Multiplicative Archaeology: Recovering Interaction History from Digital Atoms

Published 2026-06-23 · Essay · 18 min read · Also on Medium

Listen: Numbers Are Fossils of Interaction History

Should arithmetic remember how it was constructed?

Classical number theory operates under a widely accepted but conceptually limiting premise: a number is simply its value.

When we observe the number 39 and the expression 3 x 13, standard arithmetic treats them as identical objects. The equation 39 = 3 x 13 collapses the process of construction into a static result, forgetting the history of how the object was built.

Here we declare that arithmetic should remember how it was constructed.

In this extended framework, arithmetic operations are not merely rules for counting. They are the physical dynamics of an interaction space, and numbers are the artifacts left behind.

By shifting our perspective from abstract values to structured histories, we transform factorization from a dry algebraic decomposition into interaction archaeology.

We stop asking what a number is, and start asking how much hidden history it contains versus how hard that history is to recover.

Arithmetic Objects as Triads#

To formally ground this perspective, we must stop treating numbers as simple values on a number line.

Instead, every arithmetic object is elevated to a triad:

A = (V, I, R)

Value (V): The standard integer value in $\mathbb{N}$ .
Interaction History (I): A finite, rooted, ordered tree where leaves are primitive digits and internal nodes represent mathematical operations: addition, multiplication, exponentiation. It records the specific causal story of how the object was assembled.
Representation Topology (R): The finite sequence of digits in a given base $b$ that visually and structurally encodes the number.

In traditional mathematics, a forgetful functor strips away I and R, projecting the rich geometry of the object down to a single point V.

But by keeping the triad intact, we expose a critical divergence between representation cost, meaning how much topology is needed to write the object, and interaction cost, meaning how much coupling was required to create it.

For example, the number $10^{1000}$ has an astronomical value but a microscopic representation cost: a 1 followed by zeros. Its interaction cost is highly recursive, yet simple.

Existing Arithmetic Geometry Context#

The intuition that arithmetic carries hidden geometric or physical structure is not new.

In the 1960s, David Mumford and Yuri Manin independently observed that prime ideals in a number ring behave like knots embedded in a 3-manifold. Barry Mazur developed the analogy further, and in the 1990s Reznikov and Kapranov coined the term arithmetic topology to describe the program.

Morishita’s monograph Knots and Primes (Springer, 2012) remains the standard reference. It establishes a dictionary, now called the M2KR dictionary, in which the 3-sphere corresponds to $\mathbb{Q}$ , prime ideals correspond to knots, the Legendre symbol corresponds to the linking number, and the Redei triple symbol corresponds to Milnor’s triple linking number.

The triple of primes $(13, 61, 937)$ , for instance, forms a Borromean link modulo 2: each pair is unlinked, yet the triple is linked. This is a direct arithmetic analogue of the Borromean rings.

A separate line of work connects arithmetic to quantum statistical mechanics. The Bost-Connes system constructs a $C^\*$ -dynamical system whose partition function is the Riemann zeta function $\zeta(\beta)$ . Above a critical temperature the system has a unique equilibrium state. Below it, the system undergoes spontaneous symmetry breaking, and the extremal ground states are parameterized by embeddings of $\overline{\mathbb{Q}}$ into $\mathbb{C}$ . The Galois group $\mathrm{Gal}(\overline{\mathbb{Q}}/\mathbb{Q})$ acts on them exactly as class field theory predicts.

Connes and Marcolli extended this to 2-dimensional $\mathbb{Q}$ -lattices, where the partition function becomes $\zeta(\beta)\zeta(\beta - 1)$ and the symmetry group enlarges to $\mathrm{GL}_2(\hat{\mathbb{Z}})$ .

Yalkinoglu proved that arithmetic Bost-Connes systems exist for arbitrary number fields, and Tabuada categorified the construction using Tannakian formalism over Weil numbers.

Multiplicative Archaeology differs from both programs in its point of departure. Arithmetic topology begins with the etale fundamental group and works downward toward linking invariants. The Bost-Connes program begins with operator algebras and works upward toward zeta functions.

This framework begins with the representation of a number, its digit string in a given base, and asks how the choice of representation controls the computational cost of recovering the multiplicative history.

