Landscape Bootstrap Methodology — Stage 2 Synthesis

Purpose: Codify what we have learned from bootstrapping the biology arrangement landscape (Waves 0-5 + Full Unified Manifestation extension), supplementing what was already established from the entity arrangement buildout.

Position relative to canonical strategy: The existing canonical-architecture-strategy.md documents the within-arrangement discipline (4 layers, scope ladder Sc0-Sc4, 10 discipline rules). This doc documents the cross-arrangement buildout discipline — how to build out a landscape in a new arrangement, evaluate it, identify anchors, and validate the methodology.

Status: Draft synthesis based on biology + entity work. Some sections are mature (Rules 1-10, singleton triage, calibration discipline); others identify gaps (Rule 11-12 implications, projected-analysis tooling) that the next session may extend.

Part I — Preamble: What we are doing and why

The framework we are working with reduces information systems to a small number of primitives organized into chain levels (Sc0=substrate ground → Sc1=arrangement → Sc2=region → Sc3=instance → Sc4=cross-arrangement). Each instance manifestation is scored at each primitive at each chain level of its arrangement.

The landscape methodology answers: given a NEW arrangement (domain), how do you build out a manifestation corpus that reveals structural patterns, and how do you validate that the patterns you surface are real?

Two arrangements have been bootstrapped to maturity:

1. Entity arrangement — software systems; Phase 1A-1C complete with 10 inductive + declarative anchors over ~50 instance Mns + 12 cross-domain compositions/scenarios (final).
2. Biology arrangement — extant organisms; Waves 0-5 + Full Unified Manifestation extension complete with 12+ anchor candidates over 54 instance Mns (this doc).

A third spot-check (methodology arrangement) is underway in earlier work.

The cross-domain claim: the same analytical recipe (per-chain-level scoring + cluster_classical + cluster_stability + cluster_meta_stability + score_archetypes) applied to both arrangements produces coherent spine inventories of comparable shape. The framework generalizes across substantively different domains.

Part II — The 12 buildout rules

Each rule with: statement, evidence anchor, and how to apply.

Rule 1 — Don't cluster at low corpus density

Statement: Cluster analysis below ~30-40 instance Mns produces an inventory based on which categories happened to have 3 members in the early corpus, not on what the domain structures around. Anchor authoring at low density crystallizes a transient configuration.

Evidence: Biology spot-check at N=21 (Wave 0) surfaced 5 candidate anchors each with 3 members; the final inventory at N=54 (Wave 5 + Full Unified extension) bears only partial resemblance, with several Mns having migrated and several "new" zones surfaced. The early candidates over-counted because the corpus was selection-biased toward back-end model organisms.

How to apply: treat early-density clusters as gap-finders, not as anchor candidates. Defer anchor authoring until clustering has been stable across 2 consecutive waves of corpus expansion.

Rule 2 — Disciplined calibration before authoring

Statement: Before scoring a new Mn, sample 3-5 existing structurally-similar Mns. Identify the tier the new Mn occupies. Score it AT the tier unless there's a primitive-specific reason to differ. Reserve differences for actual distinguishing features.

Evidence: Wave 1 (no calibration discipline): 1 of 5 hypotheses confirmed. Waves 2-5 (calibration discipline): 13 of 13 hypotheses confirmed. The discipline turns the methodology from "buildout where authoring artifacts dominate" into "buildout where clustering is informative."

How to apply: every new Mn authoring session should begin with grep + read of 3-5 neighbor Mns, an explicit tier identification, and a justification for each primitive that deviates from tier-default.

Rule 3 — Use cluster output as gap-finder, not category lookup

Statement: Singletons in cluster output show where the corpus is sparse, not where structural attractors are. Treat singletons as targets for next-wave expansion, not as findings.

Evidence: In biology, 5 of 6 Wave-0 singletons (amphimedon, marchantia, neurospora, tetrahymena, yeast — coprinopsis added in Wave 1 partially resolved one) classified as coverage artifacts under inspection. The "vertebrates" Wave-0 singleton was a schema artifact (sc=2 aggregate). After Waves 2-5 most resolved.

How to apply: at end of each wave, list singletons. For each, predict whether it's coverage-artifact / authoring-artifact / structural-attractor / outlier-of-category. Use those predictions to plan the next wave.

