In case you didn't know, end-of-iteration retrospectives are a practice that have been gaining popularity on XP teams. At the end of every iteration, the team considers how things went in the previous iteration and how to change the process. Sort of a constant tune-up of the process. Like stand-up meetings, this is something that a lot of XP teams do, but it's not actually a core XP practice.
I've been holding retrospectives with my XP teams since my second XP team back in 2000. (We called them "post-mortems" then.) I've tried a number of approaches and I've never been very happy with them. I've tried several ways of conducting the retrospective meeting: brainstorming onto a whiteboard; brainstorming onto cards; timeboxing the retrospective. (Using cards and timeboxes worked best.)
I've also tried a bunch of ways of categorizing retrospective thoughts: "good," "bad," and "ugly;" "works well" and "needs improvement;" and SaMoLo--"same as," "more of", and "less of." (SaMoLo worked best.)
Arlo's approach is completely different and very clever. I'm going to try it next time. Rather than getting everybody around a table and brainstorming issues, he has an "emotions box." (Actually, I'm not sure what he calls it.) Throughout the iteration, whenever someone is feeling very emotional about something, they write it on a card and drop it in the box. Good, bad, whatever--any strong emotion gets recorded on a card and put in the box. As a result, the issue brainstorming that previously happened during the retrospective meeting now takes place continuously. It's a very agile-ish idea: cheaper, more frequent, more effective.
This is a neat idea for several reasons. First, writing emotions down is a good way to defuse tense situations. It's also a great way to make sure people's needs are acknowledged--and to make sure they know their needs are acknowledged. Additionally, it changes the focus of the retrospective meeting. Rather than using the retrospective meeting on the past (what happened), we now focus on the future (making improvements). There's less risk of blaming or denying and more opportunity to find solutions.
I've found retrospectives to be valuable, but they've always been slightly painful and didn't result in as many lasting changes as I hoped for. I really like Arlo's technique and I'm definitely going to use it at my next opportunity. I'll let you know how it goes.
(I should note that there's a community of people out there, like Diana Larsen, who know a lot more about facilitating retrospectives than I do. I'll refer you to them for more.)