The Algorithmic Cut

CILECT VIGS · Bilkent University COMD

The Algorithmic Cut

Agency and automation in the future of editing

A CILECT Vision & Innovation Grant project exploring what editors gain and lose when AI joins the cutting room. Five exercises, a companion book, and an ongoing pilot at Bilkent COMD.

Explore the exercises Read the briefing

Pilot sessions: Oct–Dec 2026 · CILECT Congress, Dublin: 27–31 October 2026

Why it matters

The edit is no longer just yours

Before a single cut is made, AI has already transcribed, tagged, ranked and pre-interpreted the footage. These exercises put that process under critical pressure.

The Algorithmic Native

A new generation of editors trained inside AI-saturated workflows, who may never have experienced footage without algorithmic pre-interpretation.

The Paradox of Skill

Automation handles low-complexity tasks but demands new high-complexity skills: directing the algorithm, reading its biases, reclaiming what it discards.

Interface Criticism

Reading the editing interface as an argument: every default is a claim about what good editing is. Making that argument visible is the project’s method.

What we do

Five activities. Five productive frictions.

Each exercise introduces a specific kind of resistance between the editor and the automated workflow.

The Inversion Exercise

Reclaiming what the algorithm flags as waste.

Feed and Frame

Platform logic and the Recommendation Editor.

The Answering Cut

The represented person responds.

What’s next

Project timeline

Oct–Dec 2026

Pilot phase: Inversion & Turing Cut

First two exercises run with students. Research assistants document responses. Project presented at CILECT Congress, Dublin.

Jan–Apr 2027

Production: portal launch

HD video demonstrations, multilingual protocols, visual asset library. Portal launch in March. Peer-reviewed article submitted in April.

May 2027

Grant project closes

Findings continue to circulate through the CILECT network.

Oct 2027

Book submitted

Full draft in August, reviewed by external colleagues, revised in September, submitted to a publisher in October.

Latest

From the blog

EssayInterface Criticism October 2026

How AI is Quietly Rewriting the Language of Film (And How Editors Can Fight Back)

ProjectTeam September 2026

First team meeting: from a book idea to a classroom toolkit

The project began as a book on AI-controlled editing and grew into educational tools for film schools. Read more

Documentary

Film & Documentation

Every pilot workshop is filmed by its own participants. The footage serves two purposes: a research record and the possible raw material for a documentary short. Learn more

Companion book

The Algorithmic Cut: Agency and Automation in the Future of Editing

Andreas Treske’s scholarly companion to the project, developing Interface Criticism as a method. Full draft planned for summer 2027. See the outline

Framework

Briefing: Interface Criticism

Interface Criticism reads the editing interface as an argument: its defaults, recommendations and automations express a claim about what good editing is. Making those claims explicit is the first step toward directing the algorithm rather than being directed by it.

The Algorithmic Cut is neither a polemic against AI tools nor an uncritical celebration of them. It asks a specific question: what are the aesthetic, narrative and ethical consequences of delegating decisions to automated systems? Five exercises introduce productive friction into automated workflows so that editors can answer that question for themselves.

“Every automated default is a claim about what good editing looks like. Interface Criticism makes that claim visible.”

Core Concepts

The Algorithmic Native

A generation of editors who trained inside AI-saturated workflows and may never have experienced raw footage without pre-interpretation. The exercises are designed to defamiliarise the familiar.

The Paradox of Skill

Automation reduces the need for low-complexity craft skills (rough sync, basic colour) while demanding new high-complexity skills: reading algorithmic bias, directing the machine, knowing when to refuse its suggestions.

From Montage to Prompting

The conceptual shift in the editor’s role: from constructing a sequence shot by shot to directing a system that proposes sequences. Prompting is not editing; the distinction matters.

The Recommendation Editor

An editor whose decisions are shaped not only by creative intent and the footage, but by the discovery systems that will decide whether the edit is seen at all. Platform logic enters the cutting room.

