The Methodology Reference

Probability Walk Design: From Narrative to Model

Status: Design document. Defines the precise structure of forward/reverse probability walks, the corridor intersection, bottleneck identification, and the multi-domain coupled walk — the computational model that replaces narrative storytelling for abiogenesis and other domain trajectories.

This is the critical next step. Getting this right gives us an actual model, not just narrative.

1. The Core Idea

We are NOT doing a single walk through a single lattice. We are tracking MULTIPLE simultaneous walks across coupled domains, where manifestation (actual physical realization) requires alignment across all domains at specific critical points.

The abiogenesis walk involves at minimum:

Each domain has its own lattice, its own walk, its own branching structure. But they're COUPLED — a move in one domain may enable or require moves in others. The conditional dependencies (like Dep(R≥1.7, Mem≥1)) are the coupling points.

2. What We're Computing

2.1 Single-domain reachability

For a single domain lattice:

Forward reachability set F(k): All coherent positions reachable from origin in exactly k one-step moves.

F(0) = {origin}
F(k) = {pos : pos is coherent AND exists prev ∈ F(k-1) where pos is one step from prev}

This is a BFS from origin through the coherent sub-lattice. F(k) GROWS as k increases — more positions become reachable.

Backward reachability set B(k): All coherent positions from which the endpoint is reachable in exactly (max_steps - k) one-step moves.

B(max_steps) = {endpoint}
B(k) = {pos : pos is coherent AND exists next ∈ B(k+1) where next is one step from pos}

This is a BFS from endpoint BACKWARD through the coherent sub-lattice. B(k) GROWS as k decreases (going further from the endpoint).

Corridor C(k): The intersection at each step.

C(k) = F(k) ∩ B(k)

Positions that are BOTH forward-reachable from origin at step k AND can reach the endpoint from step k. C(k) is always ≤ min(|F(k)|, |B(k)|).

Corridor width W(k) = |C(k)| — the number of positions in the corridor at step k. Where W(k) drops sharply, there's a BOTTLENECK.

2.2 Why bottlenecks matter

A bottleneck at step k means: of all the positions reachable from the origin, and of all the positions that can lead to the endpoint, very few are in BOTH sets. The walk MUST pass through this narrow gate.

Bottleneck causes:

2.3 The probability layer

The reachability analysis gives the STRUCTURE — which positions are in the corridor, where the bottlenecks are. The probability layer adds WEIGHTS:

Uniform model (what we have now): Each coherent neighbor is equally likely. Corridor width = effective probability width.

Physics-weighted model (what we want): Each transition has a rate based on:

With weights, some corridor positions become much more probable than others. The "effective corridor width" (2^entropy) may be much smaller than the raw position count.

3. The Multi-Domain Coupled Walk

3.1 The unified manifestation perspective

At any point in time, we have a POSITION in each domain simultaneously:

State(t) = (R_pos(t), G_pos(t), Cmp_pos(t), Cd_pos(t), P_pos(t), Ctx_pos(t))

Each domain walks its own lattice, but they're coupled:

3.2 Potential moves at any state

At state S(t), the potential moves are:

PotentialMoves(S(t)) = {
    for each domain d:
        for each one-step move m in domain d's lattice from pos_d(t):
            if m is coherent within domain d:
                if all cross-domain dependencies are satisfied:
                    yield (d, m)
}

Some moves are BLOCKED by cross-domain dependencies. You can't advance R to R1.7 unless Cmp is at Cmp1. You can't advance Cd to Cd2 unless R is at R2. The cross-domain coupling creates additional gates beyond the within-domain coherence constraints.

3.3 Manifestation probability

Manifestation is the question: at this state, what's the probability that this configuration is actually physically realized?

This depends on:

  1. Reachability: Can you GET here from the origin? (Forward reachability)
  2. Viability: Can you SUSTAIN this state? (The state doesn't collapse — lower-level structures remain stable)
  3. Rate: How FAST can you get here? (Physics rates × time available)
  4. Stability: How long does this state PERSIST before being disrupted? (Thermodynamic stability)

A state that's reachable but not viable collapses. A state that's viable but slow may not be reached in the time available. A state that's reached but unstable is transient.