Rule 4 — Threshold sweeps reveal more than single-threshold

Statement: Meta-stability at a single threshold misses information. Strict thresholds (e.g., 8/9 or 12/15) capture core clusters; moderate thresholds (5-6/9 or 9-10/15) capture convergent-structural zones across phylogenetically-distant lineages.

Evidence: Biology's "minimal-multicellular zone" (amphimedon + coprinopsis + trichoplax across 2 kingdoms) and "colonial-multicellular zone" (chlamydomonas + dictyostelium + volvox across 2 lineages) appear at moderate thresholds but not strict. These are the convergent-evolution findings the framework is designed to surface.

How to apply: report meta-stability at strict + moderate + (sometimes) relaxed thresholds. Anchor authoring requires strict-threshold persistence; convergent-zone findings are reported at moderate.

Rule 5 — Singleton triage has four classes

Statement: Singletons fall into four diagnostic classes:

How to apply: triage every persistent singleton across at least 2 waves before declaring it a structural attractor. Authoring artifacts get re-scored; coverage artifacts get neighbor-added; outlier-of-category insights become findings.

Rule 6 — Cluster membership is corpus-state-dependent

Statement: Existing Mns can MIGRATE between spines as corpus expands. The centroid of each spine shifts; Mns that previously fit one spine may match a different one when new members arrive.

Evidence: In biology, strongylocentrotus migrated bilaterian-invertebrate → basal-eumetazoan-ND2 (Wave 3) → planaria-pair (Wave 4) → basal-eumetazoan-ND2 (Wave 5). c-elegans dropped to singleton (Wave 4) and returned to bilaterian spine (Wave 5) when hypsibius added.

How to apply: anchor authoring should be done on a corpus that has been STABLE across at least 2 consecutive waves. Authoring at a mid-buildout state crystallizes a transient configuration.

Rule 7 — Pair-tightening also occurs under expansion

Statement: Sometimes more data NARROWS clusters: 3-member moderate-threshold zones can become 2-member strict pairs + 1 singleton when corpus expands. The reverse of "more data = better resolution."

Evidence: Biology's amphimedon-coprinopsis-trichoplax moderate-triple (Wave 3) → amphimedon-trichoplax strict pair + coprinopsis singleton (Wave 4) → coprinopsis joins schizophyllum strict pair (Wave 5).

How to apply: treat moderate-threshold triplets as "either tightens into pair + ejected singleton" OR "promotes to strict-triplet" depending on which structural feature the new corpus member tips the balance toward.

Rule 8 — The framework is chain-level-resolved AND role-resolved

Statement: Each chain level measures organismal grade at that level (substrate elaboration, organism architecture, ecological role). Convergent BEHAVIORAL patterns implemented through DIFFERENT ecological/structural roles SCATTER at the role chain level rather than converging — because the framework measures structural similarity, not behavioral analogy.

Evidence: The 3 eusocial Mns (apis-mellifera, macrotermes, heterocephalus) share the eusocial behavioral pattern (caste system + reproductive division of labor) but scatter at organism-to-ecosystem-bridge: apis joins the pollinator-mutualist cluster (with plants); macrotermes joins the decomposer cluster (with fungi); heterocephalus is a singleton. Their ecological-role IMPLEMENTATIONS dominate over their shared eusocial architecture.

How to apply: distinguish "convergent evolution" (same behavioral pattern across lineages) from "convergent structure" (same primitive-profile across lineages). The framework surfaces the latter, not the former. Patterns surfaced by the framework are STRUCTURAL convergence; patterns NOT surfaced may still be behavioral convergence that lives outside the primitive scoring.

Rule 9 — Cross-domain replication is the methodology validation

Statement: The same analytical recipe applied to a NEW domain should produce a coherent spine inventory. If it produces nonsense in a new domain, the methodology is broken. If it produces a clean inventory comparable in shape to existing domains, the methodology generalizes.

Evidence: Entity arrangement spine inventory (Phase 1C, N≈50): 4 large spines + 5 pair/triplet spines + ~10 singletons. Biology arrangement (Wave 5, N=54): 4 large spines + 6 pair/triplet spines + ~9 singletons (at strict, 3-chain-level scope). Shape comparable; silhouettes 0.30-0.55 (entity) vs 0.42-0.80 (biology — higher because biology has cleaner substrate-grade separability).

How to apply: every new arrangement should follow the same recipe (gap analysis → wave → re-cluster → triage → next wave). Spine inventory + singleton triage + threshold sweeps + anchor identification follow the same template. Deviations are domain-specific findings, not methodology defects.