Algorithmic Defaults vs. Human Editorial Values

What the algorithm optimises for	What it tends to marginalise
Technical sharpness (focus, exposure, stability)	Affective texture: soft focus, shake, room tone as expressive elements
Speech density and pace	Silence, micro-pauses, held breath, the thinking face
Engagement metrics (watch time, completion rate)	Argument, duration, the “slow burn”
Platform-standard aspect ratios	Spatial and social context discarded by auto-reframe
The “and then” cut (continuity, rhythm)	The “therefore” cut (argument, consequence)

Integrating the Framework

The exercises in the Activity Guide are not abstract theory. Each one is derived from a chapter of the companion book and is designed to be completed in a single session with standard post-production software. Instructors do not need to adopt the full theoretical framework to use them; the exercises work as standalone interventions in any post-production or documentary editing course.

Classroom exercises

Activity Guide

Five exercises that introduce productive friction into automated workflows. Activities 1 and 2 are the pilot’s core exercises; full instructor protocols and rubrics will follow. The first sessions draw on existing CILECT material, with student-filmed footage a possible later addition.

1. The Inversion Exercise: reclaiming “waste”

Goal: reverse the algorithm’s definition of low-quality footage to find narrative value it discards.

Manual prediction (no software). From 10–15 minutes of raw documentary or interview rushes, sort clips into “Keep” and “Flagged as Waste”. Write one sentence per flagged clip naming the technical cue (soft focus, shake, room tone) you expect an algorithm to reject.
Build the sequence. Cut a continuous 2-minute sequence using only flagged clips, with no stabilisation, colour or audio repair.
Reveal (optional). If a scoring plugin or retake detector is available, compare its flags with your predictions and note divergences.
Reflect. Where did a flagged clip carry more weight than a clean one? What aesthetic biases did you assume the software enforces?

2. The Turing Cut: machine pacing vs. human breath

Goal: find where automated pacing fails to replicate human affective timing.

Two versions (60–90 s) of one scene. A: automated cut using a tool such as CapCut, Descript or an AI assistant, with silence and filler removal on. B: manual cut prioritising subtext, reaction shots and natural pauses.
Blind test. Label versions “Option 1” and “Option 2”, normalise loudness (e.g. −24 LUFS), screen without revealing which is which, collect a 1–10 “human-feel” rating and the timestamp where viewers sense automation.
Analyse. Did the manual cut keep micro-pauses and eyeline shifts? Did hyper-cutting move the scene from captured reality toward the synthetic? Was the faster edit emotionally shallower?

3. Spatial Politics: auto-reframe as narrative decision

Goal: treat reframing as an uncredited editorial choice, not a format conversion.

Take a 16:9 wide shot with environmental context or several subjects. Convert it to 9:16 with Auto Reframe, Smart Reframe or Smart Conform, then document three narrative, social or spatial elements the crop erased.

4. Feed and Frame: platform logic and the Recommendation Editor

Goal: see how distribution formats and discovery systems shape editing.

Produce a baseline 16:9 cut and a 9:16 adaptation with burned-in captions, re-pacing and shot re-selection. Keep a decision log and classify each change as driven by frame constraints, sound-off viewing assumptions, or recommendation-algorithm optimisation.

5. The Answering Cut: the represented person responds

Goal: extend Interface Criticism beyond the screen by involving the person in the image. Proposed by filmmaker Nathan George, building on his documentary FUTURE.

Show the subject Version A (isolating one algorithmic operation, such as auto-pacing or reframing) and Version B (human edit). Their response is not a vote or veto; it shows where meaning entered the image and may reveal blind spots shared by editor and algorithm, even at the camera stage. The editor then records one editorial consequence:

Keep Re-edit Reframe Return to footage New image

Student Checklist: Directing the Algorithm

Embrace productive messiness: sustained silences and rough cuts over smooth templates.
Maintain structural literacy: hold the architecture of the edit in mind; construct, don’t just curate.
Make the edit visible: prefer the “therefore” cut to the “and then” cut.
Keep ontological clarity: know what is profilmic, filmic and synthesized.
Choose breath over speed when the scene needs human timing.

Laboratory stages

Roadmap

The Algorithmic Cut develops through five stages, each with its own directive, task and synthesis. The first stage is the pilot currently under way at Bilkent COMD.

Stage 1 — Critical Seeing: the database pre-edit

Directive: identify what is lost and found when AI pre-interprets raw footage.

Task: compare AI-generated selects against editor-generated selects from the same rushes; document the divergences. Run the Inversion Exercise.

Synthesis: a critical analysis of the AI confidence score as an editorial argument. What assumptions does it encode? What footage does it structurally devalue?