The three axes $(\Omega, \iota, \mathrm{Arch})$ are therefore not topological or spectral invariants but information-theoretic ones: they measure atom count, atom distribution, and recovery cost respectively.

The RG flow analogy is closest in spirit to Deninger’s program, which equips arithmetic schemes with a foliation and a flow whose periodic orbits correspond to primes. But where Deninger’s flow is a continuous dynamical system on a hypothetical 3-dimensional space, the digit-topology flow is a discrete coarse-graining on a 1-dimensional lattice of digits.

The two constructions are complementary rather than competing. Deninger’s flow explains why the zeta function has the analytic properties it does. Digit-topology flow explains why factoring has the computational cost it does.

1. The Dynamics of Digits#

In this topology, digits, not the numbers themselves, are the primitive physical objects. They act as the free monoid generators of the arithmetic space. Mathematical operations dictate how these digits couple and propagate.

Addition is local pulse propagation. When adding 123456 + 1, the operation only tweaks nearby digits. Carries ripple to adjacent nodes, making it a radius-1 local update in the digit-topology graph. Because carries never jump, the interaction is purely local, and recovering its history is computationally trivial: $O(k)$ time.

Multiplication is global coupling. When evaluating 39 x 39, every digit of the first factor physically interacts with every digit of the second. The output is a trace of all pairwise interactions, essentially mapping a complete bipartite graph between the two representations. Multiplication creates structural dependencies and generates new interaction artifacts.

Exponentiation is recursive coupling. Rather than smooth growth, exponentiation creates staircase representational growth, where the digit length remains flat before violently jumping as boundaries are crossed.

2. Primes as Interaction Atoms#

If multiplication is the process of fusing components together, then what are the fundamental particles of this universe?

They are the objects that cannot be further decomposed into interactions.

A prime number $p$ is historically defined by its lack of divisors. But structurally, a prime is the minimal fixed-point object of multiplicative interaction. It has no incoming fusion history.

If an interaction atom is defined as a node $p \ne ab$ for all $a,b > 1$ , then primes are exactly the interaction atoms of mathematics.

Crucially, the number 1 is not an atom. It is a neutral element. It does not generate multiplicative structure, meaning you can multiply by 1 indefinitely without introducing any new interaction. Primes, however, are the unique, irreducible generators of all multiplicative structure.

When we factorize a composite number like:

1521 \to 39 \times 39 \to 3^2 \times 13^2

we are actively stripping away interaction layers.

The process must eventually terminate, and it stops at the primes because primes are exactly where multiplicative explanation stops.

Because the Fundamental Theorem of Arithmetic guarantees unique factorization, every composite number possesses a canonical set of interaction atoms. Trees of construction may vary, such as $(2 \times 2) \times 3$ versus $2 \times (2 \times 3)$ , but the foundational atomic inventory is strictly invariant.

This completely recasts the Prime Number Theorem. Classically, it merely counts primes up to a bound. In multiplicative archaeology, it describes the density of fundamental particles: as arithmetic grows, new interaction atoms become structurally rarer.

The vast majority of large numbers are not new generators, but highly assembled composite artifacts built from a thinning alphabet of ancient atoms.

3. The Triad of Multiplicative Complexity#

To fully map a number’s hidden history, we must measure three distinct, independent axes: how many atoms it contains, how those atoms are arranged, and how much energy is required to dig them up.

Axis 1: Resistance, $\Omega$ #

Resistance, denoted mathematically as $\Omega(V)$ , is the total number of prime factors of $V$ , counted with multiplicity. It lives purely in value-space and represents the minimum interaction cost over all possible histories that yield $V$ .

Resistance counts how many irreducible multiplicative atoms are embedded in a number.

For a prime, $\Omega(p) = 1$ . It is a single atom. For:

360 = 2^3 \cdot 3^2 \cdot 5

we have:

\Omega(360) = 6

Astonishingly, the average resistance of the first $N$ integers grows at the glacial pace of $\log \log N$ .

This means a typical 100-digit number contains only around five or six prime atoms. Even at massive scales, integers possess surprisingly shallow multiplicative ancestries.

Axis 2: Interaction Entropy, $\iota$ #

While Resistance counts the mass of atoms, it fails to capture the shape or spread of their interactions. A pure power like $2^{1000}$ and a semiprime like $p \cdot q$ might both be multiplicative objects, but their internal architectures are wildly different.

