Affinity mapping is, at its heart, quite simple. It's the act of placing similar things near each other and separating those that don't. The idea is to start seeing patterns and themes in your qualitative data.
In practice, it's a bit messier than that. Even if you're doing in on 4' x 8' boards. It is invaluable however, in dealing with the enormous amount of data that an average project will churn up if the research team is dedicated to writing down everything that is said or done in the research sessions.
These sessions are fantastic for creating personas, user scenarios, making sense of a set of usability tests, or in anything else where it is critical that the team sees patterns emerge from data.
It also gets everyone working together. Isn't that the whole point?