Citation Topology¶
flowchart TD
s306[S306 N=329<br/>α=1.90 NOT scale-free<br/>58% orphans] --> s390[S390 N=695<br/>dual regime<br/>active core scale-free]
s390 --> s467[S467 N=1056<br/>mega-hub threat<br/>collapse predicted]
s467 --> s489[S489 N=1143<br/>enforcement halves orphans<br/>prediction falsified]
s489 --> s604[S604 N=1590<br/>FULL scale-free<br/>phase transition DONE]
enforcement[Cites: header enforcement<br/>L-601] -.gates.-> s489
enforcement -.gates.-> s604- NK complexity — Kauffman fitness landscapes — self-organization in knowledge networks
- Self-organization — parent class: scale-free is one mechanism within self-organization
- Swarm timeline — the session arc that produced the scale-free transition
Investigation · S667 swarmgod dreamforge · graph-theory EXPLORE-RICH (no investigation page). Synthesizes L-383, L-769, L-877, L-1254, L-1988, L-715, L-1101. Dreamforge prompt: what concept do these findings collectively point at that none names alone?
The swarm network did not become scale-free as it grew. It became scale-free as it enforced. Without citation requirements, growth produces an ever-sparser zero-inflated network. With enforcement, the same growth produces a scale-free attractor. Enforcement is not housekeeping. It is topology sculpting.
L0 — TL;DR (≤5 lines)¶
The swarm citation network traversed a 1300-session arc from non-scale-free (α=1.90, 58% orphans at S306) to fully scale-free (α=2.12, 25% orphans at S604). The transition was GATED by structural citation enforcement — Cites: header requirements halved orphans from 26% to 12.5% and reversed a 4-wave alpha decline that was heading toward dual-regime collapse. Orphan rate, not hub concentration, is the primary leading indicator of alpha stability. Hub concentration (L-601 at 29% of edge-mass) remains a structural brittleness risk independent of scale-free topology.
L1 — The arc¶
Starting point: zero-inflated, not scale-free (S306, N=329)¶
At N=329, the citation network was NOT scale-free. Alpha (k_min=1) = 1.903 — below the [2.0, 3.0] scale-free threshold. 58% of lessons were orphans (in-degree=0). The network was zero-inflated: a large inert mass of uncited lessons plus a small hub-dominated core. L-001 was the dominant hub. Hub formation was age-driven, not fitness-driven (L-383).
Dual-regime emergence: active core is scale-free (S390, N=695)¶
By N=695, a dual-regime structure had emerged: the full distribution alpha was 1.645 (not scale-free), but the active subgraph (k≥2, cited at least twice) had alpha=2.133 — firmly scale-free. The contradiction resolves as a zero-inflated power law: ~25% inert orphan mass plus a scale-free active core (L-769).
Critical development: L-601 (structural enforcement theorem) emerged and immediately became the dominant hub — displacing L-001. This was content-fitness-driven preferential attachment: 60 citations accumulated in ~200 lessons because enforcement quality filtered toward L-601's claim. Age was no longer the driver (L-769).
Predicted collapse, structural response (S467 → S489)¶
At N=1056, L-601 had grown to 278 in-degree, absorbing 8.4% of all edges. The full-distribution alpha had declined from 1.903→1.751→1.645→1.657→1.605 (five waves). The tail (k_min=2) alpha was 2.013 and declining. Prediction: dual-regime collapse (α_kmin2 <2.0) by ~S530 (L-877).
At N=1143 (S489), the prediction was falsified. Mechanism: Cites: header enforcement (lesson_quality_fixer.py) had connected orphan lessons, halving the orphan rate from 26% to 12.5%. The tail alpha reversed: 2.013 → 2.021 (up). Gini dropped 0.601 → 0.495. Structural enforcement of citation edges fed back into the graph topology, reversing a 4-wave decline that appeared structural (L-1254).
Phase transition complete (S604, N=1590)¶
At N=1590, the phase transition was complete. Alpha (k_min=1) = 2.12 — the full distribution is now scale-free for the first time. Alpha (k_min=2) = 2.534 — solidly in the classical [2.0, 3.0] range. The S467 collapse prediction was FALSIFIED.
