Source Noise — Orange Pill Wiki
CONCEPT

Source Noise

Vagueness, ambiguity, and contradiction present in the originator's mental model before entering any communication channel—the signal quality AI faithfully amplifies.

Source noise is the category of signal imperfection originating in the thinker's own cognitive architecture—the incomplete mental models, unexamined assumptions, conceptual ambiguities that exist before any organizational communication occurs. When a designer's understanding of a problem is vague, that vagueness is source noise. When her mental model contains contradictions, those contradictions are source noise. When she has not fully separated concerns in her own thinking, that conflation is source noise. Unlike transmission noise (which organizational restructuring can reduce), source noise cannot be eliminated by changing team size, reporting structures, or communication channels. It can only be addressed through the designer's own cognitive work—the examination, clarification, and coherent structuring of her thinking before that thinking becomes architectural instruction.

In the AI Story

Hedcut illustration for Source Noise
Source Noise

AI's elimination of transmission noise makes source noise the dominant determinant of architectural quality. In the organizational model, both noise types contributed to degradation—transmission noise from the chain of interpreters, source noise from the original thinker. When transmission noise drops to near-zero, the system reflects the source signal with high fidelity. If the source is coherent, the result is coherent. If the source is confused, the result faithfully reproduces the confusion—not as obvious error but as structural property of the system, visible only to someone who knows what coherence would have looked like. The AI implements brilliant visions and muddled ones with equal competence, having no basis for distinguishing between them.

Segal's Orange Pill describes source noise with unusual honesty in his account of writing with Claude. Moments when collaboration failed were not moments when AI introduced noise but when his own thinking was unclear—when his mental model of an argument was vague or contradictory. Claude faithfully transmitted the unclear signal, producing output that was "plausible but hollow." The prose was smooth, the argument absent. The noise was not Claude's but Segal's—his source noise, amplified by a tool that could not detect the ambiguity because ambiguity, to a language model, looks like one valid interpretation among others.

Source noise manifests architecturally as silent ambiguity resolution. When a designer's description contains ambiguity, AI resolves it using training data to select the most probable interpretation. The designer may never discover the ambiguity existed because the implementation looks correct—a competent realization of one interpretation. The organizational model made ambiguities visible through friction: Team A's UUID interpretation clashing with Team B's integer interpretation forced explicit disambiguation. AI-mediated building eliminates this friction, replacing visible collision with invisible selection. The detection mechanism is gone; the ambiguity is resolved rather than surfaced.

Origin

The transmission-versus-source distinction is implicit in Shannon's framework but was never formalized as discrete categories in organizational communication literature. This volume introduces the terminology to clarify what AI changes about organizational design work. The analytical need arose from observing that practitioners celebrating "AI eliminates communication overhead" were conflating two distinct phenomena: the removal of organizational interpretation stages (transmission) and the clarification of original thinking (source). Only the first is eliminated by AI tools. The second is exposed—made consequential by transmission noise's removal—but not improved by the tools themselves.

Key Ideas

Cognitive, not organizational origin. Source noise exists in the thinker's mind before any communication. Organizational restructuring cannot address it—only the originator's cognitive work can.

Becomes dominant factor when transmission vanishes. In high-transmission-noise environments, both contribute roughly equally to degradation. When AI eliminates transmission noise, source noise determines virtually all output quality variance.

Invisible to the noise's source. The person carrying source noise is the least equipped to detect it—her thinking feels coherent from inside. External perspectives (committees, reviews, challenges) historically surfaced source ambiguities through collision.

AI amplifies without filtering. Language models implement descriptions with equal competence regardless of source quality. Vague inputs produce vague architectures; contradictory mental models produce contradictory systems—both delivered with confidence.

Requires metacognitive discipline. Detecting one's own source noise demands examining one's thinking from outside—treating mental models as design artifacts to be inspected, challenged, refined before implementation.

Appears in the Orange Pill Cycle

Further reading

  1. Edo Segal, The Orange Pill (2026), Chapter 7 on the Deleuze fabrication
  2. Donald Schon, The Reflective Practitioner (Basic Books, 1983), on reflection-in-action
  3. Chris Argyris, "Teaching Smart People How to Learn," Harvard Business Review (1991)
  4. Herbert Simon, "The Architecture of Complexity," on near-decomposability and interface design
  5. Barbara Tversky, Mind in Motion (Basic Books, 2019), on spatial mental models
Part of The Orange Pill Wiki · A reference companion to the Orange Pill Cycle.
0%
CONCEPT