Rule 10 — Convergent-structural-zones at moderate threshold are findings

Statement: When 3+ phylogenetically-distant lineages cluster at moderate threshold (not strict), that zone is a cross-lineage structural convergence finding — exactly what the partial-level grain is designed to surface.

Evidence: Biology has multiple such zones at moderate threshold:

• minimal-multicellular zone: amphimedon (sponge) + coprinopsis (fungus) + trichoplax (placozoan) — animals × 2 + fungi
• basal-eumetazoan-intermediate-bridge: hydra + mnemiopsis + nematostella + strongylocentrotus + lumbricus — across 4 phyla
• colonial-multicellular-small-genome-eukaryote: chlamydomonas + dictyostelium + volvox — green algae + Amoebozoa
• decomposer-recycler (at organism-to-ecosystem-bridge): fungi × 4 + lumbricus + macrotermes + methanococcus — animal + fungus + archaeon

How to apply: these zones are anchor candidates as "convergent structural attractors." They should be labeled by what they STRUCTURALLY share (minimal-multicellular grade, decomposer ecological role, etc.) and tagged with the threshold + chain-level + signature combination that surfaces them.

Rule 11 — Projected-primitive analysis is methodologically necessary

Statement: When the corpus is scored at multiple chain levels, integrated meta-stability becomes substrate-dominant — fine-grained ecological-role and behavioral structure gets averaged out. Per-chain-level analysis + per-primitive-subset projection are required to surface non-substrate structural regions.

Evidence: Biology meta-stability at 5 chain levels (substrate + bridge + surface + organism-to-ecosystem-bridge + ecosystem) at strict threshold MERGES mammals + non-mammalian vertebrates into one big N=27 spine. The within-vertebrate distinction is invisible. But at organism-to-ecosystem-bridge ALONE, four distinct ecological-role clusters surface (autotroph, decomposer, engineer, consumer). The two views are complementary.

How to apply:

• Report cluster analysis at each chain level separately, side-by-side
• Report integrated meta-stability across chain levels as a separate view
• Provide projected-primitive clustering (cluster on a user-specified subset of primitives) as a navigation tool
• Anchors should be TAGGED with the projection that reveals them

The anchor inventory is multi-view. Anchors are not "the truth"; they are "the structural region that surfaces under projection X."

Rule 12 — Unified manifestation is the scoring standard

Statement: Every Mn should be scored at ALL chain levels of its arrangement (excluding levels where the Mn is a documented outlier — e.g., a Mn that operates entirely above the substrate ground). Under-scoring renders certain analytical questions invisible.

Evidence: Biology Mns were initially scored at only 5 of 9 chain levels (chemistry through organism-architecture). This rendered organism-to-ecosystem-bridge phenomena (eusociality, mutualism, ecosystem engineering) and ecosystem-level phenomena (trophic role, biogeochemical cycling) invisible. Extending to 7 chain levels (adding organism-to-ecosystem-bridge + ecosystem) immediately surfaced four new ecological-role cluster regions.

How to apply: new arrangements should be scored at all chain levels from day one. Existing arrangements that are under-scored should be extended retrospectively. The "unified manifestation" standard is part of the canonical methodology, not optional.

Part III — The buildout protocol

The wave-by-wave protocol that produced the biology Wave 5 + Full Unified state:

Stage 0: Gap analysis + buildout plan
  - Current corpus inventory by clade/category
  - Coverage analysis: well-represented vs sparse regions
  - Singleton diagnosis (Rule 5 triage)
  - Target buildout: N target, structural coverage goals
  - Candidate Mns with rationale per candidate

Stage 1: Wave-driven corpus buildout
  Wave 1 — Singleton resolution (4-6 Mns)
    - Each Mn paired with a predicted-pair-partner
    - Hypothesis stated for each pair
    - Author with calibration discipline (Rule 2)
    - Validate, re-cluster, re-meta-stability
    - Findings doc: hypotheses vs results
  Wave 2 — Phylogenetic gap fill (5-7 Mns)
    - Fill major taxonomic/structural gaps
    - Same pattern
  Wave 3 — Structural probes (4-6 Mns)
    - Test specific hypotheses (boundary cases, convergent zones)
  Wave 4-5+ — Consolidation
    - Re-check Wave 1-3 hypotheses
    - Add Mns that resolve persistent singletons
    - Stabilize spine inventory

