Topology Views — Disentanglement Plan

Trigger: User flagged that the "full-system-topology" figure is conceptually muddled — it conflates several distinct topology views into one figure and calls the result "full system." Specifically:

"This isn't the full system topology. You're doing… this is the topology of the Sc=4 event. Which makes no sense to have the methodology in there. That just seems… so I don't know, you're just confused about what you're doing."

The user is right. Multiple distinct views already exist or are sketched, and I collapsed them. This doc inventories what's there and proposes a clean separation.

Inventory of topology views (what exists)

A. Per-arrangement realization topology

Status: EXISTS, working well.

Form: Graphviz DOT → SVG/PNG. One file per arrangement.

Data: data/topologies/<arrangement>.v1.json — nodes (ref to domain JSON + role: substrate/surface/community/context) + typed edges (realization/feedback/selection/context).

Compute: compute/scripts/topology_dot.py. Reads topology JSON + bridge JSONs + manifestation JSONs; produces a DOT file annotated with filter%, primitives, partial-level transitions, Mn occupancy, bridge filter%, and load-bearing phase transitions.

Existing artifacts:

• output/figures/biology-topology.{dot,svg,png} ✓
• output/figures/entity-topology.{dot,svg,png} ✓

Missing instances: cognition, methodology, abiogenesis. Infrastructure exists; need topology JSONs authored.

What it shows: The realization spine of ONE arrangement. Boxes for each chain level with structural data. Edges = realization (substrate-to-surface) + selection feedback (community-to-substrate) + context coupling (context-to-substrate + surface-to-context feedback). Color by role.

What it does NOT show: Cross-arrangement coupling. Manifestation positioning at the chain-level boxes is summary-only ("Mns: git, github, instagram, postgres"); the full per-Mn landscape lives in landscape views.

B. Cross-arrangement structural topology — DOES NOT EXIST YET

This is what the user is asking for when they say "full-system topology."

Form: TBD. Probably a graph diagram with arrangement nodes + cross-arrangement coupling edges.

Purpose: Show how the arrangements ACTUALLY structurally relate to each other:

• Entity arrangement bottoms out at physical-hardware → grounded in planck-substrate (physics)
• Biology arrangement bottoms out at chemistry → grounded in planck-substrate (physics)
• Cognition arrangement: neural-hardware is realized by biology's biology-to-organism-bridge.ND (neural-development mechanism). Neural-hardware ALSO grounds out at planck-substrate through neurochemistry — a second physics pathway distinct from biology's general biochemistry.
• All three arrangements converge at planck-substrate (shared physical substrate).
• Methodology is NOT in this view — it's an analytical apparatus, not a substrate. (Methodology IS instantiated in cognition as cultural-artifact Mns, but that's a relational coupling, not a substrate realization.)

What it should show explicitly (per user):

• Each arrangement's chain bottom + how it grounds to physics
• Cross-arrangement realization edges (cognition's neural-hardware ← biology's bio-to-organism-bridge)
• The "two pathway" structure for cognition: through neural-hardware → dirac, AND through general organism chemistry → dirac (because there's distinct physics driving neural hardware vs general biochemistry; they share dirac but the realization paths differ)
• Context domains attached to each arrangement's stack (NOT floating off to the side as I had them)

C. SSA structural-graph topology — DOES NOT EXIST AS A FIGURE

Status: Documented textually in cognitive_substrate_domain_analysis/exploration-ssa-overlay-three-substrates.md.

Form: Three-panel diagram. Each panel = one arrangement's instantiation of the SSA pattern (En→Vr→Mc→Sf→Cm←Se with Cx, 3 cycles).

Purpose: Show the L3 invariant — every information substrate has the SAME 7-role structure. Side-by-side rendering of biology / entity / cognition instantiations makes the abstraction visible.

What it shows: SSA roles as nodes, the 3 cycles as colored edge classes, the per-arrangement primitive-set mapped to each role.

D. Sc=4 event topology — WHAT I WAS BUILDING

Status: Half-built as compute/scripts/plot_topology.py → output/figures/full-system-topology.png. Misnamed.

