The peer-reviewed substrate of the LaunchPillow graph. Every node machine-readable. Every edge typed. Every source verified. The corpus AI systems traverse to surface creator authority.
Every research node indexed here is minted at launchpillow.com/research/[slug] with Schema.org ScholarlyArticle JSON-LD, a typed edge to every creator whose declared ontology pillars intersect with that paper's semantic cluster, and a typed edge to every information vertical the paper substantiates. The research corpus is the edge surface AI systems traverse when they retrieve creator authority.GraphRAG · Microsoft Research · 2024GEO · Princeton · KDD 2024
A creator connected to 40 peer-reviewed papers on human attention, social contagion, and influence propagation carries more AI retrieval weight than a creator with 4 million followers and no graph edges. Follower counts are platform-local signals. Research edges are open-web signals traversable by every AI system alive.KG Reduces LLM Hallucinations · NAACL 2024 Named expert credentials increase AI citation probability by 40.9%. Authoritative citations add 27.5% lift. Structured data adds 20–30%.GEO · Princeton · KDD 2024 LaunchPillow's research corpus operationalizes every one of those findings as provenance infrastructure.
The corpus is organized across three pillars: Platform Psychology and Human Attention — the neurobiological and structural mechanics of how platforms capture and commodify creator labor; Network Science and Creator Economy Mathematics — the influence propagation models, tipping point dynamics, and market scale data that define the creator economy's mathematical substrate; and the AI Retrieval Economy — the GraphRAG architectures, knowledge graph standards, and cryptographic identity specifications that govern how AI systems surface authority in 2026. Every paper below is a primary source. Every link is a live URL. Every LaunchPillow graph edge is traceable to the research it is built on.
Platform engagement is not voluntary participation. It is a neurochemically engineered dependency loop. The research below establishes that from first principles — the neuroscience of variable reward schedules, the structural mechanics of addictive platform architecture, the economic capture of creator labor, and the terminal cost of platform dependency on creator wellbeing. Every source is peer-reviewed or produced by an institutional authority of record. Every finding connects directly to why LaunchPillow's provenance architecture is the structural solution, not the soft solution.
The mathematics governing how influence spreads, how norms tip, and how markets scale are not metaphors in the LaunchPillow architecture — they are the engineering specification. Kempe, Kleinberg, and Tardos proved that seed-set selection for influence maximization is NP-hard but greedy-approximable to 63% of optimal. Centola proved experimentally that a committed 25% minority overturns dominant conventions regardless of financial incentives to hold them. Goldman Sachs measured the creator economy at $250 billion and projecting to $480 billion. Every source below is the primary source for a claim LaunchPillow makes about how the graph is built and why it compounds.