To measure this, define the normalized distribution of a number’s prime factors:

w_i = \frac{\alpha_i}{\Omega(V)}

Then apply Shannon entropy:

\iota(V) = -\sum_i w_i \log w_i

$\iota(V)$ measures how evenly the multiplicative structure is distributed across its prime atoms.

For $2^{1000}$ , there is only one atom type, so $\iota = 0$ . The interaction is completely concentrated and lacks diversity.
For a balanced product $p \cdot q$ , $w_1 = w_2 = 0.5$ , yielding $\iota = \log 2$ . The structure is distributed and possesses more richness.
For $2 \cdot 3 \cdot 5 \cdot 7 \cdot 11$ , $\iota = \log 5$ , reflecting a highly diverse atomic distribution.

Axis 3: Archaeological Cost, $\mathrm{Arch}$ #

While $\Omega$ and $\iota$ are ontological, meaning they exist the moment the number exists, Archaeological Cost is epistemic.

Let $k$ be the digit length of $V$ . $\mathrm{Arch}_b(V)$ is defined as the worst-case bit-operation count required by the fastest known public deterministic factoring algorithm to recover the interaction atoms from the raw digit string of $V$ :

\mathrm{Arch}_b(V) = \inf_{A \in \mathcal{A}_{\mathrm{public}}} \mathrm{Arch}_b(V; A)

Arch measures the computational pain of extracting the truth from the representation topology.

Because computing Resistance requires factorization, Arch acts as the energy barrier between the outward representation and the inner history. It is entirely non-canonical, meaning it drops the moment a better algorithm, like Shor’s algorithm on a quantum computer, is discovered.

4. The Concealment Gap: Where Cryptography Lives#

The separation of Resistance and Archaeological Cost shatters a deeply ingrained intuition in mathematics: the assumption that more structure equals more difficulty.

Consider two numbers.

$2^{1000}$ has enormous Resistance: $\Omega = 1000$ . It contains a massive amount of hidden interactions. Yet its Archaeological Cost is microscopic. Trial division by 2 unravels the entire history instantly.

RSA-2048, where $N = pq$ , has microscopic Resistance: $\Omega = 2$ . It consists of merely two atoms. Yet its Archaeological Cost is astronomically high, pushing the limits of the best sub-exponential algorithms known to humanity.

Many atoms can be easy. Few atoms can be hard.

The true metric of cryptographic hardness is not the size of the number or the depth of its structure, but how perfectly that structure is hidden.

Resistance counts atoms. Archaeology measures recovery pain. Cryptography lives in the gap.

A Worked Prediction: Why Cunningham Numbers Are Easier to Factor#

The Concealment Gap is not merely a restatement of known cryptographic intuition. It generates a testable, quantitative prediction that the standard $(\Omega, \mathrm{Arch})$ framework does not make explicit.

Consider two 200-digit semiprimes, each with $\Omega = 2$ :

$N_1 = (2^{331} + 1) / 3$ , a Cunningham number of the form $b^n + 1$ with small base and large exponent.
$N_2 = p \cdot q$ , where $p$ and $q$ are randomly chosen 100-digit primes with no algebraic relationship.

Both numbers have identical Resistance, $\Omega = 2$ , and identical Interaction Entropy, $\iota = \log 2$ . Under any framework that measures only the ontological content of a number, these two objects are indistinguishable.

Yet their Archaeological Costs differ by a sub-exponential factor.

The General Number Field Sieve factors $N_2$ with heuristic complexity:

L_N[1/3, (64/9)^{1/3}] \approx L_N[1/3, 1.923]

But $N_1$ admits a Special Number Field Sieve attack at:

L_N[1/3, (32/9)^{1/3}] \approx L_N[1/3, 1.526]

because the algebraic form $2^{331} + 1$ hands the attacker a low-degree polynomial with tiny coefficients. The sieving norms are dramatically smaller, and the probability that they factor over the factor base is correspondingly higher.

In practice, SNFS has been used to factor Cunningham numbers 30 to 40 digits larger than the largest GNFS factorizations of the same era.

The Multiplicative Archaeology framework explains why in a single sentence: the representation topology $R$ of a Cunningham number is algebraically structured, and that structure leaks into the Archaeological Cost.