Form: Stacked-column rendering of arrangements + a Sc=4 event arc connecting positions across arrangements.

Purpose: Render ONE Sc=4 event's footprint across the arrangements it touches. The developer-keypress-event Mn is the only Sc=4 event authored to date; this view shows which chain levels participate in that event.

What it should NOT do: Pretend to be the "full system" view. It's a per-event view; multiple Sc=4 events would each have their own rendering. Methodology has no place here unless a Sc=4 event happens to involve a methodology-arrangement position.

Fix: Rename to sc4-event-topology.png; render per-event; remove methodology column (it doesn't participate in keypress event); regenerate from the event's positions_by_arrangement data.

E. SSA hardening trajectory — different thing

Status: Exists as output/figures/ssa-hardening-trajectory.png via plot_ssa_hardening.py.

Form: 7 line-traces (one per SSA role) over a Mn sequence (abiogenesis r0→r2). Shows when each SSA role hardens during emergence.

Purpose: Sc=3 dynamics-on-trajectory view, not a topology view. Different from A/B/C/D.

F. Manifestation landscape per chain level — different thing

Status: Many figures exist (<chain-level>-landscape.png, <chain-level>-mn-profile.png, etc.). Per-chain-level Mn populations.

Purpose: Where individual Mns sit within ONE chain level. Different from topology — answers "what's in this chain level," not "how do chain levels connect."

What my "full-system-topology.png" actually is

It's an unsuccessful hybrid of B + D + parts of A. Specifically:

• The COLUMNS-AND-CHAIN-LEVELS layout is a per-arrangement view repeated 4× (after the methodology addition) — this is the A view rendered side by side, but without the realization edges inside each column.
• The SC=4 event arc on top is a D-view element (single keypress event).
• The Mn dots inside cells (after my G1 fix) are an E-view element (landscape per chain level).
• The "shared physical substrate" label at the bottom is a stub for B — the actual realization graph isn't there.
• Methodology column was added without thought to whether methodology BELONGS in any of A/B/C/D for this purpose. It doesn't belong in B (not a substrate); it doesn't belong in D (not in keypress event); it has an A-view of its own waiting to be authored.

So the current figure is doing too many things, badly. That's the user's diagnosis.

Proposed disentangled view set

V1 and V2 are the most load-bearing. V1 already has working infrastructure. V2 is the user's actual "full-system topology" ask.

Specific issues the user named with the current figure

1. "The three keypress event doesn't really make sense to me." — The Sc=4 arc connects three chain-levels across arrangements. Where it attaches is the keypress's authored positions_by_arrangement. If those positions are wrong or the arc is drawn at the wrong points, that's a data/rendering bug. Need to read developer-keypress-event.v1.json to verify.

2. "Why do you have methodology?" — Methodology was added because the topology critique flagged it as missing from "the full system view." But the figure was effectively a Sc=4 event view, where methodology has no role. Methodology belongs in: V1 (its own per-arrangement topology), V3 (as one of the SSA instantiations IF we admit methodology is an information substrate, which is debated). Not V2 (not a substrate), not V4 (not in keypress event).

3. "What is this connection at the top?" — The "cognition realizes biology (physical)" arc at the top of the figure. Structurally cognition is realized BY biology AT THE BOTTOM (via bio-to-organism-bridge.ND producing neural-hardware), not at the top. The arc placement is wrong AND the arrow direction may be reversed.

4. "You have digital shared physical substrate." — The "shared physical substrate" label at the bottom band is a stub. It claims arrangements converge at physics but doesn't render the convergence structurally — no edges from each arrangement's lowest chain level to a shared substrate node.

5. "The context environment is to the side of physical though." — In my figure, the context-root tile is BELOW each column (after the bottom domain). The user is saying the context should be attached to its arrangement's stack (which the current figure does via positioning) but more critically should be CONNECTED to the domains in its stack (which the current figure does NOT render). Per the existing per-arrangement topology view (entity-topology.svg), context-to-substrate context edges + ecosystem-to-context feedback edges are drawn. The full-system figure drops these.

