P4 THEORETICAL PROPOSAL

Governance of nextness and the right to stop

Platforms turn the viewing that just occurred into a ranking condition for the next item. The right to stop tests whether exit, return, and refusal of recommendation are genuinely executable.

01
CLAIM

Central claim

A platform governs nextness not by predicting the future with certainty, but by turning the viewing that just occurred into ranking signals and using them to arrange the next field of visibility. A right to stop requires continuation, exit, return, deletion of behavioural traces, and refusal of recommendation to be intelligible, executable choices that are not covertly cancelled.

The same claim without specialist shorthand

The problem is not simply that videos are short or scrolling is fast. Pausing, skipping, replaying, and leaving can all become material for arranging what appears next. Stopping must alter that feedback chain, not merely freeze the screen.

The contemporary problem it addresses

Image research often studies how one work orders its shots, while asking less often who calculates the order between works, how viewing returns as data, and whether an interface's exit actually ends profiling, recommendation, and later recall.

02
INHERITANCE / MODIFICATION

Relations to Deleuze and later scholarship

Inheritance note

The proposition inherits the sensory-motor analysis of how perception is prolonged into action, along with the requirement that actual and virtual remain distinct. A model's probability space is computed by a technical system and cannot be equated directly with Deleuze's virtual.

Modification note

The modification moves the unit of analysis from a perception-action circuit within one shot to a feedback sequence across images: viewing is recorded, ranking is updated, and the next visible item changes. It also adds an institutional test of whether a person can leave the sequence and withdraw the conditions under which past behaviour keeps being used.

Relation by relation

4 entries
03
FILMS AND CASES

Films and case analyses

  • Research chapter

    Research chapter 11: algorithmic futures

    Observe

    Record which visible signals a viewing leaves, which enter ranking, how the next item changes, and whether a user can inspect that causal chain.

    Decision rule

    Use 'governance of nextness' only when a checkable connection shows past behaviour entering the next selection. Infinite scroll or short-video appearance is insufficient.

  • Research chapter

    Research chapter 12: generative-image provenance

    Observe

    Separate prompt, model version, training-source disclosure, seed, selection, regeneration, and publication, then test which operation narrows or reopens possible results.

    Decision rule

    A changed model output proves a change in computational conditions. It does not establish a new time-image unless the relation between image and viewing is also reorganized.

04
FALSIFIERS / WHAT WOULD MAKE IT FAIL

Counterexamples and their consequences

  • Falsifier

    The sequence does not use viewing feedback

    If found

    Checkable product documentation, experiment, or logs show that the sequence is fixed and viewing behaviour does not enter later selection.

    Recorded consequence

    The case is not feedback-based governance of nextness. Analysis should return to editing, programming, or linear exhibition.

  • Falsifier

    Stopping actually ends subsequent use

    If found

    A person can stop continuous playback, disable personalization, withdraw behavioural records, and verify that old data no longer enters later ranking, with equally legible controls.

    Recorded consequence

    For that system and version, the criticism that the right to stop is hollow does not hold. The proposition's scope must narrow.

05
SCOPE / EXCLUSIONS / LIMITS

Conditions and theoretical limits

Scope conditions
  1. Applies to systems in which viewing behaviour can enter later ranking, recommendation, generation, or recall.

  2. Platforms, models, and interfaces change. Every judgement must name product, version, date, region, and account conditions.

Not claiming
  1. It does not claim that an algorithm knows or determines the whole future.

  2. It does not claim that long content, slow images, or art images inherently resist platforms.

  3. It does not attribute 'governance of nextness' or 'right to stop' to Deleuze's vocabulary.

Evidence limitations

  1. The proposition comes from an authorial synthesis in the atlas's existing research report. Platform facts require dated official documentation, independent research, or reproducible tests; conceptual inference cannot replace that evidence.

  2. The present register does not claim to have audited a specific recommender system, and it does not treat P4's planned exercises as completed empirical results.

Simplifications it opposes

  1. Platform temporality can be fully explained by a fast-image versus slow-image opposition.

  2. A recommender system knows a future that has not happened.

  3. A model's probability space is the virtual in Deleuze's sense.

  4. A pause icon or close button is sufficient to establish a right to stop.

06