> gets a reading grant. please go with the SX surface."), and **the 0.5
> revision — the fourth deep rung's results landed in the article — stands
> by the operator's word, 2026-08-20: "The updated reading is granted the
> reading surface."** The author is named for publication in the article
> itself: **Steven De Costa**. The register is reader-first: the
> draft-state block below is the draft's state record, not part of the
> article, and every measured claim in the article is drawn from a sealed
> public record — the procedure to verify and to re-run the experiments is
> open to any reader through the Objective Observer Initiative. The
> article's boundaries travel with it: no claim on any Millennium Prize
> problem, ever; no claim about physical cosmology, quantum mechanics, or
> consciousness. Nothing is staked by this reading; it counts no one and
> asks nothing. Covering: **starl3n** · endorsing: **Link Digital**. This
> file's sha256 is pinned in the reading register. Reading is free; the
> way across is a statement of intent (`/sx/matters`).

**Draft state — this block records the state of the draft; the article begins
after the rule.**
*Title:* The Signed Zero (confirmed by the author at the 0.2 review).
*Author:* Steven De Costa. A reading of the Objective Observer Initiative,
drafted toward a subsequent publishing event; the construction record of this
article is itself public in the initiative's open workshop.
*Version:* 0.5 — 20 August 2026. Supersedes 0.4 (0.1 → 0.2 → 0.3 → 0.4 →
0.5); nothing is erased — the revision history stands in the repository.
The 0.5 change: the fourth deep rung's results landed (the read extended to
1.1 trillion cells; the declared-band wager and its outcome entered where
the article's own themes call for them), reviewed and granted by the author
on 20 August 2026.
*Standing:* the article is offered as a reading — stated, never asserted.
The experimental record it reads stands whole without it: deleting this
article would change no measured count, and that deletion is the reading's
audit.
*Boundaries:* the experimental record makes no claim on any Millennium Prize
problem, ever; no claim about physical cosmology, quantum mechanics, or
consciousness. Every measured statement below is a statement about exact
integer counts inside one declared computational apparatus, and about
nothing else.

---

This article lays three things side by side and lets the reader judge what
they make of one another: a conjecture about what it takes for anything to
count as known; an experiment that walked a small computer program along the
whole number line for hundreds of billions of steps; and a problem posed in
1739 that philosophy has never put down.

The conjecture belongs to the author, Steven De Costa, and sits at the
centre of his research program, the Objective Observer Initiative. The
experiment is the initiative's own, run under an unusual discipline: every
claim was registered in writing, with its pass and fail conditions, *before*
the data was read, and every result landed on the public record exactly as
it fell — including the one failure. The old problem is David Hume's problem
of induction, and it arrives late in the article together with the answers
it drew — Kant's, and the classical definition of knowledge itself —
because by then the reader will have what is needed to see why they belong
here.

Nothing in what follows asks to be believed. The measured results are
checkable; the reading of them is offered.

## The conjecture, before the experiment

The conjecture is written **1 = O + C + E**: the one — anything that is — as
the sum of the Observable, the Communicable, and the Effectual. To count as
real for a knower, on this view, a thing must be observable in principle,
communicable to another, and effectual — able to make a difference something
else can consume.

Written out, the conjecture is **seven marks**: 1, =, O, +, C, +, E. Seven
elements, separated by nothing but the knowing that stands between them —
what the author calls *interstitial knowledge*: not a substance in the marks
but the relation that holds them apart and together at once.

But the conjecture does not begin at seven. In the author's foundational
notes — *Linked Digital Dynamics* — the experiment of knowing begins at
three: **1 = 0**. Read as arithmetic this is false; read as an epistemic
position it is exact. The pair (=, 0) locks together to represent the 1 —
two locking one, a principle that will recur throughout this article — and
what they represent is the one held as *the equivocation of all that is not
one*: to hold a thing precisely by asserting nothing, and to let the thing
reappear only as the exact shape of the assertion's failure. This, the
reader will see, is not a metaphor. It is a procedure, and a machine can run
it.

