Structural Signals #007 · Minaret · May 12, 2026
The Sprawl
Picture a metropolitan area that annexed dozens of surrounding townships and never extended the road network. The downtown still has highways. The new districts have driveways that end at the property line. On paper the city is enormous. In practice almost everyone lives somewhere a car cannot easily reach. Minaret’s topology behaves the same way: a 749-page Fortune 500 site where three pages hold the functional core and the rest of the content exists without a road to it.
The Subject: Minaret
Minaret is a 749-page enterprise site. By the surface metrics, the structure looks reasonable enough: any two connected pages are about two clicks apart, the community algorithm picks out 25 topical clusters, the skeleton is thin but functional. The picture starts to fall apart when you look at how much of the site is actually inside that structure. The vast majority of pages receive no inbound editorial links at all. They exist on the server. They have no role in the graph.
We call this The Sprawl. In urban planning, sprawl describes development that expands outward without the infrastructure to support it — subdivisions without transit, commercial zones without sidewalks, residential blocks without connecting streets. The geography grows. The connectivity does not. Over time, the core becomes the only functional part of the system, and everything at the periphery depends on private navigation to reach anything else.
Minaret’s topology exhibits the digital equivalent. A handful of hub pages command nearly all the authority. The rest of the graph barely participates. Communities, where they exist at all, do not cross-link. The site has sprawled outward in page count and inward in connectivity at the same time — large in raw number, small in structural reality, and fragmented across both dimensions.
Structural Signal: When the connected core shrinks to a handful of hub pages and the majority of pages receive zero inbound editorial links, the topology has sprawled beyond its linking infrastructure — the site grew, the connections did not, and structural reality is a fraction of what the page count suggests.
The Five Lenses
Skeleton
Good
Circulation
Critical
Organs
Critical
Health
Critical
Nervous System
Critical
Skeleton — Size & Connectivity
749 nodes, 2,510 edges, density 0.004, average path length 2.0. The path length is short because the connected portion of the graph is shallow — nearly everything sits at depth 1. But the density tells a different story. At 0.004, this is the sparsest topology we have measured. Fewer edges per node means fewer pathways for authority to flow, fewer routes for crawlers to discover content, and fewer structural signals for AI agents to interpret. The skeleton is technically connected. It is also thin.
Circulation — Authority Flow
PageRank Gini of 0.87 — the most extreme authority concentration in this series. The top 3 pages hold 87% of all link equity. For comparison, Episode 6 (Portico) measured 0.57 with the top 1% holding 21%. Minaret’s concentration is four times higher. Zero pages qualify as overlinked hubs by out-degree, which means the concentration is purely a function of inbound links: other pages point to these three, and these three point to each other. The rest of the graph barely participates in the circulation at all. Click Hubs to see the three-node core.
Organs — Community Structure
25 communities with a modularity of 0.33. One community dominates: 501 pages, 67% of the site, absorbed into a single mega-cluster. The remaining 24 communities range from 37 pages down to singletons. Eleven communities are singletons — a 44% singleton rate, comparable to Episode 5 (Nave). But where Nave’s singletons were a frozen tail at the periphery of an otherwise healthy core, Minaret’s are pages stranded outside a system that barely organizes anything in the first place. Click Silos to see the mega-cluster and its satellites.
Health — Content Isolation
81% orphan rate — 610 pages with zero inbound editorial links. This is the highest orphan rate in the series, surpassing Episode 6 (Portico) at 31% by a wide margin. Zero dead-ends and zero islands — every page that receives a link also sends links outward, and the connected component spans the full graph. The health profile is binary: you are either part of the linked 19% or part of the invisible 81%. There is no middle ground. Click Orphans to see the scale of structural invisibility.
Nervous System — Depth & Bridges
Participation coefficient of 0.12 with a bridge rate of 0%. The depth distribution is flat: 748 of 749 pages sit at depth 1. No bridges exist because the connected portion of the graph has no separable community structure worth bridging. The participation coefficient is the second lowest in this series after Episode 5 (Nave, 0.14). Connected pages overwhelmingly link within their own community. Even the 19% that are structurally visible do not cross-link. The topology is isolated at both scales: between the connected and disconnected halves, and within the connected half itself. Click Bridges to see the flat, unbridged architecture.
What Would We Fix?
The Sprawl requires the largest intervention of any topology we have analyzed. With 81% of pages invisible and even the visible portion siloed, the fix must operate on three levels: rescue the orphans, connect the communities, and redistribute the authority that has concentrated in three pages. The goal is not to flatten the hierarchy — those hub pages are likely important landing pages — but to extend the linking infrastructure to match the content that already exists.
The optimizer identified 1,275 strategic changes: 603 orphan rescues, 132 cross-community bridges, and 540 edge redistributions.
The original topology as crawled. All edges are from the live site.
Rescue 603 Orphaned Pages
81% of the site receives zero inbound editorial links. Each orphan is connected to the most relevant hub within its algorithmically detected community, bringing every page into the structural graph.
+603 inbound links to orphan pages → Every page becomes discoverable through editorial link structure
Bridge 25 Isolated Communities
Participation coefficient of 0.12 means communities barely acknowledge each other. 132 cross-community links connect sections that share topical relevance but no structural pathways.
+132 cross-community bridge links → Communities gain editorial connections that signal topical relationships
Redistribute Concentrated Authority
Three pages dominate link equity. 540 edges redistribute from the over-concentrated core to mid-tier pages, breaking the three-node monopoly.
540 edges redistributed across sub-hubs → Authority spreads through the full graph instead of pooling at three pages
Structure is Signal
Minaret is the most structurally extreme topology in this series. A PageRank Gini of 0.87, an 81% orphan rate, a participation coefficient of 0.12, and a single community holding 67% of all pages. Every metric pushes toward the same conclusion: this is a site where content production vastly outpaced content integration. Pages were published. Links were not added. Over hundreds of publication cycles, the gap between the visible core and the invisible periphery widened until the topology became a small connected cluster surrounded by a vast structural void.
Twelve hundred seventy-five interventions — a 29% increase in total edge count — would transform it. Every orphan becomes discoverable. Community pairs gain editorial bridges. Authority spreads from three pages to dozens. The page count does not change. The structural reality of the site expands by a factor of five.
Each topology in this series has revealed a different failure mode. Episode 1 found content islands. Episode 2 found gravity wells. Episode 3 found a monolith. Episode 4 found an archipelago. Episode 5 found a frozen tail at the periphery of a healthy core. Episode 6 found a compound fracture where three problems struck at once. Minaret takes a different path: the sprawl, where the connected core shrinks to three pages and the invisible periphery expands to 81% of everything. The site did not neglect its structure. It never built one for most of its content in the first place.
Methodology & Disclaimer — This analysis was performed using web topology crawling and network science methods including PageRank, Louvain community detection, and betweenness centrality. Navigation, header, and footer links are excluded to isolate editorial linking structure — only in-content links are analyzed. All data represents publicly accessible page structure only — no content, metadata, or user data was collected or stored. All identifying information has been anonymized. The embedded graph is a lightly perturbed version of the underlying topology — structurally equivalent in shape, community composition, and the qualitative patterns described in the post, but specific node-level metrics on the live site are more extreme than what the perturbed graph displays. Structural patterns are presented for educational purposes only.