The Debate Format Gives Equal Time to Unequal Evidence. What Should Change?

A symmetrical running order makes a claim about the material before anyone speaks. Two chairs, two guests, equal clocks: the shape says that two positions were worth assembling and that they were comparably worth assembling. Sometimes that is true. When it isn't, the shape keeps asserting it for the whole episode, and no amount of careful talk from the host undoes the arrangement that the audience is watching.

So the answer is not to abandon adversarial conversation. It is to decide, before casting, which part of your material is actually in dispute—and to let that answer choose the structure. That means four moves: inventory the material by claim type, work out what your format's shape will make viewers infer, pick the structure that fits the type of claim you have, and give the host a defined job when the frame starts to slip.

Unpack the proposed question before casting its sides

Start with a list, not a title. For each claim you intend to put on air, write down what supports it, what the support covers, and what would still be unresolved if the claim were settled tomorrow. Those last two categories do most of the work, because they tell you whether you have an evidentiary dispute at all or a disagreement about priorities wearing a factual costume.

Take an invented case, used throughout this piece: a shared woodworking room with roughly twenty regular users has to decide where to keep its hand tools. Two layouts are proposed. Layout A hangs everything on a pegboard along the long wall; the floor stays clear. Layout B uses a five-drawer rolling cabinet parked at the end of the assembly bench, occupying about 0.6 square metres of floor. The room's observation record—invented for this exercise—contains this much: on one Saturday morning, one member ran the same task twice on each layout, alternating between them, timed while retrieving and returning the eight tools needed to cut and assemble a small frame. Layout A: 3:20 and 3:05. Layout B: 2:40 and 2:55. Mean times of 3:13 and 2:48, a difference of about twenty-five seconds.

That record supports one bounded statement: under those conditions, with that member, on that task, in that order, the cabinet layout produced shorter elapsed times. Someone who says "the cabinet is faster" has moved from that statement to a general one. The gap between the two is where formats get into trouble.

Three more things sit in the same pile. Someone argues the cabinet's drawers rattle and the room should stay quiet. Someone objects that the pegboard's top row is too high. Someone else wants the floor kept clear because it doubles as standing room when two people are working. Only one of these is straightforwardly a priority claim. Floor space is a value: how much clear floor is worth relative to twenty-five seconds is a judgement, and no further testing will convert it into a fact. Noise is half-factual—a measured decibel figure would exist if anyone had taken one—and half-preference, since whether a rattle is objectionable is a judgement about how the room should feel. Reach is where producers most often fool themselves. "I can't use the top row" is a testable claim about a person and a shelf height, and the record contains no measurement of it. The room may hold it as a value ("tools shouldn't be above eye level"), in which case it belongs in a preference discussion, but it is currently being argued as though it were a fact that has been checked.

Settle the classification before you settle the casting. A factual claim with support on one side and none on the other is not a debate between equal positions; it is an episode whose real subject is something else—usually the priority question underneath, which is often worth the airtime on its own.

Warning against one shortcut: do not let a pleasing binary title do this work for you. "Pegboard or cabinet?" is a catchy frame that will quietly tell your researchers to find two equivalent evidentiary positions, and there aren't two. The title is a marketing artefact. The claim inventory is the editorial one.

Test what the adversarial structure makes viewers infer

Now map the machinery. Introductions assign standing. Alternating turns imply that what one guest says is answerable by what the next says. Rebuttal segments imply that a claim was engaged on the merits. Graphics imply quantification—a bar chart comparing two layouts will read as a measurement even if the underlying record is four stopwatch readings. The host's closing summary implies the episode's verdict. Each part distributes authority whether or not anyone intended it to.

Here is the failure version of this episode, in outline. Cold open: "Two ways to store the tools. Which is faster? Our guests disagree." Four minutes each, alternating, with two-minute rebuttals. A graphic of the two mean times, unlabelled as to sample. Close: "We've heard both sides. The room will have to decide."

