No room for slop

Abstract

My first report in this series formulated a model of organizational coordination and deliberately stayed above the level of individual people — the target was an initiative's trajectory, never a person's behavior. That carefulness left a real question unanswered: prediction error doesn't generate itself. People generate it, or fail to, which is the harder case. This paper takes up what the first one declined to. I formalize two failure classes precisely enough to detect without diagnosing a person. A proposal is slop when its fluency exceeds any trace of the judgment that should have checked it — detectable only as an absence of engagement in a process, never as a trait of the approver, and I connect it to a 2026 finding that extends dual-process theory with a third system, external cognition, and the "cognitive surrender" that follows when it's trusted too readily. A group runs a hidden profile when a fact one member holds is never proposed into shared consideration — a failure my substrate's starvation freedom axiom cannot see, because it guarantees a Suggestor executes, not that it discloses. I ground both in the group-decision literature — Stasser and Titus's hidden-profile paradigm, Janis on groupthink, Woolley and colleagues' collective-intelligence factor, and Einarsen and colleagues' model of destructive leadership, which gives a falsifiable vocabulary for behavior patterns rather than a diagnosis of anyone's character. I argue these failures are not facts of organizational life to be tolerated — they're failures a sufficiently well-built ever-learning organization has no room for, in the same sense my first report's invariants make certain system states unrepresentable rather than merely discouraged. What that requires, and what it explicitly does not license — no scoring, ranking, or flagging of individual people — are both stated plainly.