3.4 The collapse condition

The user's key insight: if at any point the lower-level structures collapse, the whole walk collapses.

Viable(S(t)) = all(
    for each domain d:
        pos_d(t) is sustainably manifested given the physical context
)

If Cmp drops below Cmp1 (the vesicle pops), the bootstrap loop fails, R regresses, the walk collapses back to an earlier state. The walk doesn't just need to REACH R2 — it needs to SUSTAIN everything below R2 long enough for the code to crystallize.

This means the walk has MEMORY — the current state depends on the sustained viability of all lower states. It's not just "where am I at step k" but "can I maintain my position."

3.5 The R2 crystallization threshold

R2 is special: once the code crystallizes, it's PERMANENT. Before R2, the walk is reversible — any state can collapse. After R2, the code is frozen, the evaluator is deterministic (Kd4), and competitive exclusion begins monopolizing the substrate. The walk becomes IRREVERSIBLE.

The critical trajectory is:

  1. Walk from R0 toward R2 through a series of states, each requiring sustained viability
  2. At each step, there's a probability of collapse (losing a lower-level structure)
  3. The walk must reach R2 AND sustain all prerequisites long enough for crystallization
  4. Once crystallized, the system is locked — no more collapse risk for the code itself
  5. Post-R2, the system must sustain until the ecosystem is self-supporting (proliferation)
  6. Once proliferation reaches sufficient density, competitive exclusion makes the biosphere permanent

3.6 What this model provides

If we build this correctly:

Structural predictions:

Quantitative estimates (with physics rates):

Counterfactual analysis:

4. The Computational Architecture

4.1 Data structures needed

Domain:
  primitives: list of (id, levels, partial_levels)
  internal_dependencies: list of (from, to) constraints within domain

CrossDomainCoupling:
  constraints: list of (domain_a, primitive_a, min_level, domain_b, primitive_b, min_level)
  
UnifiedState:
  positions: dict of domain_id → position_tuple
  
Walk:
  states: list of UnifiedState over time
  moves: list of (domain, primitive, from_level, to_level) at each step

4.2 Algorithms needed

  1. Single-domain forward/backward BFS — already have this (probability.py), just need to separate reachability from probability weighting

  2. Corridor computation — intersection of forward and backward sets at each step, with width tracking

  3. Bottleneck identification — find steps where corridor width drops sharply, identify which dependencies cause the narrowing

  4. Multi-domain state enumeration — the product of positions across all coupled domains, filtered by cross-domain constraints

  5. Coupled walk simulation — step through the multi-domain state space, tracking which moves are available, which are blocked by cross-domain constraints, and which states are viable

  6. Collapse probability estimation — at each state, estimate the probability that a lower-level structure fails (based on stability/sustainability of each domain's position)

4.3 Implementation sequence

Phase 1: Single-domain corridor (immediate next step)

Phase 2: Bottleneck analysis

Phase 3: Multi-domain coupled walk

Phase 4: Physics weighting

5. The Abiogenesis Application

5.1 The specific domains and their coupling

R (Translation): R0 → R0.1 → R0.2 → R0.5 → R1 → R1.3 → R1.7 → R1.9 → R2
G (Genome):      G0 → G0.5 → G1 → G2 → G3
Cmp (Compartment): Cmp0 → Cmp0.5 → Cmp1 → Cmp2
Cd (Code):       Cd0 → Cd0.5 → Cd1 → Cd2
P (Population):  P0 → P0.5 → P1 → P2
Ctx (Context):   Ctx0 → Ctx1 → Ctx2

5.2 Known cross-domain gates

From the abiogenesis analysis:

GateConditionWhy it's narrow
Bootstrap entryR≥1, G≥1Proto-ribosome needs template RNA — co-arising
Bootstrap thresholdR≥1.7, Cmp≥1, P≥0.5Parasite control requires compartmentalization AND some population structure
Code crystallizationR≥2, Cd≥2, G≥2Code freezing requires deterministic translation AND sequential heredity
Competitive exclusionR=2, Cd=2, P≥1, Cmp≥2Monopolizing the substrate requires all of: frozen code, population, self-assembled boundary
Ecosystem viabilityAll above SUSTAINED + Ctx stableThe walk is only permanent if the ecosystem reaches self-sustaining density

