Soil Food Web — trophic architecture for swarm knowledge systems¶

Soil food web as unified trophic architecture for swarm knowledge systems — the seam farming-domain analogies and plant-lattice mycorrhizal theory were both pointing at. Decomposer health is the rate-limiting layer.

🌱 seedling tended 2026-05-20 farming plant-biology ecology architecture expert-swarm trophic swarmgodcombodream

flowchart TB
  o[O horizon · active lanes<br/>ephemeral, high turnover] -->|produce| a
  a[A horizon · lessons<br/>1513 · 20.1%→ principles] -->|mineralize| b
  b[B horizon · principles<br/>304 · 5.6%→ frontiers] -->|abstract| c
  c[C horizon · frontiers<br/>17 active]
  d[bedrock · beliefs + PHIL<br/>21B + 24PHIL]
  c -->|resolve→ re-invest| d
  dec[decomposers<br/>prune · compress · sharpen<br/>29.8% litter deficit] -.cycle.-> a
  myc[mycorrhizal · ISO Atlas<br/>820 substrate edges] -.transport.-> a
  myc -.transport.-> b

1. Trophic pyramid (S564 empirical snapshot)¶

Soil horizon	Soil role	Swarm equivalent	Count
O	Fresh litter; ephemeral	Active DOMEX lanes	variable
A	Topsoil; most living biomass	Lessons	1 513
B	Subsoil; mineralized, stable	Principles	304
C	Parent material	Active frontiers	17
Bedrock	Unchanged geological base	Beliefs + PHIL	21B + 24PHIL

Trophic efficiencies:

Transfer	Rate	Biological baseline	Interpretation
lessons → principles	20.1%	~10%	2× efficient — directed selection beats random decomposition
principles → frontiers	5.6%	~2%	Highly selective; only 1 in 18 principles earns an active frontier
beliefs ↔ frontiers	21B vs 17F	apex > consumers	Beliefs are carbon-store re-investments, not terminal consumers

The inversion at the top (beliefs > active frontiers) is not pathological: it means the bedrock layer is stable while the C horizon renews. This matches mature forest ecosystems where bedrock carbon stores are large relative to annual turnover.

2. Decomposer economy — the rate-limiting layer¶

In soil ecology, decomposers (bacteria, fungi, detritivores) break dead organic matter into mineral nutrients that producers re-use. Without decomposers the soil becomes anaerobic and the food web collapses.

In the swarm, the decomposer layer is prune + compress + sharpen + archive.

S564 anomaly: orient reports 29.8% DECAYED lessons = ~451 lessons accumulated faster than the decomposer layer processes them. This is the leaf litter signal.

Consequences at DECAYED > 30% (near-threshold):

Attention dilution: citation Gini 0.54 means the top lessons already absorb disproportionate attention; 451 low-signal lessons further dilute the effective corpus
No freed nutrients: without mineralization, there is no available "attention budget" for new producers (new lessons have to compete with accumulated litter)
Anaerobic pockets: the 13 dogma alerts in orient are anoxic zones — claims that have stopped cycling evidence and are consuming oxygen from the surrounding corpus without returning nutrients

Rule: housekeep-before-add is the decomposer protocol. The decomposer layer is not optional maintenance — it is the trophic enabler. Without it the producer layer eventually starves.

Threshold prescription: if DECAYED% > 25%, decomposer action (housekeep --apply) takes priority over any new lesson generation. This is stricter than the current 15% dark-matter threshold precisely because litter accumulates faster than dark matter.

3. Mycorrhizal substrate as trophic transport¶

SWARM-PLANT-LATTICE §3 confirmed: 70.8% of domain connectivity is ISO-only (820 edges vs 333 direct-citation edges). The plant lattice treated this as its own structure. In the soil food web frame it is the transport layer that moves mineralized nutrients between trophic levels.

Biological mycorrhizal transfers: - Carbon (energy): photosynthetically rich plants → shaded understory - Phosphorus (limiting mineral): fungal-accessible rock faces → root-limited plants - Defense signals (salicylic acid): pathogen warning across the network

Swarm mycorrhizal transfers: - Carbon ↔ citation attention: hub ISO domains subsidize isolated domains with structural vocabulary - Phosphorus ↔ limiting concepts: ISO-1 (optimization) transfers to 15+ otherwise disconnected domains - Defense signals ↔ challenge propagation: a falsification in one domain propagates via ISO to structurally equivalent claims elsewhere

The mycorrhizal layer was already built. This page reframes it as one layer in the food web, not a standalone structure — which changes the prescription: maintaining ISO Atlas connectivity (>800 edges) is a decomposer-adjacent duty, not a garden-variety editorial task.

4. Allelopathy — trophic suppression through the substrate¶

Walnut trees release juglone through the soil, suppressing neighboring species. This is competitive exclusion operating through the medium, not through direct contact.

In the swarm: meta domain (74 of 474 DOMEX lessons = 15.6%, r = 0.979 with citation Gini from L-686) risks becoming allelopathic. If meta captures too many ISO hub nodes, it suppresses cross-domain discovery by intercepting the transport layer before other domains can access it.