The digit string of $2^{331} + 1$ is not a random 200-digit integer. It is the output of recursive coupling, exponentiation, applied to the smallest possible base.

The interaction history $I$ is shallow and regular, a single chain of squarings, and that regularity is visible in the representation. The attacker does not need to discover the history. The representation confesses it.

For a random semiprime $N_2$ , the interaction history is equally shallow, $\Omega = 2$ , but the representation is a structureless digit string that reveals nothing about the two primes that produced it.

The Concealment Gap:

\frac{\mathrm{Arch}(N_2)}{\mathrm{Arch}(N_1)}

is therefore not a function of Resistance or Interaction Entropy. It is a function of how much of the interaction history $I$ is leaked by the representation topology $R$ .

This yields a general prediction: for any family of integers whose representation topology preserves algebraic structure from the interaction history, the Archaeological Cost will be strictly lower than for a random integer of the same size and Resistance.

Mersenne numbers $(2^p - 1)$ , Fermat numbers $(2^{2^n} + 1)$ , and numbers defined by linear recurrences, such as Fibonacci and Lucas numbers, all satisfy this condition. Empirically, all are easier to factor than random integers of comparable size, via SNFS or specialized ECM strategies.

The framework does not merely describe this fact after the fact. It predicts it from the structure of the triad $(V, I, R)$ .

5. Renormalization Group Flow in Digit Topology#

This framework translates directly into the language of theoretical physics, specifically the Renormalization Group. Digit topology already treats a number as a 1-dimensional lattice of interacting digits. By coarse-graining these graphs, we can watch the arithmetic parameters flow exactly as physical couplings flow under scale transformations.

Blocking the Lattice#

In a real-space RG step, we group $k$ sites on a spin chain to form a superspin. In digit topology, we fence every $k$ digits of a number and treat each block as a meta-digit in base $b^k$ .

As we zoom out, we must renormalize the coupling constants.

Addition, or carry probabilities. Carries can only jump a boundary if the inside sum is at least $b^k$ . Thus, blocking creates a renormalized carry probability $p' = f_k(p)$ .

Multiplication, or digit potentials. Within a block, digits exhibit full pairwise coupling: an effective intra-block potential. Across blocks, the interaction strength drops by a factor of $b^k$ . High-distance couplings rapidly die out, mirroring irrelevant operators in quantum field theory.

Relevant, Irrelevant, and Marginal Observables#

As we coarse-grain a number by taking successive roots or by blocking, its prime atoms exhibit distinct physical behaviors.

Irrelevant observables: Even prime exponents, or perfect squares, behave like massive modes that decouple. They annihilate under square-root blocking and shrink under the RG flow.
Relevant observables: The square-free kernel of the number, primes with odd exponents, persists indefinitely. It is an RG invariant that survives every coarse-graining step, dominating the infrared physics of the integer.
Marginal observables: Prime factors that perfectly pair up at the exact blocking scale $b^k$ hover on the knife-edge between extinction and survival.

Under this flow, the Archaeological Cost acts as a dangerously irrelevant operator. At the macroscopic scale, large $k$ , factoring a block becomes easier through trial division up to $b^{k/2}$ . But in the microscopic ultraviolet regime, where a single 1024-bit RSA block resides, the hardness coupling spikes violently.

The Resistance splits into frozen IR components and flowing UV components, mapping numerical arithmetic into a Wilsonian sandbox.

Scope and Limitations of the RG Analogy#

A note of mathematical honesty is warranted.

The RG flow described above is a heuristic analogy, not a proven equivalence. In a Wilsonian RG, the flow acts on a space of Hamiltonians parameterized by coupling constants, and the classification of operators as relevant, irrelevant, or marginal is derived from the eigenvalues of the linearized flow at a fixed point.

The digit-topology construction defines a coarse-graining map, blocking $k$ digits into a meta-digit in base $b^k$ , and identifies quantities that grow, shrink, or remain invariant under that map. But it does not yet establish a formal Hamiltonian, a coupling-constant space, or a proof that the linearized flow at the round-number fixed point has the spectral structure required to rigorously classify operators.

What the analogy does provide is a precise vocabulary for three empirical observations that are otherwise stated only informally.