Stage 1.5: Full unified manifestation extension
  - If chain levels were under-scored, extend ALL Mns at missing levels
  - This is mandatory for the canonical standard
  - Re-run all analyses post-extension

Stage 2: Methodology synthesis (this doc)
  - Codify rules with evidence anchors
  - Compare to other-domain experience
  - Identify open questions

Stage 3: Anchor authoring
  - Tag-by-projection (Rule 11)
  - Strong inductive anchors from strict-threshold spines
  - Convergent-zone anchors from moderate-threshold zones
  - Projection-specific anchors from chain-level-restricted or primitive-subset analyses
  - Full JSON anchor files matching the existing corpus_role=anchor schema

Part IV — Analytical lens hierarchy

The framework supports multiple analytical views. They are complementary, not competing. An anchor or pattern should be tagged with the view that surfaces it.

View 1 — Integrated meta-stability across all chain levels

The "default" view. Co-cluster across (N levels) × (3 signatures) configs at strict + moderate threshold.

Surfaces: broad substrate-grade clusters (mammals, vascular plants, etc.). The within-grade variation washes out.

Use when: you want the dominant structural regions of the corpus.

View 2 — Per-chain-level cluster + stability

Apply cluster_classical + cluster_stability at each chain level separately. Compare the resulting cluster assignments side-by-side.

Surfaces: chain-level-specific structural patterns. Substrate-level clusters reveal substrate-grade attractors; bridge-level clusters reveal developmental-program attractors; organism-architecture clusters reveal organism-form attractors; organism-to-ecosystem-bridge clusters reveal ecological-role attractors; ecosystem clusters reveal trophic-role attractors.

Use when: you want to see what the corpus structures around at a specific chain level.

View 3 — Primitive-subset projected clustering

Cluster on a user-specified subset of primitives (across one or more chain levels). E.g., (Cm, RD, CDif) for "social/reproductive architecture" or (Pd, Cs, Cy, Dec) for "trophic role."

Surfaces: structural patterns localized in a few primitives that washout under uniform Euclidean distance.

Use when: you have a hypothesis about which primitives carry a particular convergent signal.

Status: not yet fully scripted as a standard analytical lens. Currently done ad hoc.

View 4 — Cross-arrangement coupling

When a Mn is realized at Sc=4 (multi-arrangement context), look at coupling between arrangements. This is documented in earlier work (how-to-apply-cross-arrangement-coupling.md).

Use when: you want to see how arrangements relate (substrate sharing, evaluator sharing, etc.).

Part V — Anchor authoring: how and when

Anchors are not "the truth" about cluster membership — they are documented structural regions that emerge under specific projections + thresholds.

Anchor types

1. Strong inductive anchor (multi-view) — surfaces in BOTH integrated meta-stability AND per-chain-level views; persists across 2+ waves of corpus expansion. Examples: vascular-plants (entity = vcs-system).

2. Single-view inductive anchor — surfaces in ONE view only. Examples: decomposer-recycler (organism-to-ecosystem-bridge projection only).

3. Convergent-zone anchor — surfaces at moderate threshold across phylogenetically-distant lineages. Examples: basal-eumetazoan-intermediate-bridge (4 phyla converge at bridge profile).

4. Pair anchor — only 2 documented members at strict threshold. May graduate to inductive anchor if more members are found. Examples: choanoflagellate (monosiga + salpingoeca-rosetta).

5. Declarative anchor — analyst-proposed region with members specified, scored declaratively rather than emerging from clustering. Used to capture intuitive structural regions that the framework might not surface. Examples: interactive-web-platforms (entity).

When to author anchors

• Strict-threshold spine persistent across 2+ waves (substrate-grade anchor)
• Moderate-threshold zone persistent across 2+ waves (convergent-zone anchor)
• Single-view cluster reproducibly persistent (projection anchor)
• After Full Unified Manifestation extension is complete (no under-scoring)
• After Stage 2 methodology synthesis (this doc)

Anchor metadata

Each anchor JSON should record:

• corpus_role: "anchor"
• origin: "inductive_clustering" | "declarative_population"
• derived_from_cluster: which analysis run + cluster ID + threshold + chain-level + signature
• exemplars (3-5 representative Mns)
• positions (centroid scoring across all chain levels)
• evidence narratives per chain level

Part VI — Connection to canonical-architecture-strategy.md

The canonical strategy has 10 within-arrangement discipline rules. This doc adds 12 cross-arrangement buildout rules.

