Two speakers, one talk, three handovers

A co-presented talk is one talk delivered by two people. It occupies one slot, answers to one abstract and is judged as a single submission, which is the first thing most pairs of speakers misunderstand about it. The part that decides whether it works is not the material and not the slides. It is the handover: the two or three moments when one speaker stops and the other starts, each of which costs the room a little of its attention and gives it back only if the pair have rehearsed the join. Almost everything that goes wrong with a two-hander goes wrong there.

How a review committee reads a joint proposal

A committee reading a joint submission is buying one slot and taking on two speakers. That is two sets of travel and accommodation to fund if the event funds any, two tickets, two bios to place in the programme, and two chances that somebody withdraws in the fortnight before the day. The proposal has to be good enough to carry that, and the question the committee asks is short: why does this need two people?

There is one answer that works and several that do not. A proposal that could not honestly be given by either speaker alone justifies itself immediately. A proposal that reads as two colleagues who each wanted to speak, or who divided a subject in half to halve the preparation, is competing against single submissions on the same subject and losing to them. The tell is usually the abstract, which describes two topics joined by the word “and”.

Reviewers also look for signs that the pair have worked together before. A note saying the two have given the talk at a user group answers the question the committee cannot ask directly, which is whether these two people can share a stage.

The handover is the part that fails

Each transition costs somewhere between fifteen and thirty seconds of the room’s attention, and the loss is not the silence. It is the restatement. The incoming speaker, uncertain where the previous section finished, begins by summarising it, and the audience hears the same point twice in a talk that had no time for it once. Do that three times in forty minutes and two minutes of material have gone, along with the sense that the pair know what they are doing.

The repairs are mechanical. Use fewer and larger blocks, so a forty-minute talk has two handovers rather than five. Script the first sentence each incoming speaker says, and make it a continuation rather than a restart. Have the outgoing speaker hand over on a question or a named claim, so the next section has somewhere to begin. And agree what the silent speaker does, because a second person visibly checking notes, adjusting a laptop or quietly correcting a detail draws more attention than the person talking.

Rehearsal costs more than double

A solo speaker runs a talk through three or four times, alone, at whatever hour suits. A pair needs the same number of runs plus the ones spent finding out that the join does not work, and no run counts unless both are present at the same time. In practice the figure is eight or nine, and the constraint is not effort but calendars, which is why paired talks are so often under-rehearsed by the week of the event.

The runs also have to be complete and in order. Rehearsing one half alone is the most tempting shortcut available to a pair with no shared time, and it produces two competent halves and a talk that falls apart between them. A remote pair, or one spread across time zones, should assume the rehearsal problem is the main risk of the format and plan the sessions before the slides exist.

When a pairing is the right form

Two cases justify it consistently. The first is a case study with two genuine perspectives: the engineer who built the system and the operator who ran it at three in the morning, or a supplier and the customer who deployed the thing. The content is the difference between the two accounts, and the moments where they disagree are the reason the room stays awake. Neither speaker can give that talk alone, because neither holds the other half.

The second is a mentor paired with a first-time speaker. The experienced speaker takes the opening and the close, where the risk of losing a room is highest, and the newer speaker takes a defined middle block that is entirely theirs, with their own material and their own questions to answer. This produces a real slot on a real programme rather than a rehearsal in front of a lightning-talk audience, and it works because the division of labour is stated rather than improvised.

What does not justify it is shared authorship. Two people having built something together is a fact about the work, not a reason for both to stand up, and a committee reading “we both worked on this” in a proposal is reading the most common reason a joint talk is declined.

Forms that are not co-speaking

Several formats look like a two-hander and are governed by different rules. A panel has three or more voices and a chair whose job is to keep them apart. An interview format has one subject and one questioner, and is judged on the questions. A demonstration with a driver at the keyboard has one speaker and one pair of hands. Naming these correctly matters, because a committee with room for a panel and none for another two-hander reads the label before the abstract.

There is also the option of a single-named submission with a credited collaborator: one speaker on stage, one acknowledged in the abstract and thanked from the podium. For a pair whose real difficulty is that only one of them wants to be on a stage, that is the honest form, and it costs the event nothing.