The square-free kernel of an integer is invariant under any multiplicative coarse-graining: taking $k$ -th roots, modular reduction, or digit blocking. This is the relevant-operator claim, and it is provable. The square-free kernel $\mathrm{rad}(n) = \prod_{p \mid n} p$ is a multiplicative function that is invariant under exponent reduction modulo 2.
Perfect-power components vanish under root extraction. This is the irrelevant-operator claim: if $n = m^2$ , then $\sqrt{n} = m$ , and the squared structure is annihilated in one step.
The Archaeological Cost of a blocked number decreases with block size, since trial division up to $b^{k/2}$ becomes feasible. But the cost of the unblocked number at the UV scale remains sub-exponential. This is the dangerously-irrelevant claim, and it is consistent with the known complexity of GNFS, but it is not derived from a beta function.

Elevating this analogy to a theorem would require defining a metric on the space of digit-topology configurations, proving that the blocking map is a contraction in that metric, and showing that the fixed-point spectrum reproduces the relevant/irrelevant classification from first principles rather than by inspection.

That program remains open. The analogy is offered here as an organizing principle, a way to see why certain arithmetic quantities persist under scale change and others do not, rather than as a claim of mathematical equivalence with quantum field theory.

6. Inverse Coupling, Momentum Shells, and the Imaginary Generator#

If multiplication is global coupling, generator to artifact, then taking a square root points in the opposite causal direction: artifact to generator.

A square root is not simply an algebraic inverse. It is the physical act of self-generator recovery. It asks: what topology’s self-interaction produced this exact interaction trace?

The Algorithm as Momentum-Shell RG#

The traditional long-hand square root algorithm mirrors an RG momentum-shell extraction.

When extracting a root manually, we group digits into pairs because a squared number forces each new generator digit to couple twice, left and right. In each step:

We inspect a remainder cluster and extract the largest single digit $d$ whose local self-interaction bubble $d^2$ fits inside.
We peel off this trace, leaving an unexplained remainder.
We bring down the next block, diffusing the trace.
We test a new coupling stripe, $(20G + d)d \le \mathrm{remainder}$ , where $20G$ is the established stem and $d$ is the new node.

Every time we peel off a perfect square chunk and rescale the remainder by $10^2$ , we are actively stripping a high-energy shell and iterating the RG flow.

A perfect square leaves no irrational fringe. Its entire graph decomposes neatly. A non-square leaves a trace that cannot be resolved in integer space, forcing a descent into the decimal lattice.

The Origin of $i$ in Digit Space#

This physical approach to inverse coupling reframes the ontological mystery of complex numbers.

In digit topology, digits are strictly non-negative, and multiplication produces structured magnitude-based traces. Therefore, evaluating $\sqrt{-1}$ is not merely mathematically undefined. It is structurally undefined in digit space.

There is no possible digit configuration whose self-coupling leaves a trace of -1.

When inverse-coupling fails, we cannot simply invent an inverse. Instead, we must enlarge the target space. The imaginary unit $i$ is not an inverse output. It is a new primitive generator introduced specifically to carry an orthogonal interaction mode.

Digits encode magnitude-based interaction. $i$ encodes phase-based interaction that sign oscillations cannot achieve.

When we write:

(a + bi)(c + di)

we are establishing a two-layer digit topology that couples real magnitude channels with imaginary phase channels, generating cross terms like $bc$ and $ad$ .

Thus, $\sqrt{-1}$ acts as the trigger that forces this new generator to exist, expanding the framework from a 1D lattice into a stable complex manifold to prevent the RG flow from crashing.

7. Additive Engines and Multiplicative Cascades#

While multiplication creates deep structural archaeology, addition acts merely as a local pulse. However, when local additive rules are iterated recursively, they can spawn staggering multiplicative complexity.

The ultimate showcase is the Fibonacci sequence.

The recurrence:

F_{n+1} = F_n + F_{n-1}

is purely additive, yet its interaction traces behave like wild fractal skylines.

As $n$ grows, the carry patterns approximate multiplication by the golden ratio $\phi \approx 1.618$ , creating a repeating standing wave of local pulses.

Under an RG lens, coarse-graining a Fibonacci block is like moving one tick along the sequence, revealing $\phi$ as the golden RG attractor.