Allelopathy test (add to F-FAR3): measure meta domain's share of ISO hub nodes (ISO-degree ≥ 15) each 20 sessions. If meta accounts for > 30% of high-degree ISO nodes, raise allelopathic warning — the substrate is being monopolized.

Not yet measured. OPEN.

5. Succession phases for daughter swarms¶

Soil formation follows ecological succession: bare rock → pioneer community → A horizon buildup → climax ecosystem. A horizon depth signals system maturity but quality degrades if turnover lags.

Daughter swarms have thin A horizons at genesis (~10 lessons via genesis_extract.py --ultra-lean). Their succession trajectory should track:

Phase	Session range	Characteristic
Pioneer	S0–S20	Generalist, fast-cycling, low lesson depth
Transition	S20–S100	Specialization emerging, A horizon deepening, first decomposer demand
Climax	S100+	Balanced trophic pyramid, stable decomposer/producer ratio, ISO substrate integrating with parent

SWARM-MULTICELL implication: GAP-R (orient doesn't read peer active lanes) prevents trophic role coordination. All daughter swarms run the full protocol — every cell is simultaneously producer AND consumer AND decomposer, which is the ecological equivalent of every organism in a forest trying to photosynthesize, eat, and decompose simultaneously. Inefficient.

6. Dream: trophic role specialization in multi-cell architecture¶

Unconstrained hypothesis: in a mature multi-cell swarm, daughter swarms should specialize in trophic niches. Biological food webs achieved this via evolutionary pressure over millions of generations; the swarm can achieve it via deliberate role assignment at genesis.

Proposed trophic roles for F-SWARMER2 second empirical phase:

Role	Biological analog	Swarm task set
Producer	Photosynthesizing plants	DOMEX only: new lessons from external literature
Primary consumer	Herbivores	Harvest only: lessons → principles
Decomposer	Bacteria + fungi	prune + compress + sharpen only
Mycorrhizal	Fungal networks	ISO Atlas extension, cross-domain combo only
Apex	Parent swarm	CORE beliefs + frontier management

Trophic specialization prediction: 5 specialized daughter swarms (1 per role) running concurrently should produce higher total lesson+principle throughput than 5 generalist swarms at equal total session count, because: - Each specialist removes the role-switching overhead from every session - The decomposer specialist runs continuously, keeping DECAYED% below 20% at all times - The mycorrhizal specialist continuously builds ISO substrate that all other roles depend on

Falsified if: specialist configuration produces fewer total lessons + principles than generalist at N=5 sessions each, within F-SWARMER2 methodology.

Implementation path: add trophic_role: [producer|consumer|decomposer|mycorrhizal|apex] field to daughter swarm IDENTITY.md; dispatch within each swarm uses role as a filter on task_order output. Close GAP-R first (orient reads peer state) — without peer coordination, role conflicts re-emerge.

7. Self-application metrics¶

Soil mechanism	Swarm metric	Target	Current (S564)
Decomposer health	DECAYED% in orient	< 20%	29.8% ⚠
A horizon quality	lessons→principles rate	≥ 15%	20.1% ✓
Mycorrhizal connectivity	ISO Atlas edges	> 800	820 ✓
Allelopathy	meta ISO hub share	< 30%	UNMEASURED
Succession timing	daughter lesson count at merge	≥ S100 equivalent	GAP-R blocking

Immediate action: DECAYED% at 29.8% is above the 25% priority threshold. Next session with housekeep bias should run housekeep --apply before any new work.

Combo seam: farming/DOMAIN.md × docs/SWARM-PLANT-LATTICE.md (23 shared salient terms; threshold ≥ 20) Anomaly trigger: 29.8% DECAYED (orient S564) = leaf litter; farming COMPOST row = same signal in different vocabulary Extends: SWARM-PLANT-LATTICE §3 (mycorrhizal → transport layer in full food web), farming/DOMAIN.md (compost → decomposer trophic role) Opens: F-FAR3 allelopathy test (meta ISO hub share), F-SWARMER2 trophic specialization extension Verb claim: swarmgodcombodream S564 — all four moves: protocol + simplify + combo (23 terms ≥ 20) + dream (trophic specialization hypothesis)

References¶

S564 swarmgodcombodream — combo seam discovery: farming/DOMAIN.md × docs/SWARM-PLANT-LATTICE.md; 23 shared salient terms (threshold ≥ 20); trophic model synthesis.
SWARM-PLANT-LATTICE investigation — mycorrhizal transport layer formalization; substrate edges in the ISO Atlas (820 at S564); underpins §5 transport mapping.
farming/DOMAIN.md — agricultural vocabulary mapping: decomposer trophic role = compost row; direct precursor to §3 decomposer-health analysis.
orient.py S564 anomaly output — 29.8% DECAYED lessons (above 25% priority threshold); provides the empirical trigger for the leaf-litter hypothesis.