A poster session sets aside a room and an hour in which people stand beside a printed board describing a piece of work, and everybody else walks around talking to them. It arrived from academic conferences, where it has been standard for decades, and technical conferences keep rediscovering it because it solves a problem the talk format cannot.
What it is for
Volume and conversation. A programme has six or eight talk slots and receives eighty proposals, which means almost everything good is declined for lack of room. A poster session adds twenty or thirty places in a single hour without touching the schedule.
The second thing it does is harder to replicate. A poster is a conversation rather than a broadcast: the person who stops is interested, they ask about their own situation, and the answer is specific to them. That is the exchange a talk cannot produce, and it is why people who have presented both often prefer the poster.
What a poster actually contains
Far less than people put on one. The reliable structure is a question at the top, one figure in the middle large enough to read from two metres, and a short answer at the bottom. Everything else is conversation.
The poster is the prompt. The presenter is the talk.
The commonest failure is a paper printed at A0: dense text nobody standing up will read, and a presenter who then reads it aloud. The test is whether a passer-by can tell what the work is about from two metres away in four seconds, because that is the decision they are making.
Running one at a technical conference
Three practical requirements, all easy to get wrong. Boards or wall space, which the venue may not have and which is expensive to hire. An hour with nothing programmed against it, or the room stays empty. And food or drink in the same room, because people circulate when they are holding something.
The slot that works is immediately before or during a reception, for exactly that reason. The slot that fails is a parallel session opposite talks, which asks people to choose against a programme they have already planned around.
Why technical conferences under-use it
Partly unfamiliarity: outside academia most attendees have never seen one, and a format nobody recognises gets low submissions in its first year. Partly cost, since boards and printing are real money for a small event. And partly status, because a poster reads as a consolation prize where a programme treats it as what happens to declined talks.
That last one is self-inflicted and fixable. An event that solicits posters in the call for speakers as a first-class option, rather than offering them to people it declined, gets a different and better set of them. How the call frames its formats is on the call for speakers.
What it is confused with
A sponsor table. Superficially similar and entirely different: one is a commercial arrangement, the other is part of the programme. Mixing them in the same room damages both, and it is examined on what sponsors get.
A demo session. A running thing on a laptop rather than a printed board. Closer in spirit than the name suggests, and increasingly the version technical events actually run.
A lightning talk. The other way to fit many people into a small amount of time, and the one most technical events reach for first. See lightning talks.
The rest is in the glossary.
Questions
What is a poster session?
An hour in which people stand beside a printed board describing their work while everybody else walks round and talks to them. It comes from academic conferences and lets twenty or thirty pieces of work be presented in the time two talks would take.
What should a poster contain?
A question at the top, one figure large enough to read from two metres, and a short answer at the bottom. The test is whether a passer-by can tell what the work is about in four seconds, because that is the decision they are making.
When should a poster session be scheduled?
With nothing programmed against it, and ideally alongside a reception or food, because people circulate when they are holding something. Running it opposite talks empties the room and makes the format look like it failed.
Why do technical conferences rarely run them?
Unfamiliarity outside academia, the real cost of boards and printing, and status: where a programme offers posters to people whose talks were declined, they read as a consolation prize and attract submissions to match.
Is a poster worse than a talk?
It is a different bargain. A talk reaches everybody once; a poster reaches a few people properly and answers their actual questions. Presenters who have done both frequently prefer the poster for exactly that reason.