Watch what a viewer takes away. First, that "which is faster" is genuinely open at the level the guests were contesting—when the record is small but the disagreement on air is a general one, viewers reasonably read the gap as a gap in the evidence. Second, that the two positions have comparable support, because they received comparable treatment. Third, and least visible, that the twenty-five seconds and the clear floor are competing evidence claims, when one is a measurement from a single morning and the other is a preference that no measurement will settle.

That third inference is worth naming precisely, because it is the one producers usually miss. The episode has not merely flattened a strong claim and a weak one. It has presented a fact and a value as though they belonged to the same category, and then split the airtime between them. A viewer who leaves thinking "the room weighed speed against floor space and didn't reach a conclusion" has been badly served, because the room did not weigh anything—it heard one timed observation and some preferences, arranged as if they were commensurable.

This is not an argument against disagreement. A producer who treats conversation as inherently defective will end up with a lecture. The problem is a specific misleading inference, and it can be located and prevented: the inference that equal treatment reflects equal support. Preventing it requires the host to hold four pieces of context—what the record is, who produced it, what task it covered, and what it never measured—and to be able to say, in one sentence, which of the on-air disagreements is not a factual dispute at all.

Compare an exchange with an inquiry-led sequence

Where does adversarial exchange earn its keep? Where the disagreement is genuinely about priorities. Contest is good at extracting trade-offs. Put the cabinet's defender and the clear-floor advocate opposite each other with real time and a host who presses, and you will get the thing that matters: "Would you give up the clear floor for twenty-five seconds?" The pressure is the point. It forces each position to state what it would sacrifice, and it produces disagreement that is real, live, and about something the audience can weigh.

Note that even here, guests reach for factual proxies. Someone defending the pegboard may argue the time difference is meaningless rather than argue that clear floor is worth more than twenty-five seconds—the second argument is stronger, but the first feels safer to make. The host should be ready to catch the substitution and hand the argument back: "You're not disputing the times. You're saying they don't matter much. Say that."

For a contested factual premise, the sequence inverts. Before preferences, examine the record: describe it, let it be questioned, and establish what can and cannot follow. In this case, four questions do real work. Why one member out of twenty? Does the narrowing gap between the first and second runs—forty seconds apart on the first pass, ten on the second—suggest the member was learning the task rather than the layouts differing? What happens when a second person is working at the same bench? And what about the untested claims: nobody measured drawer noise, nobody measured reach, nobody checked whether the cabinet's castors mark the floor.

The last question is the one that protects the format from its own tidiness. An inquiry-led sequence can smuggle in a verdict just as easily as an adversarial one—not by crowning a winner, but by omitting the counter-claims nobody got round to testing. If a guest mentioned floor damage in passing and the running order never returns to it, the close reads as though the only open question is the preference one. It isn't. An untested claim is not a false claim, and the final account should carry it as unresolved alongside the speed result.

An inquiry-led form has two hard preconditions. The first is substantive preparation: the host cannot ad-lib the boundaries of a record they have not read, and "the evidence is limited" as a vague gesture is worse than silence, because it sounds like a finding. The producer's notes should carry the numbers, the sample, and the specific gaps. The second is disclosure before recording. Tell contributors in advance that the bounded version of the factual claim is the one on the table. A correction that lands as a surprise is an ambush, and the audience will side with whoever looks ambushed, whatever the numbers say.

Then there is the quieter duty: representing positions as the people actually hold them. A guest whose position is "the floor should stay clear" must not be edited into a guest who disputes the times. Those are different claims held by different reasoning, and the cut can convert one into the other without anyone lying. The same discipline runs the other way: do not let an inquiry format produce a satisfying two-part episode—facts then values—if a contributor raised a factual point that was never addressed.

Give the host a usable correction role

The host needs a taxonomy they can apply in real time, because the three kinds of statement require three different responses.

An unsupported assertion is a claim presented as established that the material cannot carry. The response is a restatement at the level the record supports, with the gap named: "That's four runs by one person on one morning." Resist the temptation to convert scope into a statistic. "One of about twenty members" invites the phrase "five per cent of users," which is not a finding; it's a number invented by division.

