Pitch a Documentary Series When Different Witnesses Cannot Be Reconciled
Pitch a Documentary Series When Different Witnesses Cannot Be Reconciled
A pitch document has to read as an argument. It needs a shape, a through-line, a page a commissioner can follow from the first line to the last. That pressure is exactly what produces a deck in which two people spent the same twenty minutes in the same room and the deck reports one of them.
The flattening is rarely dishonest. It is editorial hygiene. Two accounts, one chronology, problem gone. But the smoothed chronology then does a job nobody assigned it: it decides something. And once the deck has decided, every later choice — the narration, the graphics, the temp music, the shot length — inherits that verdict without ever having argued for it.
The alternative is not vagueness. It is organizing the disagreement precisely: write down what is established, write down the exact claim that is not, audit each record for the narrow point it can actually bear on, then choose a structure whose shape shows the weight of the evidence instead of smoothing it. Disagreement does not have to be resolved to be organized. It does have to be located.
The example running through this piece is invented for the exercise: a theatre opening that started late, one technician and one stage manager who remember the same delay in opposite orders, a delay log, and a backstage camera. Nothing here has been assessed. No witnesses have been interviewed, no footage reviewed, no records obtained.
Separate agreement from the exact point of conflict
Start with two columns. In the left one, put every statement that would survive whichever account turns out to be accurate. In the right, put the statements that cannot both be true.
For the invented dossier, the left column holds three things: the premiere's curtain went up later than advertised, and the delay log establishes the lateness; somewhere inside that delay the auditorium's house lights went out; a cue bell sounded. Both speakers place both of those events inside the delay. The right column holds their conflict, and it is narrow: the technician remembers the house lights failing and then the bell; the stage manager remembers the bell and then the house lights.
That narrowness is worth noticing, because it tells you what kind of disagreement you are holding. These are two first-person claims about a sequence each person says they perceived — not one person's recollection against another's inference about what must have happened. That is not always the case, and it changes the work. When one account is a claim of direct perception and the other is a reconstruction from consequence ("the show started late, so something must have failed first"), you have a disagreement about evidence, not about memory, and it should be written up that way.
Next, put a name, a role, and a place in the room beside each claim. The pitch should say where each speaker was and what that position let them see. If the technician was at the lighting desk when it happened, that is one kind of claim; if they were on the fly floor and heard about the blackout afterward, it is a different kind entirely. And if the bell was one the stage manager sounded or called, then their account of when the bell happened rests on their own action, while their account of the house lights rests on observation — a distinction the pitch can draw without deciding who is right.
Which brings up the sentence-level question: are you writing "the house lights failed" or "the technician says the house lights failed"? The attributed version is not weaker. It is the sentence whose scope matches what you have. Attribution preserves a claim's source, not its truth. Captions are not verdicts, but viewers frequently read them as currency — a named, credentialed face in a frame is a kind of evidence in itself — so the deck should know that naming someone is doing work beyond identification.
Two procedural failures are worth checking for here.
Scaling a detail into a person. If two accounts differ about the order of two events, the honest statement is that two accounts differ about the order of two events. It is not that one witness is unreliable. Sequence memory is narrow and fragile: a person can be scrupulous and accurate about everything else that night and still be wrong, or right, about which of two things happened first. A pitch that treats an ordering disagreement as a character finding has made a claim it cannot support and has usually made it for narrative convenience.
Counting one account twice. Two witnesses are not automatically two independent supports. If these two debriefed together afterward and settled on a version between them, the series now has one account with two mouths. Independence is a question about provenance: when did each person first commit to a version, to whom, and from what? If the answer is "together, in a bar, that night," the deck should say so and stop counting heads. This is one of the few places where more reporting genuinely improves the pitch — not to re-test anyone's memory, but to find out how the version they hold was formed. Accounts that descend from one retelling look like corroboration and are not.
Which records actually bear on the disagreement
For every item the deck proposes as corroboration, write three things: where it came from, what it covers, and the single narrow point it can address. Then write the sentence the film could say on the strength of it, with the item named.
Run the invented dossier through that.
The delay log. It covers lateness. The film can say: the delay log records that the curtain went up twenty-two minutes after the advertised time. That sentence is available. The order of the bell and the house lights is not, from this item, at any level of interpretation. The log's own provenance is a question the pitch should state rather than assume — who wrote it, when, and from what — because a log completed later from memory carries the same ordering problem as the witnesses, in a more authoritative-looking font.