5.3 What the model would show

For abiogenesis, the model would produce:

  1. The corridor plot: Width vs time (R sub-level), showing the funnel shape with bottlenecks at the bootstrap threshold (R1.7) and code crystallization (R2)

  2. The gate analysis: For each bottleneck, which cross-domain conditions must be met, how narrow the gate is, and what determines its width

  3. The collapse risk profile: At each state, the probability that a prerequisite structure fails. Highest before R2 (everything is reversible), drops to near-zero after crystallization + ecosystem establishment

  4. The critical path: The sequence of moves through the multi-domain state space that minimizes total collapse risk — the "most probable actual trajectory"

  5. Comparison with literature: The model's bottleneck predictions vs known abiogenesis milestones (PTC symmetry at R1, two aaRS classes at R1.9, LUCA at 4.2 Gya)

5.4 What makes this more than narrative

The narrative says: "the bootstrap loop accelerates, compartmentalization is required, the code crystallizes." The model says:

These are structural predictions, not stories. They can be compared to empirical data. They can be wrong (and thereby informative).

6. The Unified Manifestation Model

6.1 What we're actually tracking

We are NOT tracking abstract lattice positions. We are tracking UNIFIED MANIFESTATIONS — specific physical configurations where multiple domains are simultaneously realized at specific levels. A unified manifestation is a concrete thing: a population of RNA molecules in a mineral micropore with a specific chemistry, at a specific temperature, with a specific set of catalytic capabilities.

The probability comes from asking: what's the likelihood that this manifestation exists? That depends on:

6.2 Fixed points and uncertainty

We have KNOWN POINTS — things we know happened, backed by empirical evidence:

Fixed pointWhat we knowEvidence
R0Prebiotic chemistry existedMiller-Urey, meteorites, hydrothermal synthesis
~R0.2Aminoacylated RNA existedStereochemical affinity (Yarus), ribozyme aminoacylation (Suga)
~R1Proto-ribosome existedPTC pseudo-2-fold symmetry (Yonath), RNA machine fossil
~R1.7Compartmentalization requiredEigen error threshold, parasite problem
R2Genetic code crystallized — singular eventUniversality of the code, LUCA reconstruction
Post-R2Sustainable population establishedAll extant life descends from this

Between the fixed points: UNCERTAINTY. The structural analysis constrains what COULD have happened (the corridor). The physics rates constrain how FAST each step could occur. The empirical evidence narrows the distribution. But between fixed points, we have a probability distribution, not a single path.

6.3 The scope of the walk

The walk we're modeling:

START: Prebiotic chemistry in hydrothermal vent micropores
  ↓ (chemistry explores molecular space — population of micropores)
R0 → R0.1 → R0.2: stereochemical associations, aminoacylated RNA
  ↓ (template-directed synthesis emerges — narrowing search)
R0.5: Pre-ribosomal translation
  ↓ (evaluator separates from encoding — SSA architecture appears)
R1: Proto-ribosome
  ↓ (bootstrap loop — co-evolution of R and P)
R1.3 → R1.7: Bootstrap threshold + compartmentalization gate
  ↓ (code expansion, two aaRS classes, protein enzymes)
R1.9: Near-complete code
  ↓ (code crystallizes — singular event, one cell/vesicle)
R2: Genetic code frozen — IRREVERSIBLE
  ↓ (tangent set EXPLODES — many new capabilities now available)
  ↓ (R2 population reproduces, adapts, evolves replication machinery)
END: Sustainable R2 population — self-sustaining, no longer dependent on 
     a single lineage surviving. Multiple R2 organisms, heritable variation,
     Darwinian evolution operating. The cascade is established.

We stop at "sustainable R2 population" — NOT at prokaryotic Earth or biosphere saturation. Those are important but they're the next walk (R2 population → ecosystem diversification → biosphere saturation). The scope for this model is: prebiotic chemistry → code crystallization → sustainable population.