A legitimate uncertainty is a claim the material cannot settle in either direction. This one is not a correction. Mark it and say what would settle it—a second round of runs with several members, a crowded room, a measuring tape at the top row.

A value judgement should not be corrected at all. The host's job is to make the trade-off visible and, where it matters, identify who bears the cost: the twenty-five seconds fall on whoever fetches tools most often, and the floor goes to whoever needs somewhere to stand.

There is a fourth case that catches people out: a checkable factual claim that nobody has checked. It is not unsupported, it is unresearched. It goes to a referral, not to the host's judgement.

Where corrections land matters as much as what they say. Put the frame in the introduction, once, before anything implies a different one. Return to it at the pivot, at the moment a guest generalises from the record—"that was one morning; go on"—and state it in the close. Three correction beats in a twenty-minute episode is usually enough. A host who qualifies every sentence will smother the preference discussion that the second half of the episode exists to hold, and the audience will read the caution as evasion.

Finally, define the routes. Before recording, anything that could be heard as a verdict on a contested factual question should go to whoever owns editorial standards for the commissioning body. After recording, the commissioner should have chosen its correction path in advance: an on-page note, a correction at the top of the next episode, or a re-cut. Australian broadcast editorial guidance on impartiality, in the version available for this outline, treats fairness to contributors and balance as distinct obligations and allows the weight of the evidence to bear on how much airtime a position receives. That is a useful reference point, and it is scoped to its own jurisdiction and institution. It is not a US rule and does not settle what any particular commissioner will accept.

A running order that keeps its frame

The table below is the corrected structure. Two columns: what happens, and what the segment is for. The correction points are marked.

What separates the two is where the factual frame sits. In the failure version, it sits inside the disagreement, where it can be argued with, softened and split. In the corrected version, it sits in front of the disagreement, stated once, referred to twice, and left intact—so the contest in the middle of the episode is about the thing the room actually has to decide.

Frequently asked questions

What does a symmetrical debate running order assert before anyone speaks?

Two chairs, two guests, equal clocks implies that two positions were worth assembling and comparably worth assembling. Sometimes that is true; when it is not, the shape keeps asserting it for the whole episode. The answer is not to abandon adversarial conversation but to decide before casting which part of the material is actually in dispute and let that choose the structure.

In the woodworking-room example, what does the observation record actually support?

The record supports one bounded statement: under those conditions, with that member, on that task, in that order, the cabinet layout produced shorter elapsed times. It does not support the general claim that the cabinet is faster. Floor space is a value judgement, drawer noise is half-factual and half-preference, and reach is testable but was not measured.

When is an adversarial exchange better than an inquiry-led sequence?

Adversarial exchange earns its keep where the disagreement is genuinely about priorities; contest is good at extracting trade-offs and forcing each position to state what it would sacrifice. For a contested factual premise, the sequence inverts: examine the record, describe it, let it be questioned, and establish what can and cannot follow. An inquiry-led form needs substantive preparation and disclosure before recording.

How should a host respond to different kinds of claims in real time?

An unsupported assertion should be restated at the level the record supports, with the gap named. A legitimate uncertainty should be marked, and the host should say what would settle it. A value judgement should not be corrected; the host makes the trade-off visible and identifies who bears the cost. A checkable factual claim nobody has checked goes to a referral, not the host's judgement.

What caveat is given about Australian broadcast editorial guidance?

The guidance discussed is a reference point scoped to its own jurisdiction and institution. It treats fairness to contributors and balance as distinct obligations and allows the weight of the evidence to bear on how much airtime a position receives. It is not a US rule and does not settle what any particular commissioner will accept.

Further reading

• Use Factual Research to Change the Series Question, Not Fill the Deck — Use this to move from a claim inventory to editorial decisions, especially when findings should change the question rather than decorate the deck.
• Write a TV Running Order That Connects Content to the Broadcast Clock — After fixing the factual frame, use this to put each segment's job, duration and handoff against the broadcast clock.
• Pitch an Investigative Series That Might Disprove Its Central Allegation — Read this when a factual claim rests on the existence of an accusation; it traces citation chains and maps what would weaken or leave the claim unresolved.