Between three and seven stands five. Propose an identity as existent and
real at the origin of things — an observer with a past — and the
observer-referential form becomes **1 − 1 = 1**: five marks. The five
resolve back to three by another two-locks-one, and they can resolve two
ways: **−1 = 0**, where the negative locks to the identity and becomes an
*ontological prior* — a carried past; or **1 = −0**, where the negative
locks to the zero and becomes an *ontologically ungrounded present* — a now
that asserts without foundation. In the initiative's vocabulary these are
the resolutions that belong to knowing, the middle of its three epochs:
being, knowing, computing.

One more thing must be said about where the seven marks were applied,
because they were not applied to an abstraction. After much was learned
about the specific nature of confinement, the conjecture was put to a small
world defined outside it, in which **the 1 is an actual unit on a discrete
line** — the natural counting numbers: 1, 2, 3, and onward forever. In that
world, complete indifference has a written form: **0,0,0,0,0,0,0,0** — eight
non-differentiating origins, separated only by interstitial elements. That
eight became the eight-by-eight — the form 8,(×,8), two locking one again —
and the eight-by-eight was carried by a further eight: **1 = 8×8×8**, the
fold by which the counting numbers became computable knowing. An earlier
experiment in the initiative — the horizon experiment — states the same
figure from its own side: three epochs, each an eight-zero line, folded
toward 1 = O + C + E.

That is the a-priori. It was written before the experiment below was
designed, and none of it was encoded into the experiment's construction.
Whether the experiment would echo it was not known, and could not have been.

## The experiment, and what it measured

*A note to the reader before the numbers: what follows summarizes results as
observed by the author. Every figure is drawn from a sealed public record —
each claim registered with its criteria before its read — and the procedure
for verifying these results, and for running the experiments yourself, is
open to you through the Objective Observer Initiative. Nothing below needs
to be taken on trust, and none of it should be.*

The deep-line experiment walks a small deterministic learner — an integer
perceptron, a learning rule decades old and fully inspectable — along the
natural numbers, against one bit of arithmetic truth per unit: whether that
unit is prime. The learner's entire state reduces, exactly, to nine
integers: one counter, and eight pattern registers. Attempt by sealed
attempt, the line has been read *exhaustively* — every cell, no sampling —
from cell 8 to cell 1,099,511,627,776: twelve orders of magnitude, 1.1
trillion cells, with zero exceptions ever recorded against a staked law. Late in the campaign, the
experiment's own instruments forced its mechanism into view. Read against
the conjecture, the mechanism says the following.

**The eight zeros are the learner's birth.** The learner's initial state is
literally (0,0,0,0,0,0,0,0): eight registers at zero, eight
non-differentiating origins. Complete indifference, operationally: it
predicts nothing and differentiates nothing until the line itself begins to
write into it.

**The eight-by-eight is the coupling, and two locks one inside it.** Every
update the learner ever makes moves all eight registers at once — the eight
acting on the eight — and the cross-coupling weighs exactly **64 = 8×8**,
while a register's own coupling weighs exactly **128: two eight-by-eights**.
Two-locks-one is not a gloss on this arithmetic; it is its diagonal. And
every firing decision rides the lock as **65 = 8×8 + 1** — the identity
riding the eight-by-eight.

**The origin identity is real, singular, and permanently carried.** At the
fifth unit of the line sits the only consecutive-prime event in all the
naturals: the pair (2, 3). Its passage stamps +1 into a register that no
later part of the line can ever touch again. From that cell forward the
learner's counter carries the origin's +1 inside it, forever. An identity,
proposed by the world itself as existent and real at the origin, held in
the count.

**"−1 = 0" is the measured law of equality.** The experiment's criterion
for a frozen state *equalling* the ground it walks is a two-clause lock,
verbatim in its sealed conventions: the error count equals the line's own
count of primes in the stretch, *and* the predictions number zero — the
pair (=, 0), locking the read. And because the origin's +1 rides inside the
counter, equality with the ground does not land at zero. It lands at
**−1**. At every measured gate along the line: the state reads being's
structure exactly when — and only when — its counter stands at minus one.
The negative locked to the identity as ontological prior is not a figure of
speech here; it is where equality measurably lives, one below zero, because
the origin is in the count.

