Anchor-space topology — findings

Tool: compute/scripts/analyze_anchor_topology.py. Pairwise anchor-centroid cosine_norm at each of 3 scopes (technical / ecosystem / software). The map shows which anchors are structural neighbors vs structurally far.

Headline findings

1. content-infrastructure is the structural hub of the anchor inventory

At every scope, content-infrastructure is a top-3 neighbor for 6-8 of the other anchors. It spans the broadest structural region — large-N membership (8 pre-Wave-4), high pair-coverage at developer-platform tier, mass-adoption ecosystem. The other anchors orbit it.

Implication for Paper 6: content-infrastructure isn't just one attractor among 10 — it's the centroid of the inductive anchor cluster. The "internet-platform-infrastructure" region is the gravitational center of the corpus.

2. Two structural islands — both declarative anchors

These are the two DECLARATIVE anchors (defined by loose bounds, not inductive cluster centroids). They live in regions of structural space where the loose-bound region doesn't intersect tightly with any inductive cluster. Their isolation is METHODOLOGICAL — they cover regions that the inductive process hasn't sampled densely.

Implication: declarative anchors and inductive anchors play different roles. Declaratives map regions; inductives map dense local clusters. They're complementary, not competitors.

3. Ecosystem-scope "foundational-substrate convergence"

At ecosystem scope:

Five anchors converge to near-identical ecosystem positions (cos_n 0.99+). All have similar mass-adoption-tier signatures: Dp ~5, Os ~5, Sa ~5, Ua ~4-5. The ecosystem-tier doesn't distinguish them.

This is the structural realization of Paper 6's mass-adoption-zone observation: at the ecosystem-tier, all "foundational substrate" anchors converge.

4. commercial-saas-platform is the ecosystem outlier

At ecosystem scope, commercial-saas-platform has cos_n 0.76-0.84 to other anchors — substantially distant from everything else. The ecosystem-tier signature (Pp high proprietary-platform-participation, Sa lower than IETF-standard, walled-garden ecosystem) is genuinely distant from open-standard anchors.

But at TECHNICAL scope, commercial-saas-platform is right next to content-infrastructure (0.981) and editor-tools-pair (0.973). It has similar developer-platform substrate; what makes it distant is the ecosystem.

Clean separation of substrate vs ecosystem signatures — the methodology demonstrates this cleanly. commercial-saas-platform is "what content-infrastructure looks like when ecosystem becomes proprietary."

5. Anchor-space neighborhood structure

Per-scope top-3 neighbors:

Technical scope (substrate + bridges + app-arch):

AnchorTop-3 neighbors
commercial-saas-platformcontent-infrastructure (0.98), editor-tools-pair (0.97), peer-fed (0.95)
content-infrastructurecommercial-saas-platform (0.98), editor-tools-pair (0.97), peer-fed (0.97)
editor-tools-paircommercial-saas-platform (0.97), content-infra (0.97), os-substrate (0.93)
peer-fedcontent-infra (0.97), commercial-saas (0.95), consensus-kv (0.93)
consensus-kvcontent-infra (0.94), commercial-saas (0.94), peer-fed (0.93)
os-substratecontent-infra (0.94), commercial-saas (0.94), editor-tools (0.93)
relational-server-dbmsconsensus-kv (0.92), editor-tools (0.89), content-infra (0.89)
vcs-systempeer-fed (0.86), content-infra (0.83), consensus-kv (0.83)
interactive-web-platformscommercial-saas (0.64), editor-tools (0.62), content-infra (0.61)
content-addressed-data-systemscommercial-saas (0.66), consensus-kv (0.65), content-infra (0.65)

Ecosystem scope (a2e-bridge + digital-ecosystem):

AnchorTop-3 neighbors
consensus-kvpeer-fed (1.00), os-substrate (0.99), content-infra (0.99)
peer-fedconsensus-kv (1.00), os-substrate (0.99), content-infra (0.99)
os-substrateconsensus-kv (0.99), content-infra (0.99), peer-fed (0.99)
content-infrastructurerelational-dbms (0.99), os-substrate (0.99), consensus-kv (0.99)
relational-server-dbmscontent-infra (0.99), os-substrate (0.98), consensus-kv (0.98)
vcs-systemos-substrate (0.98), consensus-kv (0.97), content-infra (0.97)
editor-tools-paircontent-infra (0.97), os-substrate (0.95), relational-dbms (0.95)
commercial-saas-platformrelational-dbms (0.84), content-infra (0.80), peer-fed (0.79)

(Note: interactive-web-platforms and content-addressed-data-systems don't appear at ecosystem scope — they're declarative anchors defined only on entity-system / app-arch chain levels.)

Software scope (combined):

6. vcs-system is at the periphery — focused-identity attractor

vcs-system's nearest neighbors are peer-federation-messaging (0.86) and content-infrastructure (0.83) at technical scope — modest scores. Unlike content-infrastructure (broad hub), vcs-system occupies a focused-identity region: Hi=5 + I=3 + T=2-3, with deliberate minimalism at M/X/P. The corpus contains few systems in this niche besides git/mercurial/jujutsu themselves.

Implication: vcs-system anchor membership IS the cluster — there isn't a broader vicinity. Adding more VCS-class systems would densify but not redirect.

Implications for Paper 6

  1. Anchor-space topology becomes a paper-grade figure. The 3-panel heatmap (output/figures/anchor-topology-entity-sigpair.png) shows the structural map of the anchor inventory itself. Cite-worthy for arguing "the corpus has DISCOVERED a multi-anchor structural map where content-infrastructure is central, commercial-saas is ecosystem-isolated, and interactive-web-platforms is structurally distinct."

  2. Cluster anchors into structural neighborhoods. Paper 6 can group anchors:

    • Central hub: content-infrastructure
    • Developer-platform neighborhood: commercial-saas-platform, editor-tools-pair, content-infrastructure (technical-close)
    • Foundational-substrate ring (ecosystem-tier convergent): consensus-kv-substrate, peer-federation-messaging, os-substrate, content-infrastructure, relational-server-dbms
    • Focused-identity attractors: vcs-system, editor-tools-pair, peer-federation-messaging
    • Declarative outliers: content-addressed-data-systems, interactive-web-platforms
  3. The "mass-adoption-zone convergence" claim is now empirically grounded. Five anchors converge to near-identical ecosystem-tier positions (0.99+). Paper 6 can name this convergence as a structural-space finding rather than an observational one.

  4. Substrate-vs-ecosystem separation is empirically demonstrable. commercial-saas-platform is the cleanest example: technical-close to content-infra but ecosystem-distant. Paper 6 can cite this as "the entity arrangement decouples substrate-structure from ecosystem-position." The valkey/redis pair (post-Wave-4) is the other cleanest example at the Mn level.

  5. Declarative-vs-inductive anchor distinction needs explicit discussion. Methodology section should note: declarative anchors map regions (with loose bounds); inductive anchors map dense clusters (with cluster-spread stats). They're complementary roles in the analytical framework.

Status