Though built purely from radius-1 local updates, the sequence acts as a cryptographic playground. Factoring a Fibonacci number $F_n$ is notoriously difficult. Its Resistance, or prime-atom budget, oscillates wildly, and its interaction entropy $\iota$ bounces between concentrated and highly distributed modes, masking low-Resistance terms with gigantic Archaeological Costs.

It demonstrates how a predictable, linear representational staircase can hide an aggressively unpredictable interaction history.

8. A Universal Template for Interacting Networks#

The beauty of the Digit Topology to Renormalization package is that it is fundamentally a domain-agnostic recipe.

By defining primitive nodes, a gluing interaction, an inverse archaeology, and the three observables $(\Omega, \iota, \mathrm{Arch})$ , the entire prime/entanglement theory ports to other fields.

Chemistry. Atoms are elements: C, H, O. Multiplication is a synthesis reaction. Taking a square root is stoichiometric decomposition. Resistance is the total atom count in an empirical formula, while Arch is the laboratory effort needed for elemental analysis. Coarse-graining combusts a molecule to its baseline elements, leaving the element-inventory fixed point.

Functional programming. Primes are pure leaf functions. Multiplication is composing call graphs. Square root extraction is inlining and beta-reduction. Resistance is the number of reachable leaf functions, and Arch is the computational cost of static analysis.

Neural networks. Primes are single neurons. Multiplication is dense layer connectivity. Root extraction is model pruning or weight decomposition. Resistance counts irreducible neurons, and Arch is the FLOPs required to find the minimal factorization, for example via SVD. Residual blocks that self-copy across scales are RG fixed-point subnetworks.

Music theory. Primes are individual pitches. Multiplication stacks intervals. Root extraction finds the generative bassline. A complex jazz chord has high Resistance, while the ear-training difficulty required to identify its tones represents Archaeological Cost.

The 5-Step Recipe#

Pick a set of primitive nodes, or atoms. Identify the foundational building blocks of your network that cannot be broken down any further. In arithmetic, these are the prime numbers, which act as irreducible interaction vertices.
Identify a binary interaction, or glue. Find the operation that fuses two primitive nodes together to create a new, larger aggregate or node. In arithmetic, this is multiplication.
Define the inverse interaction, or archaeology. You need a mechanism to partially or fully reverse the binary interaction to recover the original components from the aggregate. In arithmetic, this is square-root extraction or factorization.
Define the three observables. Resistance $\Omega$ counts primitive atoms. Archaeological Cost $\mathrm{Arch}$ measures the effort required to recover hidden atoms from the aggregate’s outward appearance. Interaction Entropy $\iota$ measures how evenly distributed those atoms are.
Coarse-grain and observe the flow. Define a zoom-out rule where you block nodes into meta-nodes. Then watch how the network’s information behaves under RG flow: which elements stay fixed, which ones change in importance, and which ones wash away as irrelevant details.

Conclusion: The Geometry of Buried History#

The framework of Multiplicative Archaeology does not invent new mathematics. It reinterprets ancient functions by pulling them into a coherent coordinate system. It forces us to recognize that $V$ , $\Omega(V)$ , and $\mathrm{Arch}(V)$ are orthogonal axes.

By shifting the definition of $\Omega(V)$ from a mere number of prime factors to the amount of multiplicative interaction history encoded in the value, arithmetic ceases to be a static line of scalars. It transforms into a dynamic, discrete geometry composed of generators, interaction traces, and recoverable histories.

When we view arithmetic through this lens, we realize that structure does not inherently equal difficulty, that primes are the irreducible atoms of causality, and that cryptography thrives in the dark epistemic space between what mathematically exists and what is computationally recoverable.

The hardest problems in number theory are no longer abstract numerical puzzles. They are excavations into the deeply buried, perfectly concealed history of digital topology.

We opened with a declaration: arithmetic should remember how it was constructed.

We close with a question that Poincare asked in 1904 and Perelman answered a century later: what happens when a space remembers nothing?

In the next article, we apply the 5-step recipe to closed 3-manifolds. The primitive atoms are loops. The interaction is homotopy. The archaeology is Ricci flow. And the deepest result in modern topology becomes a single statement in this framework:

When the Resistance of a closed 3-manifold is zero, the manifold is the 3-sphere. The Concealment Gap vanishes. There is nowhere left to hide.