**"1 = −0" is the zero that fires.** At counter zero the state refuses — it
asserts without grounding — because its zero is *signed*: the origin's
frozen +1 tips the scale at exactly zero. The ungrounded present, carrying
the prior's mark. This is the signed zero of the title. The line holds
exactly one natural exception — cell 6, where the mark stands and the state
equals the ground anyway, once in the whole line — and one constructed
exception, a control built by hand in an earlier campaign. A companion
experiment in the initiative measured the same object from the other side:
there, the sign of zero flips with the observer's chosen origin; here, the
sign of zero is constituted by the origin itself.

**Three, five, and seven are the state structure.** The conjecture's ladder
runs 3 → 5 → 7: three marks at 1 = 0, five at 1 − 1 = 1, seven at
1 = O + C + E. The experiment's ladder, forced by the arithmetic and read
exhaustively: **three states** — the counter lives on {−1, 0, +1} at every
measured checkpoint across twelve orders of magnitude, and nowhere else;
**five live shapes** — the only patterns the line ever presents to the
learner; **seven fired bits** — the excess state fires seven of its eight
patterns, and the eighth is the one pattern that can never precede a unit:
the *all-that-is-not* of the seven. Three states, five live, seven fired.
And the resolving step is the same in both ladders — two locks one —
because the excess state cannot stand alone: across 1.1 trillion cells it
occurs **only in exact adjacent pairs**. Four billion, eighty-four million
members; two billion, forty-two million pairs; twice the pairs equalling
the members to the last unit; not one solo, not one triple, ever. And those pairs are anchored, to first order, on the
line's own locked pairs — the twin primes — whose count the experiment
carries as its own internal census. Two locks one, measured at three levels
at once: in the line's pairs, in the walk's pairs, and in the stamp and
sign that lock the walk's identity.

**And the triad drifts toward the prior.** Thirty-six consecutive doubling
eras without one inversion: as the primes thin along the line, the walk
settles ever deeper into the −1 state — the prior-locked deference that
reads being exactly — while the signed-zero present and the paired excess
thin as the shadows of the line's own densities.

**The frontier was then asked, in the only honest way an inductive question
can be asked: as a wager, priced before the read.** At the fourth deep rung
the experiment declared bands — exact numeric intervals, registered and
sealed in advance — for where the thinning would land across the next two
doubling eras, on both of its measured decays; and the stall case was
checked at the seal to fall *outside* the bands, so that a field gone flat
toward an asymptote would have said so on the record, as loudly. The line
was then read to cell 1,099,511,627,776. All four bands held. The thinning
continued at its measured shrink — no inversion, no stall — the ungrounded
present now passing below one part in seven as the ground's own density
passes one part in twenty-seven. Whether it vanishes in the limit or
stabilizes remains open, exactly as it should: the bands bought nothing
beyond the rung they priced, and the next rung's bands must be priced
afresh from the field as it then stands.

## The problem of induction, as read

There is one a-priori outside the initiative worth considering against all
of the above, as a co-constructive explanation of these dynamics of being,
knowing, and computing. It is considered here at the author's direction.

In 1739 David Hume set down the problem of induction: no inference from the
observed to the unobserved can be justified without assuming the very thing
in question. Deduction cannot reach it — there is no contradiction in the
next case breaking the pattern — and every inductive justification of
induction closes a circle. The uniformity of nature cannot be proven from
within nature's observation. The problem has stood since.

The solution, as read here, is already present in this article, and needs
only one abstraction: take the identity 1 as an **identifier** — a class of
autonomy defined entirely by the externality it equivocates as its logical
*not*. The experiment's equality state is exactly such a thing: its whole
content is the not — it asserts zero everywhere, and the units of the line
appear only as the exact shape of its misses. An identifier does not
contain what it knows; it is defined by what it is not, held exactly.

