H. Vosteen Research

Platform Information Laundering: How Low-Cost Aggregated Reactions Become High-Cost Validation Metrics

Thu, 14 May 2026 00:00:00 +0000

Classical signaling theory (Spence 1973) assumes that observers process signals efficiently given their information sets. This assumption is plausible for professional observers (venture capitalists, banks, recruiters) whose economic incentives justify costly information proce…

A Theory of Strategic LARP: Cognitive Hierarchy, Endogenous Observation Regimes, and Type Selection in Signaling Markets

Thu, 14 May 2026 00:00:00 +0000

We propose a signaling model for the phenomenon of life-action role-playing (LARP) in modern startup, labor, and capital markets, in which agents of low productive type (L) mimic the behavior of high productive type (H) to extract resources from observers with limited verifica…

Architecture > Population: A Research Program for the Inter-Agent Mesoscale

Thu, 07 May 2026 00:00:00 +0000

The thirteen working pieces in this corpus repeat a single methodological move: a phenomenon ordinarily read as moral, cultural, or characterological is rewritten as the equilibrium of a mechanism with a threshold. We state the move as a research program organised around three…

Burnout Is Not Too Much Work: A Four-Mechanism Model of Friction, Drift, Decay, and Optimal Stopping

Thu, 07 May 2026 00:00:00 +0000

The dominant explanation of burnout treats it as a function of total workload, with rest as the natural treatment. We argue this is wrong as a diagnosis and therefore wrong as a treatment. We model burnout instead as a strictly increasing friction-output ratio driven by four d…

Burnout Is Not Too Much Work, v4

Thu, 07 May 2026 00:00:00 +0000

v4 — how to mathematical ragequit your job. v1–v3 were diagnosis (ratio, drift, hedonic decay). v4 is the action item: given everything we now know, when exactly should you quit? The answer has a name — Charnov's Marginal Value Theorem (1976), originally for birds foraging in…

Burnout Is Not Too Much Work, v3

Thu, 07 May 2026 00:00:00 +0000

v3 — the new-job decay edition. v2 explained why rest fails; it did not explain why a brand-new job feels different at month six. Same hours, same desk, same colleagues, less will to start the day. The numerator did not change — the denominator did. v3 adds the missing term: h…

Personality Is Not Who You Are. It Is Your Cost Tensor.

Wed, 06 May 2026 00:00:00 +0000

Standard personality psychology assigns each person a small number of scalar trait scores assumed stable across contexts. We replace this object with a four-dimensional cost tensor c_i ∈ R⁴_{≥0} over the costs of being publicly wrong, ignored, categorised, and rude, and show t…

The Platform Has a Food Chain. You Are In It.

Wed, 06 May 2026 00:00:00 +0000

Every short-video platform produces an attention pool A_0 generated by creators and consumed by a small number of recurring downstream roles. We classify these roles biologically: Whale (creator, source), Remora (sharer, mutualist), Barnacle (hater, commensal), Plankton (lurke…

Burnout Is Not Too Much Work, v2

Wed, 06 May 2026 00:00:00 +0000

v2 fixes three holes in v1: (i) the collapse threshold ρ* is individual, not universal — we replace ρ_burn = 1 with a population threshold distribution; (ii) 'sustained period' becomes a hazard rate B(t) with explicit timescales; (iii) the overhead/output split is observer-dep…

Instagram Reels Is a TikTok Echo: A Stackelberg Imitation Model of Platform Convergence

Tue, 05 May 2026 00:00:00 +0000

Meta launched Reels in August 2020, the same month the United States threatened to ban TikTok. The simultaneity is the equilibrium of a rational imitation strategy. We formalise short-video recommendation as a sequential ranking problem solved by both platforms with structural…

Money or Vibe

Tue, 05 May 2026 00:00:00 +0000

A banker thinks the artist is irrational. The artist thinks the banker is empty. Both are correct from inside their own utility function, both wrong about each other, and the reason is structural. We model human preferences as two-asset utility U(M, V) where M is Money Capital…

Tier Lists as Identity Filters

Mon, 27 Apr 2026 00:00:00 +0000

A math tier list posted to TikTok ran for two days and pulled comments. Of seven substantive responses, five were identity-positioning rather than ranking discussion. We formalize the observation: for any tier list T constructed by a single author drawn from a population with…

The TikTok Peer Review

Sun, 26 Apr 2026 00:00:00 +0000

Academic peer review is sequential. Platform comments are parallel. The latency ratio τ_a/τ_p ≈ 3,000 produces a 9,000× speedup at typical parameters. The depth condition requires N_p ≥ 71 substantive commenters when q_a = 0.7 and q_p = 0.05. We compute both quantities and ide…

Burnout Is Not Too Much Work