1. Global Analysis Protocol
A four-step structure for a research assignment or a single classroom session, built around the tools already on this site rather than a separate worksheet.
Step 1 — Macro exploration & intuition
Objective: unsettle a textbook definition by looking at intra-mode divergence.
Step 2 — Proximity Matrix & custom weighting
Objective: show that stylistic kinship shifts with the analytical lens.
- Open the Proximity Matrix and locate the pair.
- Toggle the similarity weighting between Ethics & Subject and Form & Rhythm (the two named presets under "Advanced: dimension threshold & custom weighting").
- Record how the pair's score moves, and who becomes each filmmaker's nearest neighbour under each weighting.
Step 3 — Comparator deep dive & verification
Objective: cross-examine the heuristic radar against the students' own critical reading.
- Load the pair into the Comparator to overlay their eight-axis profiles.
- Isolate the widest gaps — shot rhythm, voice, sound design.
- Check each filmmaker's confidence score on Data & Confidence or their Profiles card, and discuss whether a lower figure signals a data gap or an unusually hard-to-quantify style.
Step 4 — Blend Simulator & plastic practice
Objective: apply the structural model to an actual editing or writing decision.
- Open the Blend Simulator and sequence the pair (plus a third preset, if useful) across a short film's acts.
- Read the seam row's hard-constraint text where the engine flags one.
- Have students write a short director's note on how they would bridge that seam in an actual edit — a title card, a sound bridge, a hard cut.
2. Fully Annotated Case Study: The Collision of Reality
The protocol above, worked through with two filmmakers from different theoretical branches: Frederick Wiseman and Michael Moore. Direct link to this pair already loaded: Comparator: Wiseman vs. Moore →
Step 1 — Theoretical framing
We compare Frederick Wiseman (Observational) and Michael Moore (Participatory).
This pairing is an immediate contrast for students: documentary style is not just what a camera records, but an active structural choice about time and sound.
Step 2 — Matrix reconfiguration
On the Proximity Matrix:
- Under Ethics & Subject weighting, Wiseman and Moore sit near the far ends of the corpus: Wiseman toward detachment, Moore toward direct intervention.
- Under Form & Rhythm weighting, an unexpected point of overlap appears: neither filmmaker's profile allows staged reenactment, a shared boundary on the truth axis despite the opposite execution.
Use this step to show that style is not a single sliding scale. Two filmmakers can be worlds apart ideologically while sharing structural rules about what they refuse to do.
Step 3 — Comparator deep dive
Loading the pair into the Comparator surfaces two clear contrasts:
Editing & temporal rhythm
Wiseman — long, organic shot durations. The montage follows the institution's own timeframe rather than an imposed pace.
Moore — rapid, rhetorical collage; frequent ellipsis; shots clipped short to create dialectical collisions and sustain pace.
Sound & voice profile
Wiseman — purely diegetic; no score, no explanatory voice-over. The audience reads the raw sonic environment unassisted.
Moore — constant guiding voice-over, paired with ironic music cues that reinforce the argument.
Point students to each filmmaker's confidence score on Data & Confidence: Wiseman confidence 0.85, Moore confidence 0.80 — both toward the top of the corpus, reflecting large, well-documented critical bodies of work. Section 3 below picks this thread back up with a filmmaker who sits lower, to show what that difference does and doesn't mean.
Step 4 — Act suture in the Blend Simulator
The Blend Simulator's own worked example threads a third filmmaker in: Agnès Varda opens, Wiseman holds the middle, Moore closes. Load it directly: Blend Simulator: Varda → Wiseman → Moore →
Musical score or soundtrack; Explanatory narration. That is not a coincidence of the data: those two prohibitions are absolute-severity rules that belong to Wiseman specifically, so they surface at whichever seam touches his act, regardless of which neighbour is on the other side.Ask students why the same constraint appears twice. The answer is structural, not narrative: an absolute-severity rule travels with its filmmaker, not with a direction of travel. Then ask them to write a short director's note on how they would bridge each seam in an actual edit — a title card, a sound bridge, a hard cut to black.
3. Reading Confidence Scores Critically
A core pedagogical layer of this platform: teaching students to question a data model rather than accept it at face value. Not every preset carries the same evidentiary weight, and the corpus is explicit about why on Data & Confidence.
Case study: Chris Marker and the limits of quantification
Chris Marker (Reflexive) is as canonical a figure as this corpus contains — decades of French and Anglophone film-theory writing, a stable place in the essay-film tradition, a preset the corpus classifies as fully consolidated with a solidly documented theoretical grounding. And yet his profile carries a confidence 0.65, near the lower end of the corpus's 0.60–0.90 range — noticeably below Wiseman (0.85) or Moore (0.80).
Why does a well-documented, canonical filmmaker still score lower?
- It isn't a documentation gap. Marker's status as
consolidatedwithdoc_assertedgrounding puts him in the same evidentiary tier as Wiseman. The volume of critical writing about him is not the limiting factor. - It's a measurability gap. Marker's method is associative and essayistic; his own profile flags a critical-severity refusal of linear narrative structure. A style built on digression and montage-as-argument resists being reduced to eight fixed axes more than an observational or expository style does — the corpus's own methodology names this directly as "a practice that resists conventional measurement."
- The two are independent. A high confidence score and a caution flag can sit on the same profile, and a low confidence score does not imply the filmmaker or the underlying scholarship is thin — only that this particular quantification had less to grip onto.
Research & thesis assignment prompt: have students pick a filmmaker whose confidence score sits at 0.65 or below, choose one specific film, manually sample roughly fifteen minutes of footage against the eight axes, and write a short critical assessment of where the platform's heuristic model holds up and where the chosen style simply resists the eight-axis frame.