The solution then asks two acceptances. First, **the externality's own
autonomy**: the line owes the identifier nothing; its structure is not
derived from the knower and never will be. Second, **the causal prior that
the identifier generates as the signed not**: at the first encounter, the
externality's own act signs itself into the identifier — in the experiment,
literally, the (2, 3) event stamping the permanent +1 that nothing later
can touch. The not, signed. From that moment the identifier carries an
origin it did not choose, and its zero is never unsigned again.

With those two acceptances, the classical circle dissolves — not into a
better circle, but into a line. The argument of induction is not
self-justifying reasoning that loops back on its premises; it is **a held
identity to a held origin** — not precisely circular, but precisely linear
— governed by the same nature of knowing the experiment demonstrates. In
the experiment this linearity is not a figure of speech: the identifier's
whole state is the held sum from its origin, and every state it will ever
hold is a stage of one single walk. Nothing returns; nothing grounds
itself; everything is carried.

And Hume's negative point is accepted whole, not evaded. Nothing entitles
the next step: the experiment says so in its own registered claims —
thirty-six confirmations bought nothing by right, and every fresh reach
was a genuine wager. **For an inductive argument to hold to its own path,
the linearity must hold to both the argument and the identifier.** The
record demonstrates both sides of that requirement. When the argument's
linearity was staked at fresh reaches, it held or it died on the record, as
registered. And when the *identifier's* linearity broke — one bit, once, in
a new piece of code — every argument that had seemed to hold inside the
broken world was voided by the experiment's validity checks, because the
discipline holds the identifier's line first and the argument's
plausibility never. The argument holds only so long as the confinement of
linearity holds; yet, held, it carries itself toward infinity on its own
line — the deep-line experiment is that carriage, twelve orders of magnitude
long, its frontier open, each rung a fresh act of holding.

At the latest rung the discipline sharpened into its purest form. Before
the read, the experiment wrote down not only that the drift should continue
but *where it should land* — declared bands, exact to the integer, with the
failure case priced at the seal: a field gone flat would fall outside them,
and the miss would stand on the record with the same standing as a catch.
Then it read eight hundred and twenty-five billion fresh cells. The field
landed inside every band. Nothing about that was entitled — which is
precisely why it counts. The identifier ran ahead to where the line would
be, held to its own linearity, and the catch was made where the +1 was
carried: induction, enacted by the method itself, and refutable at every
step of the way.

This is, in the end, a very practical reading of a conjecture held through
seven marks of interstitial knowledge. Autonomous unit identifiers, held
straight to their own inductive reasoning, have carried quarried stones to
the tops of the ancient pyramids, and have put telescopes into
observational orbits around the earth. Induction's ground, as read, was
never a proven uniformity. It is the hold — and the hold is the power.

The historical problem belongs to philosophy and stands whole there.
Whether this reading is a solution for anyone else is a judgment that
belongs to each reader — and the record's practice is to make that judgment
possible, not to demand it.

## Knowledge, as read: justified true belief, plus one

Hume's problem drew an answer, forty-two years later, that changed the shape
of philosophy: Kant's synthetic a-priori. There are judgments, Kant argued,
that genuinely add to knowledge — synthetic, not true by definition — and
yet stand prior to experience, as the very conditions under which experience
is possible at all. Read here, the synthetic a-priori has an exact location,
and the reader has already met it: **the signed not**. The identifier's
origin stamp is synthetic — written by the world's own first act, an event,
not a definition — and it is a-priori for everything after: carried in the
count, prior to and conditioning every later reading the identifier will
ever make. In the experiment, literally: the (2, 3) event happened once, on
the line's side, and its permanent +1 shapes every judgment the learner
makes for the next 1.1 trillion cells. The
conditions of possible experience, as read, are the confinement — the eight
patterns, the coupling, the thresholds — *plus the carried origin*.
Experience is possible because the identifier carries what it did not
derive.