The backstage camera. This one fails twice, and the two failures are independent. Its field of view does not include the auditorium, so it cannot show either of the two disputed events; no amount of interpretation or enhancement gets them into frame. And its clock is not aligned to any other record in the dossier, so even something useful inside the frame could not be placed on a shared time base against the log. Fixing one problem does not touch the other: a perfectly synchronized camera that never saw the room is still no help, and a camera that saw the room with an untraceable clock cannot be dated. In the dossier the camera is also merely described, so its coverage and time base are claims about equipment rather than established facts. Those should be settled before it appears in a deck as support for anything.
Then the distinction the whole section exists to protect: an image that illustrates a claim is not evidence that independently supports it. A shot of a dark auditorium under a technician's account shows what the claim describes. It does not date anything, does not come from that night, and does not establish that the darkness preceded the bell. It will still read as corroboration, which is why the episode needs one stated visual convention — what archive looks like, what reconstruction looks like, what a stand-in looks like — applied without exception, rather than a judgment call made in the edit under deadline.
Two smaller disciplines. First, absence in a record is not evidence of absence; logs get kept for billing, safety, or maintenance, and things nobody needed to write down do not appear. Second, if you cannot write the plain sentence with the record's name attached and no verbs of interpretation, the record is not in scope.
A bounded timeline or attributed accounts
Now the structure decision, using the same materials in both routes.
Event-led. A timeline restricted to the delay itself: doors opened, advertised curtain, actual curtain, twenty-two minutes between. What it does well is make the lateness a fact the viewer can hold and give the episode a spine. What it costs is subtler. An empty stretch of timeline reads as either nothing or concealment, and a timeline that has demonstrated it can place events on a line implies the next tick is coming. The form promises a resolution the evidence may not contain.
Account-led. Two interviews, matched framing, no graphic, each speaker holding the sequence they remember. What it does well is keep the disagreement legible as a disagreement about perception. What it costs is equipollence: parallel treatment announces that these are claims of comparable weight. In this dossier the weights happen to be close — both rest on unsupported first-person perception — but only close, and the film should not adopt a symmetric form until it knows whether that is true. In plenty of projects the weight will not be comparable at all, and symmetry will then be a lie told by shot lengths.
There is a decent test for which route an episode needs: name the claim the episode is actually making. If it is this production ran late and here is what the lateness cost, an event-led structure carries it. If it is these two people have worked together for six years and now they describe the same interval differently, an account-led structure carries that. The structure should follow the claim, not the other way around.
A hybrid is possible — agreed events marked on the timeline, the timeline stopping at the disputed delay, the accounts filling it, the line resuming at curtain — but it has its own hazard. Bounding a disagreement implies containing it. A slot for each account implies equal contents in the slots.
The rule that keeps any of these honest: if the evidentiary weight is uneven, the form must be uneven. Give the supported claim the graphic. Give the unsupported claim a name, a face, and its own words. Do not give both the same shot length, the same slow push, and the same music cue and then insist in the narration that no conclusion is being drawn.
Write one passage without letting style settle it
Take one passage and build it two ways. Times and staging details below are invented for the exercise, standing in for whatever the delay log actually shows.
Event-led, restricted to the delay:
Narration: Doors opened at seven. The curtain was advertised for half past, and went up at 7:52. The delay log records those twenty-two minutes and nothing that happened inside them.
[Graphic: one horizontal line. Marks at 7:00 doors, 7:30 advertised curtain, 7:52 curtain. The space between the last two marks is blank — no markers.]
Narration: Two people were in the building that night. They describe those twenty-two minutes differently.
Account-led, opening the disputed sequence:
[No graphic. Two interviews, same lens, same seat, same background.]
Technician: The house lights went first. Black, then the bell.
Stage manager: The bell went, and then the house.
Now audit them side by side.
The event-led passage establishes the lateness and the boundaries of the delay. It implies that the delay is a container whose contents the film will now disclose — a promissory implication the next beat has to honour, or the handoff reads as a stall. It also implies the times are agreed. "The delay log records" is defensible; "everyone agrees the show was late" is not, until you know both witnesses would accept those times, which this dossier does not establish.
The account-led passage establishes that two named people hold opposite orders in their own words. It implies equivalence, because two interviews of matched length in matched frames say these are equally supported, and that is close to true here but not verified. It also invites a quiet mistake in casting: if you write one speaker as flatly certain and the other as hedging, you have built a credibility verdict into performance and word choice while the narration claims neutrality. Both versions above are deliberately comparably plain.
Then watch for the moves that settle the matter without saying so.
Reconstruction. Any staged re-enactment has to pick an order to shoot. Whichever order gets shot becomes the film's version regardless of what the narration says over it. Shooting both, once per account, is more defensible — but repeated reconstruction footage accumulates authority with each pass, and the second version tends to read as a correction of the first, which is itself a verdict. A reconstruction based on one account must never quietly become the shared timeline.