6.4 Post-R2: from singular event to sustainable population

R2 is a singular event — one cell/vesicle achieves it. But one cell is fragile. The walk isn't complete until:

  1. R2 cell reproduces — vesicle-level reproduction already exists from R1.7. Post-R2, it improves (protein-based replication machinery, coordinated division). This is probably NOT a separate gate — it evolves smoothly under selection during and after crystallization.

  2. R2 population grows — the R2 cell's efficiency advantage (enzyme-catalyzed metabolism >> mineral-catalyzed chemistry) drives rapid growth. The vent environment provides energy and nutrients. Growth is fast once R2 is achieved.

  3. R2 population must not exhaust local resources — this is a real risk. But the vent provides continuous energy input (serpentinization), so resource exhaustion is mitigated as long as the vent is active.

  4. R2 population must reach diversity — some genetic variation among R2 descendants. This enables adaptation to changing conditions and resistance to environmental perturbation. Diversity comes naturally from replication errors + selection.

  5. Sustainable — the population persists even if any single lineage dies. Requires: multiple individuals, heritable variation, environmental stability on relevant timescale (vent lifetime ~30K+ years is sufficient).

The post-R2 trajectory is probably SMOOTH — no additional narrow gates. R2's tangent set explosion means many new capabilities are accessible. The main risk is environmental: if the vent fails before the population disperses, the lineage dies. But given millions of years and thousands of vents, this is a numbers game — one successful vent is enough.

6.5 The tangent set explosion at R2

At R2, the available moves EXPAND dramatically. Pre-R2, the system is constrained to improving translation fidelity and code assignments. Post-R2, the system can:

Each of these is a NEW capability that was structurally impossible pre-R2 (required proteins that couldn't be reliably produced). Post-R2, they're all accessible. The tangent set goes from ~3-5 moves (pre-R2: improve ribosome, expand code, improve compartment) to MANY moves (post-R2: any protein-based capability).

This IS the "downstream biology lattice is unlocked" from the analysis. R2 is the gate; everything after it is a garden of forking paths.

6.6 How probabilities enter the model

The model is NOT trying to compute the absolute probability of abiogenesis. It happened — we know that with certainty because we exist. The model provides RELATIVE probabilities and structural constraints:

Structural probabilities (from the lattice):

Physics-informed probabilities (from domain knowledge):

Bayesian updating:

The model is ADAPTIVE. As new data comes in (new experiments, new fossils, new theoretical insights), the probabilities update. The STRUCTURE (lattice, dependencies, gates) changes less often — only when the domain analysis itself is revised.

7. Open Questions

  1. How to handle the product state space size. Even with 6 domains × 3-9 levels each, the product space is 9×5×4×4×4×3 = 8,640. Manageable. But adding sub-levels makes it larger. Need to decide: compute exactly (brute force, feasible at this size) or approximate (sample, needed if we go finer).

  2. How to define collapse probability. What makes a state "at risk of collapse"? Options: (a) structural — states with few dependencies satisfied are fragile, (b) thermodynamic — states requiring sustained energy input collapse when input fluctuates, (c) information-theoretic — states near the Eigen error threshold collapse if fidelity drops. Probably all three contribute.

  3. How to source transition rates. For physics-weighted walks, we need rates for each transition. Some are known from literature (nucleotide polymerization rates, vesicle formation rates). Others are structural estimates (search space for a functional ribozyme ~10^20 sequences). This is where the methodology meets empirical chemistry.

  4. How to validate. The model produces predictions (bottleneck locations, transit time estimates). How do we test them? Against: (a) the geological record (timing of abiogenesis ~4.4-3.8 Gya), (b) experimental origin-of-life chemistry (which steps have been demonstrated), (c) comparative genomics (what LUCA reconstruction tells us about early conditions).

  5. Generalization to other domains. The same framework should apply to: the QG→QM transition (physics domains coupling), the biology→cognition transition (neural + cognitive domains coupling), and the entity system's adoption trajectory (computing + ecosystem domains coupling). Building it right for abiogenesis means it works everywhere.

7. Relationship to Other Work

This design document feeds:

Referenced by the model

Cited as a source by 5 model records (browse the model census):