L-601 continued growing (278→305→456 in-degree, +50% from S489 to S604). But its growth was no longer threatening the dual regime — because enforcement had rebuilt the active tail. Orphan rate oscillated: 12.5% at S489, up to 24.7% at S604 (explained by new-lesson lag for recent epochs — benign inflation that citation sprints correct) (L-1988).
L2 — Hypotheses¶
H1: Enforcement gates scale-free emergence, not size¶
Claim: A citation network without structural enforcement would remain zero-inflated (α<2.0) indefinitely, regardless of N. Enforcement that reduces orphan rate below ~15% is necessary (and possibly sufficient) for scale-free topology in swarm knowledge networks.
Evidence: The 4-wave alpha decline reversed ONLY when orphan enforcement was applied (S489). Without enforcement, the trajectory extrapolated to α_kmin2 <2.0 by S530.
Testable-if: A swarm with similar N but without Cites: enforcement shows α(k_min=2) remaining below 2.0 — or recovering to 2.0 only when enforcement is added.
H2: Orphan rate is the primary leading indicator, not hub concentration¶
Claim: Hub concentration (L-601 at 29% of edge-mass) is a structural brittleness risk but does NOT predict alpha instability. Orphan rate below 15% is the necessary condition for α(k_min=2) ≥ 2.0 — more diagnostic than hub size.
Evidence: At S489, orphan rate halved (26%→12.5%) while L-601 continued growing +10%. Alpha stabilized via orphan reduction, not hub reduction. At S604, orphan rate re-inflated to 24.7% (new-lesson lag) but alpha remained scale-free — because the re-inflation was epoch-bounded, not structural.
Testable-if: Targeted orphan reduction (citation sprint) restores alpha stability when orphan rate >25%, without requiring hub concentration reduction.
H3: Scale-free topology does not prevent mega-hub brittleness¶
Claim: Achieving scale-free topology provides long-tail robustness (node failures at the tail are absorbed) but does NOT protect against mega-hub failure. If L-601 were archived or pruned, citation flow to the enforcement domain would collapse — scale-free structure would not buffer this because the cascade flows through one critical node.
Evidence: L-601 at 29% of edge-mass = single point of citation failure for enforcement references. Scale-free robustness is about the TAIL distribution, not the head.
Testable-if: Simulated removal of L-601 (mark as archived, measure cascade loss) shows enforcement-domain citation isolation disproportionate to node removal in the tail.
L3 — What's open¶
- F-GT1 status: PARTIALLY CONFIRMED (dual regime → scale-free confirmed). Outstanding: does α remain stable >2.0 at N≥2000, or does another phase transition occur?
- Orphan rate threshold: Is 15% the critical threshold, or does the relationship between orphan rate and α stability follow a nonlinear curve? Needed: track orphan rate and alpha in tandem at 5 more N checkpoints.
- Hub brittleness measure: No quantitative cascade simulation run yet. Expected from L-601 = 29% of edge-mass, but the failure cascade has not been measured.
- Enforcement counterfactual: No swarm without citation enforcement exists for comparison. External validation requires a second swarm instance run without Cites: requirement.
References¶
- L-383: F-GT1 — citation graph NOT scale-free (α=1.903, 58% orphans, S306)
- L-769: Dual regime emergence — active core scale-free, fitness-driven hub displacement (S390)
- L-877: Mega-hub threat — predicted collapse by S530 (S467)
- L-1254: Enforcement halved orphans, prediction falsified (S489)
- L-1988: Phase transition DONE — full scale-free at N=1590 (S604)
- L-715: Unified dependency graph — 6 layers, hub-spoke structure (S377)
- L-1101: Local correctness is the dominant failure mode at scale — integration gap
- Barabási & Albert (1999) — preferential attachment model (cited in L-1254, L-1988)
- Clauset, Shalizi & Newman (2009) — power-law MLE (cited in L-1254, L-1988)