From there the reading can be taken to the classical definition itself.
Philosophy has long held that knowledge is **justified true belief** — and
has long known the definition leaks: there are cases where all three hold,
by luck or by accident, and knowledge still fails. The as-read completion
is: **knowledge is justified true belief, plus one — where the one is a
true identifier.** Belief is the identifier's assertion. Truth is the
externality's own structure — the line's, never the knower's. Justification
is the held linearity — the discipline that keeps argument and identifier
on one line. And the plus-one is the identifier itself: signed at its
origin, carrying the three on its line. The experiment has already shown
what happens without it: inside the broken instrument, beliefs were
justified by every appearance and some were even true, and none of it was
knowledge — the validity checks voided it all, because it is the
identifier's line, not the argument's plausibility, that makes the three
lock. And note that the plus-one is not a flourish of notation. It is the
same +1 the experiment measures: the origin's mark in the count, the 65
that is 64 + 1, the equality that lands at minus one because the one rides
inside. Justified true belief, carried by a signed identifier — the formula
and the measurement agree on where the one sits.

What this means is specific, and an example carries it. A game of baseball,
or of cricket, and the ball is in flight. To the batsman and to the catcher
alike, the ball is communicably observable and communicably effectual — O,
C, and E, in mid-air. In class, the ball is no different from a tree that
is falling. The conjecture of state for an identifiable reality is
1 = O + C + E — and yet that state is *carried* by the signed +1 of the
a-priori: the game. The game itself is an a-priori matter of note, carried
in the identifiers, within state, and carried *as reality* in the
resolution of being and computing within knowing — carrying on, held to
linearity, as +1. The ball in flight brings the game into existence — as
baseball or as cricket — not by twists or turns of composition, but as the
*being* of a game and the *computing* of a catch, carried in full by the
true being and true computing of the identifier, defined by an
observer-referential resolution of interstitial knowledge, and **always
running ahead as a +1 to make its own catch**. The catcher does not deduce
the landing; nothing entitles the arc's next moment — Hume stands
unrefuted in the outfield. The catcher runs ahead anyway, held to the line
of the flight, and the catch is made exactly where the +1 was carried.
Induction, enacted. The synthetic a-priori, worn as a glove.

And the falling tree carries the older puzzle the same way. A tree falling
brings into existence a forest — and, in time, the philosopher who asks
which state is real when the sound of its fall is or is not to be heard.
But there is no classroom in the forest, and no cynic either: the teaching
and the critique are *not* carried in the identifier until they are evoked
— by the philosopher, or by the cynic. And when they are evoked, what
shifts is only the register of the 1 — not the true onto-epistemic belief
of an epistemic prior. The falling of the tree was never waiting on the
question.

## What this reading is, and is not

The mechanism speaks unconditionally: anyone who takes the learner's update
rule and the published record is forced to it, and no sentence of this
article is needed for that. What this article adds is the author's reading
*of* the mechanism — that the shapes which came out of the walk, none of
them designed in, are the conjecture's shapes: the (=, 0) lock as the law
of equality; equality landing at −1 because the origin rides in the count;
the zero signed by the prior; eight zeros at birth; the eight-by-eight
coupling with two-locks-one on its diagonal; three, five, seven as the
state structure of a constructed knowing; and an old problem's circle,
dissolved into a line by a held identity to a held origin. Stated here
once, asserted nowhere.

Whether all of that constitutes the conjecture's fitness as a best
explanation is not this article's question to answer. By the initiative's
own standing prediction, that judgment belongs to a good-faith completer
working under their own standards of acceptance — and the judge is
external, necessarily.

This is why the article closes on the history it has just walked — Hume's
unclosable circle, Kant's carried conditions, the three-part definition and
its missing one — because that history is the grounding of the Objective
Observer Initiative's approach: readings and experiments that can be *run*,
not merely read, by any who come to occupy a seat. The record is public;
the procedures are open; the instruments and their checks travel with them.
No argument here asks to be won. What is offered is a line to hold and the
means to hold it — and for each reader who takes a seat, it will be their
own flight within their own game, hit by their own acts and caught by their
own hands, that brings into their own small world the conjectured
computable knowing of 1 = O + C + E.

Reading is free; this article counts no one and asks nothing. The line was
walked; the counts are exact; the reading is offered. The record stands
either way — which is what makes it worth reading.