Music and delivery. A swell under one account is an endorsement. So is a calmer reading. A narration that says the bell, then the house lights has resolved the conflict in the voice, and a narration that says no one can say has resolved it too, in the direction of flatness, which is a claim about the evidence that this dossier does not support.
Silence and black. Not neutral. It reads as solemnity or as concealment, both of which are statements.
The practical test for a passage is to split it. Read the narration with the images covered: if the words alone state a settled order, the passage has resolved the conflict in narration. Then imagine the images alone: if a single sequence of pictures states a settled order, the passage has resolved it visually. Most pitches audit only the first of those. Run both on the rough cut, not in the deck.
What stays unresolved, and why the pitch should say so
For this invented dossier, the hybrid is the likeliest honest default: established events on the timeline, the timeline paused at the delay, two attributed accounts, the line resumed at curtain. Likeliest is not the same as automatic. The pitch should state the structure as a decision with reasons, and state the one thing that structure cannot settle: the order of the bell and the house lights inside those twenty-two minutes.
Then say why that unresolved thing matters, which is the step most decks skip. Order is what converts a technical delay into a story about responsibility. If the series' larger premise is a production in trouble, this small sequence may be carrying more of that argument than it can bear. Test it directly: name what the series actually argues, and ask whether the argument survives with the order unresolved. If it does, say so plainly and move on. If it does not, the premise needs a record, not a better edit.
List the records you would ask for, in the pitch, as requests rather than promises: the lighting console log, if it exists and if its clock can be traced to a common time base; the call report from the stage manager's desk; the delay log's provenance; the camera's coverage and time alignment. Say what each could settle and what it could not.
Then handle the hypothetical, and mark it as one. Suppose — and this is a supposition about the fictional dossier, not a claim about any record — an independent item with a verified time base placed one of the two events first. What changes is narrow. The event-led timeline absorbs it: one marker moves, the delay gets a line of narration, and the account-led passage becomes asymmetric rather than collapsing, since asymmetry is what it should have been built to carry all along. What does not change matters just as much. That record would not validate the rest of the supported account, would not discredit the other witness, and would say nothing about causes, intentions, or what anyone had done before the doors opened. Sequence memory is narrow. A record that settles the order settles the order.
Two endings to refuse. The first is the confident one, where a persuasive voice or a well-cut reconstruction quietly closes the gap. The second is its opposite: everyone has their own truth. When two claims concern the same room, the same delay, and the same night, that sentence is not a finding — it is an abdication, and it invites the viewer to treat testimony of different weight as interchangeable. Unresolved is not the same as ambiguous. An unresolved point in a documentary has a shape: these two claims, this delay, this absent record. Ambiguity is atmosphere. This is a structure.
Close the pitch the way the film will close: with the chosen structure, the limit it must communicate on screen, and the unresolved point it preserves. Attribution stays with the claim, in the frame, at the moment the claim is made — not in a footnote after the narration has already settled the matter. Confidence in a documentary is a resource. Spend it where you can establish something, and let the stretch where two people disagree remain visibly a place where the film is still asking.
Any question about whether a real person's account can be trusted, and any legal or privacy exposure arising from actual allegations, belongs to qualified review. It is not a decision a timeline should make.
Frequently asked questions
How can a pitch organize two witnesses who cannot be reconciled?
Separate agreement from the exact point of conflict. In the theatre example, both accounts place the house lights and the cue bell inside the delay; only their order differs. Write the established statements in one column and the statements that cannot both be true in the other, then name each speaker's role and position in the room.
What does counting one account twice mean?
Two witnesses are not automatically independent supports. If they debriefed together afterward and settled on a version, the series has one account with two mouths. Independence is a question about provenance: when did each person first commit to a version, to whom, and from what?
Why does the backstage camera fail as corroboration?
It fails twice for independent reasons. Its field of view does not include the auditorium, so it cannot show either disputed event. Its clock is not aligned to any other record, so even something useful in frame could not be placed on a shared time base against the log. Fixing one problem does not fix the other.
When should an episode use an event-led timeline versus attributed accounts?
Name the claim the episode is actually making. If it is that the production ran late and here is what the lateness cost, an event-led timeline can carry it. If it is that two people who have worked together for six years describe the same interval differently, an account-led structure carries that. The structure should follow the claim.
What should the pitch say about the unresolved order of the bell and house lights?
State the chosen structure with reasons and state the one thing it cannot settle: the order of the bell and the house lights inside those twenty-two minutes. Explain why that unresolved point matters, because order is what can turn a technical delay into a story about responsibility. Then ask whether the series' larger argument survives with the order unresolved; if it does not, the premise needs a record, not a better edit.