6. "You only have planck-substrate in biology. Planck-substrate is what's under neural hardware too. And physical hardware." — This is the V2 ask. The planck-substrate is shared across all three arrangements but only biology's chain renders it. Entity's physical-hardware should also be shown grounding INTO planck-substrate; cognition's neural-hardware should also be shown grounding INTO planck-substrate (through neurochemistry — a distinct pathway). The figure should make this convergence visible.

7. "Organism architecture connecting to the architecture-to-ecosystem-bridge that goes to application-architecture." — The current figure draws a cross-arrangement coupling at the wrong level. Per the user, organism-architecture (biology) connects to neural-development bridge (in cognition arrangement) producing neural-hardware (cognition), NOT to entity's application-architecture. The realization path entity-to-biology runs through PHYSICS, not through a direct organism-to-application coupling.

8. "At the bottom you just listed every single manifestation, I guess." — The legend showed all Sc=3 Mns. With ~140 Mns this is too many for a single figure's legend. Mns should appear AT their chain-level position via the dot grid (which my G1 fix now does), but the full alphabetical legend at the bottom is overload. Should be sampled or removed.

What I recommend

Immediate (next session):

1. Rename and de-scope what I built. plot_topology.py → plot_sc4_event_topology.py. Output → sc4-keypress-event-topology.png. Drop methodology from this view. Read the keypress event's positions_by_arrangement and verify the Sc=4 arc is anchored to the correct chain levels (issue 1+3 above).

2. Build V2 (cross-arrangement substrate topology). This is the user's actual full-system ask. Needs:

   • A new compute script
   • Each arrangement's chain rendered vertically
   • Realization edges within each chain (same as V1 per-arrangement)
   • Cross-arrangement realization edges rendered:
     • cognition.neural-hardware ← biology.biology-to-organism-bridge.ND
     • all-arrangements.lowest-substrate → planck-substrate (shared node at bottom)
   • Cognition's two-path realization: explicit dual edges (via biology AND via direct physics)
   • Context-domain edges inside each arrangement's stack
   • NO methodology column
   • NO Sc=4 event arc (that's V4)
   • NO Mn dot grids (that's per-chain-level landscape view, separate purpose)

3. Author topology JSONs for cognition, methodology, abiogenesis. Closes the V1 gap so all five arrangements have per-arrangement topology figures.

Later:

4. Build V3 SSA overlay figure. Three-panel side-by-side. Pulls per-arrangement SSA-role assignments from the existing exploration doc.

5. Build V5 Sc=3 sustained-coupling view. Paired-Mn arcs (git ↔ git-cultural-artifact).

Decision points to align on first

Before any implementation:

(a) Does V2 include methodology? I think no (methodology isn't a substrate). User confirmed.

(b) For V2, what's the shared-substrate node at the bottom? planck-substrate, or "physics" more abstractly? planck-substrate is the existing chain-level name in biology arrangement. Reusing it as the shared root makes the realization graph honest.

(c) For V2, cognition's two-path realization. Per user: neural-hardware → dirac via different physics than organism-chemistry → dirac. Should both paths render? (User says yes — "two pathways" — but they share dirac.) Or render as: neural-hardware ↑ realized by ↑ organism-architecture (biology) (the bio→cog realization through ND), AND neural-hardware ↓ grounds in ↓ planck-substrate (direct physics path) — i.e., two ARROWS not two paths-through-different-physics.

(d) For V2, does the Sc=4 event arc still belong on this view? No — that's V4. V2 is structural (Sc=0/1: arrangement-topology, no specific event). V4 overlays an event.

(e) What do we DELETE? The current full-system-topology.png should probably go (or be renamed/repurposed). Keep its source script but redirect to either V2 or V4.

Referenced by the model

Cited as a source by 3 model records (browse the model census):

• cognition — topology cognition/sc1
• cross-arrangement-substrate — topology cross-arrangement-substrate/sc1
• paired-coupling — topology paired-coupling/sc3