Suggested integration: add this doc as §11 — Landscape Bootstrap Methodology to the canonical strategy doc. The within-arrangement rules + cross-arrangement rules together form the complete methodology.

Connection points:

• Rules 6-7 here (corpus-state-dependence, pair-tightening) connect to the canonical scope ladder (Sc0-Sc4) — chain-level integration is one form of scope shift.
• Rule 8 here (chain-level + role-resolved) connects to the canonical layer 4-5 distinction (substrate primitives vs surface/ecosystem primitives).
• Rule 11 here (projected-primitive analysis) connects to canonical Layer 5 navigation primitives.
• Rule 12 here (unified manifestation) connects to canonical §2.4 (within-arrangement Mn discipline).

Part VII — Open questions + gaps

Things that the biology buildout surfaced but did not fully resolve:

1. Projected-primitive analytical tooling is not yet scripted. The (Cm, RD) projection and the organism-to-ecosystem-bridge-only cluster were done ad hoc. We need a cluster_projected.py that accepts a list of (chain-level, primitive) pairs and runs clustering on the projected subspace. This would make Rule 11 fully operational.

2. Entity arrangement chain-level coverage check. Entity Mns score 5-6 of 8 chain levels. Are physical-hardware + hardware-to-computing-bridge + digital-computing actually uniform across software systems, or have we been under-scoring entity Mns too? Should re-check using the same Rule 12 lens.

3. Eusocial-projected analysis. Even at organism-to-ecosystem-bridge, eusocial Mns scattered. But is there a primitive subset where they DO cluster? Try (Agg, Pop, RD, CDif). If yes, that's a methodologically interesting finding (eusociality has a localized signature). If no, eusociality is genuinely an emergent behavioral pattern not captured by structural primitives, and we should be explicit about this limit.

4. Authoring artifact in marchantia. Wave 5 showed marchantia drops to singleton when anthoceros joins physcomitrium. Diagnosis: PF1 in marchantia vs PF3 in physcomitrium + anthoceros. Is PF1 correct (reflecting marchantia's actually-lower polarity elaboration) or is it under-scored? Could be re-examined under Rule 2 calibration discipline.

5. Per-chain-level vs integrated anchor inventories. We have 12 anchor candidates from 3-chain-level scope (Wave 5). We have 4 ecological-role clusters from organism-to-ecosystem-bridge alone. Authoring both sets as anchors creates a 16+ anchor inventory. Is this the right anchor density, or should we consolidate? Worth deciding before Stage 3.

6. Methodology arrangement. This is the third arrangement that hasn't been fully built out. Once Stage 2-3 are complete for biology, methodology arrangement is the next domain to bootstrap — and it will test whether the rules generalize beyond biology + entity.

7. Cognition arrangement Mn-kind partitioning. Earlier work noted cognition's heterogeneous Mn-kinds (developer keypress is Sc=4 cross-arrangement; cultural-artifact Mns presuppose neural-hardware). Cognition buildout requires Mn-kind partitioning first per canonical strategy Rule 8.

8. Ecosystem-level anchor identification. The organism-to-ecosystem-bridge analysis revealed four ecological-role clusters but we haven't authored anchors for them. The decomposer-recycler cluster (fungi + earthworm + termite + methanogen) is one of the cleanest cross-kingdom convergent findings — should be authored as a convergent-zone anchor.

Part VIII — Where to next

Three viable directions from here:

A. Author the anchor inventory (Stage 3) — 12-16 anchors, mechanical authoring informed by Wave 5 + organism-to-ecosystem-bridge findings. ~1 session.

B. Build the projected-primitive analytical script — formalize Rule 11 as cluster_projected.py. Enables further empirical investigation of eusocial-projection, trophic-role-projection, etc. ~1 session.

C. Apply Rule 12 retrospectively to entity arrangement — check whether entity Mns are properly scored at all chain levels; extend if not. ~1-2 sessions depending on scope.

Recommended order: A → B → C. Author the biology anchors first to capture the current understanding. Then build the projected-analysis tool to enable continued investigation. Then return to entity for the symmetric check.

After all three, the methodology is mature enough for paper-integration work — feeding the findings back into Paper 6, Paper 11, and other relevant papers.

Appendix: Cross-domain comparison summary

The shapes are comparable; biology has cleaner separation (higher silhouettes) likely because evolution has produced sharper grade transitions than software-system design has.

Files