Stage 2 — Temporal Craft: pacing and the affective cut

Directive: locate where automated pacing cannot replicate human affective timing.

Task: run the Turing Cut experiment; analyse the blind-test data; identify the precise moments where viewers detect automation.

Synthesis: a taxonomy of the micro-decisions that distinguish a human cut from an AI cut in emotionally complex scenes.

Stage 3 — Spatial and Platform Politics

Directive: map the editorial consequences of format conversion and platform optimisation.

Task: document the spatial, social and narrative elements erased by auto-reframe (Exercise 3); build a family of platform edits and log each decision (Exercise 4).

Synthesis: an analysis of the Recommendation Editor as a new editorial role shaped by discovery systems.

Stage 4 — Ethics of Representation: the Answering Cut

Directive: extend Interface Criticism to the represented subject.

Task: run Exercise 5 with documentary subjects; record editorial consequences; analyse what the subject’s response reveals that neither the editor nor the algorithm could see.

Synthesis: a framework for the ethical dimensions of algorithmic representation in documentary editing.

Stage 5 — Publication and Dissemination

Directive: translate pilot findings into transferable resources for the international film education community.

Task: portal launch, multilingual protocols, peer-reviewed article, CILECT dissemination, companion book.

Synthesis: The Algorithmic Cut: Agency and Automation in the Future of Editing submitted to a publisher.

Curriculum Mapping

Exercise	Relevant courses	Key skills
1. Inversion Exercise	Introduction to editing, Documentary production	Critical viewing, footage evaluation, aesthetic judgement
2. Turing Cut	Advanced editing, Post-production for documentary	Affective timing, audience analysis, comparative editing
3. Spatial Politics	Multi-platform storytelling, Social media production	Frame composition, spatial analysis, format literacy
4. Feed and Frame	Digital distribution, Transmedia production	Platform literacy, decision logging, algorithm awareness
5. Answering Cut	Documentary ethics, Advanced documentary	Subject relations, editorial ethics, representation analysis

Scholarly companion

The Book

The Algorithmic Cut: Agency and Automation in the Future of Editing by Andreas Treske is the scholarly companion to the project. It develops Interface Criticism as a method and argues for structural literacy as the editor’s response to automation. Each chapter closes with an Interface Critique: a practical exercise that makes the argument testable in the classroom. The book is in development, with a full draft planned for summer 2027.

Not a polemic against AI tools, but a demand for critical literacy: understanding what the algorithm optimises for, what it marginalises, and when it is the right tool for the job.

Introduction: The Crisis of the Cut

The cut was once a deliberate, human act. Now much of the territory of possibility is shaped before an editor watches a frame. The introduction presents the Algorithmic Native, the Paradox of Skill, the shift from montage to prompting, and Interface Criticism as the book’s method.

Part I: The Logic of the Database

Selection. How algorithmic systems restructure the way footage is encountered, organised and pre-evaluated before any creative decision is made.

Chapter	What it argues	Interface Critique
1. The Ontology of the Bin	The digital bin is no longer a neutral container but an active agent that sorts, ranks and pre-interprets footage. Draws on Manovich’s database logic and examines the AI Confidence Score and metadata as pre-edit.	Confidence Score Deconstruction
2. The Phenomenology of the Rush	How editors watch, or should watch, raw footage. Extends Barthes’s punctum to moving images, with Sobchack’s haptic visuality, emergent structure and the paper edit as analog counter-practice.	Active Listening Exercise

Part II: The Architecture of Time

Structure. How an editor thinks in duration, and why algorithmic systems optimise time for engagement rather than meaning.

Chapter	What it argues	Interface Critique
3. Paralysis and the Interface	Infinite tracks, unlimited undo and AI suggestions can immobilise the editor. Reads the blank-canvas problem through Flusser’s apparatus and proposes strategies of commitment.	Anti-Template Workshop
4. The Essayistic Mode	The essay film as the strongest counter-practice to AI-optimised editing: the argumentative “therefore” cut against the smooth “and then” cut.	Argument Sequence

Part III: The Techno-Cultural Format

Form. How platforms format, distribute and recommend edited work, so that format and distribution shape editorial meaning.

