Chapter 03 · the bet
Every module is a hypothesis.
Seven hypotheses held against the evidence, two of them refuted and redesigned. Three distinct ways to solve the problem, scored honestly. Then the one idea you can try for yourself.
Change · who · impact · how much · by when · and the verdict so far
Each hypothesis, held against the evidence.
| # | Hypothesis | Level | Evidence so far | Verdict |
|---|---|---|---|---|
| H1 | Small teams re-key script changes into 3+ prep documents by hand. | Problem | Vendor admission of manual entry; 0 of 20+ tools sync live; seen on own sets. | Supported |
| H2 | Students and indies won't pay studio prices: under ₹500 a month is the ceiling. | Value | Value-for-money is Final Draft's weakest score; StudioBinder is "too high for independents"; writer fees under ₹50k. | Supported |
| H3 | If prep reads live from the script, ≥30% of finished drafts add a breakdown within 14 days. | Value | KOAN went script → breakdown → shots → schedule in one tool (n=1). | Early signal |
| H4 | Marking the real page beats form-based tagging for first-time ADs. | Feature | The form version was ignored in favour of printed PDFs; the marked-page version replaced it. | Early signal |
| H5 | Numbered steps fix shot-list findability. | Design | Tab names failed with a real user, then the redesign shipped. | v1 refuted → redesigned |
| H6 | Theory tools ("psychological geometry") help directors prepare. | Feature | Produced nothing a crew could use. | Refuted → retired |
| H7 | Compact documents keep 50 users within Firestore's free quota. | Feasibility | Architecture designed to the budget: computed plans, writes only on change. | To measure at scale |
Outcome → problem → solution · RICE
Three distinct theories. One compounds.
| Option | Theory | R | I | C | E | Score |
|---|---|---|---|---|---|---|
| A · Glue | Keep the stack; build import and export bridges. | 8 | 1 | 50% | 3 | 1.3 |
| B · Generator | AI reads a finished script and spits out prep. | 8 | 2 | 40% | 5 | 1.3 |
| C · One source | The screenplay is the source; prep documents are live views of it. | 6 | 3 | 60% | 8 | 1.4 |
The scores are close, and saying so is the honest finding. C wins on impact, because it removes drift (P1) instead of speeding up re-keying, and because it compounds: every module makes the others more useful. A is easy to copy. B produces a one-shot copy that drifts at the next rewrite, so AI belongs inside C as an assistant.
Seven problems · each mapped to the modules that solve it
What each solution solves, and for whom.
| Problem | Solved by | For students | For indie filmmakers |
|---|---|---|---|
| P1Drift: rewrites never reach prep | Production scriptBreakdownShots & BoardsSchedule & Call Sheets | Their thesis paperwork stays right without a coordinator. | No re-keying, and no wrong day on set. |
| P2Cost and fragmentation: 4–6 paid tools | One workspaceFree-tier architectureImport FDX/PDF | A student can afford it. | One subscription; bring old scripts in. |
| P3Craft gap: first-timers don't know the paper | Industry-layout sheetsStripboardShooting bookPaperwork (India) | Learn by making the real artefact, with no lectures. | Crew-ready paper on day one. |
| P4Fragmented time: writing in bursts, on phones | Phone layoutFragmentsScene historyOffline sync | Write between classes. | Rewrite on set from a phone. |
| P5Story scattered: characters, world, outline go stale | CharactersWorld & LoreOutlineResearchPitch | One place to develop the idea. | Pitch decks that match the draft. |
| P6Collaboration and trust: feedback in WhatsApp, leaks | Comments & presenceSharing rolesActor's copyDelivery packRevision mode | Mentors comment on the exact line. | Each crew member gets only their pages. |
| P7Long-form continuity: series and micro-dramas | Series & franchiseSecrets matrixContinuity watchdog | Web-series projects stay consistent. | Micro-drama seasons with canon tracked. |
10⁻¹⁸ metres · the fundamental unit
Three quarks hold it together.
- One sourceThe screenplay is the truth; everything else is a view of it.
- Paper and data in one actMark the page and the sheet writes itself. Either prints. No coaching prompts, no theory sliders: every field feeds a real artefact.
- The writer stays the authorStructure is optional, out-of-order drafts are fine, and AI suggests while the writer decides.
"Three quarks for Muster Mark!" — Joyce, the line Gell-Mann named the quark after
Under the scene · start with the data flow
Script-bound modules read. They never write back.
Functional
Write, import and export (FDX, PDF, Fountain), break down, plan shots, schedule, issue call sheets, share by role, comment in context, keep series canon.
Non-functional
Never lose a word (revision stamps + compare-and-set). Screen = paper (one pagination engine). Access enforced in the database. Works offline. Runs on the free tier for the first 50 users.
Scene 12 · ⅞ of a page · live demo
Change one word. Watch production follow.
NIDHI (34) sits on the parapet, a red notebook open on her knees. The city hums below.
ARJUN (O.S.)
You didn't tell them.
Tap INT / EXT or DAY / NIGHT / DUSK up there.
Every sheet on the right rewrites itself.