Immanent critique is the Husserl volume's name for the continuous self-evaluation that occurs within the stream of creative activity. In temporally articulated work, the builder continuously evaluates the just-completed step against the accumulated understanding of the project's requirements, the aesthetic standards internalized through years of practice, the functional criteria the product must satisfy. This evaluation is retentionally grounded: it requires holding the just-completed step in awareness while simultaneously bringing evaluative standards to bear on it. When retentional articulation collapses under AI-augmented work, immanent critique weakens. The builder evaluates each AI response against the immediately preceding prompt rather than against the full, temporally articulated sequence of decisions that have shaped the project. Evaluation becomes local rather than global — assessing each response in isolation rather than in the context of the project's evolving trajectory. This is the mechanism behind the Deleuze error: a passage evaluated and found satisfactory within the narrow retentional window of the immediate interaction, but disconnected from the genuine context that retentionally deeper engagement would have recognized.
The concept specifies what phenomenological depth contributes to evaluative activity. The builder possessing the full retentional articulation of the project can evaluate each new element against the whole — asking whether the latest addition fits, whether it preserves or disrupts the project's trajectory, whether it meets the standards the earlier work established.
The builder whose retention has collapsed can evaluate only locally. The question becomes 'does this response answer the prompt I just issued?' rather than 'does this addition advance the project I have been building?' The local question is easier. The global question requires phenomenological depth the absorbed state does not provide.
The concept also illuminates the specific danger of the fluency trap: AI output is often locally satisfactory — it answers the prompt, it fits the immediate context, it reads coherently. Its failures are typically global: it does not fit the larger project, it contradicts earlier material, it draws on reference frames inconsistent with the work's trajectory. Without immanent critique operating at global scale, these failures are invisible.
The Deleuze error Segal describes is a paradigmatic instance. Locally, the passage connecting Csikszentmihalyi's flow to Deleuze's 'smooth space' read as genuine insight. Globally — in the context of what Deleuze actually argued — the connection was wrong. Immanent critique at global scale would have caught the error; the collapsed retentional field operating at local scale did not.
The concept draws on the Frankfurt School tradition of immanent critique (Adorno, Horkheimer) while giving it specifically phenomenological grounding. The Husserl simulation in the Orange Pill cycle articulates the phenomenological mechanism by which immanent critique operates and the specific way retentional collapse undermines it.
It connects to Blair's analysis of ars critica — the Renaissance art of critical reading developed in response to the first information abundance — as the intellectual ancestor of the evaluative discipline AI-era work demands.
Ongoing, not episodic. Immanent critique operates continuously within activity, not only at designated evaluation moments.
Retentionally grounded. The capacity requires holding the project's temporal trajectory in awareness while evaluating current additions.
Local vs global. Local evaluation answers 'does this fit the immediate context?'; global evaluation answers 'does this advance the project?'
AI output is often locally satisfactory, globally problematic. The failures that matter are typically invisible at the scale of the interaction.
Retentional collapse converts global to local. When the temporal articulation of the project is lost, evaluation defaults to interaction-scale assessment.