Chapter	What it argues	Interface Critique
5. Verticality and the Square	The spatial politics of the 9:16 frame and of automated reframing, analysed through platform determinism.	Reframing Analysis
6. The Loop	The loop as the default temporal form of platform culture, and the difference between the concealed loop and the revealed loop.	Seamless Loop Exercise
7. The Feed and the Frame	The systems that decide whether an edit is seen at all: algorithmic discovery, captions, and building a family of edits for a family of platforms.	Repurposing Cascade

Part IV: The Ethics of the Black Box

Agency. What editors are responsible for when their tools make decisions they may not understand, endorse or even notice.

Chapter	What it argues	Interface Critique
8. Interface Criticism: Transcript-Based Editing	The book’s fullest demonstration of Interface Criticism as method. Treating a transcript as the footage reduces film to language and discards sighs, silences and held breaths.	Transcript Comparison
9. The Turing Cut	Adapts Turing’s test to the editing suite: can viewers tell a human edit from an AI edit, and what does their judgement reveal about automated pacing?	Turing Cut Experiment
10. The Truth of the Splice	Uses Souriau’s levels of filmic reality, from the profilmic to the synthesized, to ask what truth claims an edit makes in an age of generative images.	Souriau Analysis

Conclusion: The Director of the Algorithm

Refusing both uncritical enthusiasm and reflexive rejection, the conclusion proposes the editor as Director of the Algorithm: a practitioner who commands automated tools rather than being commanded by them, and who looks ahead to generative systems with structural literacy intact.

From book to classroom: the portal’s exercises are adapted from the book’s Interface Critiques and tested in the Bilkent pilot. The Inversion Exercise grows from Chapter 1, the Turing Cut from Chapter 9, Spatial Politics from Chapter 5 and Feed and Frame from Chapter 7.

Timeline

Project Schedule

The Algorithmic Cut runs from autumn 2026 to the end of May 2027 as a CILECT Vision & Innovation Grant project, with the companion book continuing through 2027. Teaching dates follow the Bilkent semester.

Period	Project and resources	Companion book
Sept 2026 Foundation	Research team in place and first project meeting held. Portal architecture drafted; exercise protocols and chapter map in preparation.	Chapter outlines complete and drafting under way.
Oct–Dec 2026 Pilot phase	Inversion Exercise and Turing Cut run with students at COMD, with research assistants documenting responses and collecting feedback. Rubrics revised on the evidence. Student case studies compiled by December. Project presented at the CILECT Congress, Dublin, 27–31 October 2026.	Chapters closest to the pilot exercises drafted alongside the classroom work and informed by pilot data.
Jan–Apr 2027 Production	Jan: HD video demonstrations; French and Spanish translations begin. Feb: visual asset library and multilingual protocols completed. Mar: Portal launch Apr: peer-reviewed article on pilot findings submitted. Spring: presentation at the CILECT GEECT Regional Meeting.	Drafting continues through Parts I–III: selection, structure and format.
May–Jun 2027 Closeout	Grant project completes at the end of May.	Part IV, including the philosophically demanding chapter on the truth of the splice, in draft.
Jul–Oct 2027 Book	Findings continue to circulate through the CILECT network.	Introduction and conclusion written last. Full draft in August, read by external colleagues, revised in September, and submitted to a publisher in October.

Teaching and pilot dates are fixed to the semester; the resource and book dates may shift as the pilot teaches us what the exercises need.

Documentation

Film & Documentation

Every pilot workshop is filmed by its own participants. The footage serves two purposes: a research record of what happens when students encounter automated editing tools, and the possible raw material for a short documentary.

Research documentation

A systematic record of student responses, editorial decisions, and moments of friction between human and algorithmic editing. Research assistants Nehir and Çiçek document the pilot exercises and collect feedback for rubric revision.

Documentary possibility

The workshop footage may become a short documentary about the project itself: students engaging with algorithmic tools, encountering their limitations, and finding the limits of their own assumptions. Co-developed with Nathan George.

Collaboration with Nathan George

Filmmaker Nathan George (director of FUTURE, Béziers, France) is a co-researcher on the project. His contributions include:

Proposing Exercise 5, The Answering Cut, which involves documentary subjects in the evaluation of their own representation
Co-designing the documentation approach and the possible documentary
Bringing the perspective of a working documentary filmmaker to the theoretical framework

The Change of State Rule

Document the moment when something changes state: when a student first sees the AI’s version of their footage, when they choose to override or accept a suggestion, when they cannot tell their own edit from the machine’s.

