[jofer]

Having given quite a few talks and even more posters, my experience is that you get to communicate on a _much_ deeper level with a poster. (Note: I'm not in CS. I'd imagine this varies a lot by field.)

You get absolutely zero feedback and discussion generated from a talk. Maybe a few questions, but no actual discussion. If you're really, really, really lucky, someone might track you down or e-mail you later. For the most part, though, you don't interact with people during a talk.

In contrast, during a poster session, you talk yourself hoarse over three to twelve hours. You wind up in intense arguments and deep discussions. I've never seen that happen during talks.

Personally, I find a 12.5 minute talk is a lot more limiting than a poster, where you'll be talking to people about your work for several hours. Figures are the main focus of any scientific presentation, and a poster gives a great format to walk people through the concepts using figures.

That having been said, at the tech conferences I've been to, posters have always been an afterthought. It's a real shame.

[iak8god]

> you get to communicate on a _much_ deeper level with a poster

I used to hate poster sessions until I realized this. Something poster sessions provide, that other formats like podium talks really fail at, is pretty fine-grained control over the level of engagement. You can spend five seconds glancing at a poster, or a minute or two skimming the whole thing; you can spend a few minutes listening in while the presenter talks a bit about the work to someone else; you can skim and then ask one or two clarifying questions; or you can engage the presenter for as long as you're both interested in discussing.