Moments always documented

First encounter with AI selects or AI-generated rough cut
Decision to accept, override or ignore an AI suggestion
Blind-test screening and audience response (Turing Cut)
Subject response in the Answering Cut exercise
Debrief discussions where students articulate what they noticed

Three Camera Functions

Evidence

Recording what actually happens in the workshop: student decisions, moments of hesitation, unexpected discoveries.

Illustration

Close-ups of interfaces, comparison screens, the before and after of an algorithmic edit. Making the invisible decision visible.

Reflection

Students and instructors on camera after the exercise: what surprised them, what the machine got right, what it could not do.

Principles

Participants film the workshop; the documentation is itself an exercise in the project’s themes.
No footage is used without the consent of everyone in it.
The documentary, if made, will be subject to the Answering Cut: subjects will see how they are represented.
The research record and the possible documentary are kept as separate projects; the second does not compromise the first.

Notes from the pilot

Blog

Notes, observations and findings from the pilot at Bilkent COMD, written by the project team and student research assistants.

October 2026 · EssayInterface Criticism

How AI is Quietly Rewriting the Language of Film (And How Editors Can Fight Back)

Before you make a single cut, your editing software has already tagged faces, transcribed dialogue and ranked your rushes. Six takeaways on becoming the Director of the Algorithm rather than its curator: from montage to prompting, what the machine flags as noise, the silence problem in text-based editing, the Recommendation Editor, “and then” versus “therefore”, and the Answering Cut.

September 2026 · ProjectTeam

First team meeting: from a book idea to a classroom toolkit

The project began as a book on AI-controlled editing and grew into educational tools for film schools. At the first team meeting, the research team gathered in Ankara and online to introduce themselves: research assistants, co-researchers in Turkey and France, and colleagues from the Media Archaeology Lab.

The meeting settled the shape of the work: five exercises that test automated editing against human editors. The fifth, proposed by filmmaker Nathan George, has people film themselves and then compare how AI and human editors represent them. The team agreed to start the first exercises with existing CILECT material, to film the workshop process on phones and cameras for documentation and possible further use, and to plan the first workshop together.

People

Project Team

The Algorithmic Cut is a CILECT VIGS project based at the Bilkent University Department of Communication and Design, developed with colleagues in Ankara and in France.

Project Lead

Andreas Treske

Project Lead · Chair, Dept. of Communication and Design

Media theorist and author of Video Theory and Heaven’s Delight. Founder of COMD’s Media Archaeology Lab.

Research Assistants

NÜ

Nehir Üstündağ

Research Assistant

Graduate student at COMD interested in media archaeology and narratology. Studies how digital interfaces and systems influence meaning-making.

Çiçek Ertuğrul

Research Assistant

Graduate student at COMD. Enthusiast of analog media formats and experimental cinema.

Co-researchers

Nathan George

Co-researcher · Filmmaker, France

Filmmaker and editor based in France, currently establishing the independent production company Quartier Libre Films. His recent works include VERTICAL, FUTURE and FURAX, exploring duration, representation and editorial agency. He joins The Algorithmic Cut as a co-researcher, developing The Answering Cut and an audiovisual documentation method examining the relationship between algorithms, editors and the people represented in images.

Melih Aydınat

Co-researcher · Instructor, COMD

Teaches post-production in the spring semester.

Team Members

İNÇ

Ípek Naz Çelen

Team Member · Editor

First assistant editor on a music documentary.

Ali Olcay Aracı

Team Member

Graduate student in COMD interested in experimental film, contemporary art and media archaeology. Studies film archives and their relation to culture.

Alinur Bağcı

Team Member · Head Assistant, Media Archaeology Lab

Supports the project through COMD’s Media Archaeology Lab.

Contact

Questions or interest in joining the pilot? comd@bilkent.edu.tr

Bilkent University, FADA · Department of Communication and Design · 06800 Çankaya, Ankara

Supported by a CILECT Vision & Innovation Grant (VIGS) and hosted at Bilkent University, Department of Communication and Design · comd@bilkent.edu.tr