Tuesday, May 11, 2010

Using affinity estimating to choose the Sessions for the Lean Agile Scrum Conference in Zurich

The LAS Conference Organizing Committee met last night to select the sessions for the LAS Conference in September. We had some 26 proposals from 19 speakers from Europe, North America and Asia. Everyone is under time pressure. How do we get through the task of sifting through all the proposals and agreeing on a program quickly and effectively?

While co-teaching a CSM Course with Peter Hundermark, I saw a new estimating technique (well, new for me ;-) 'affinity estimating' as an alternative to planning poker. I thought this would be useful for selecting the talks and tutorials, so we decided to give it a try.

  1. Everybody read all the submissions (46 pages!)
  2. For each Submission, I created a card with the title of the submission. I used yellow cards for talks, blue cards for tutorials and green cards for submissions that could go either way. (Next time, I would also put on the page number of the submission in the printout)
  3. I created column headers (white) Great, Good, Maybe and Questionable.
First we discussed our criteria for a good presentation, decided to rate the talks on their intrinsic quality first, and then handle in a second step the evaluation criteria necessary to create an attractive, balanced program. We decided we wanted 10 selections (for the program) and some alternates (just in case!). Then we agreed to try out the Affinity Estimating to rate the submissions.

Affinity estimating is pretty simple: for each story size, one column (e.g. 1, 2, 3, 5, 8, 13, 20, Too Big). To estimate a story, put its card down in a column, and anybody can slide it around to another column. If it stops moving, you have an estimate. If not, you need to discuss the story to come to a consensus. It can be very useful for (preliminary) estimates of large backlogs, because it is very fast and requires little discussion.

So we set up our columns, put down 6 cards, and moved them around until they stabilized. If they didn't stabilize, we put them in a 'to be discussed' column to discuss at the end. When all 6 cards were placed, we put down 6 more, et cetera until all cards were placed.

At this point, we put all the 'maybe' and 'questionable' cards aside. We were left with 14 cards.

We discussed the criteria for creating the program. There were a number of points: the mix of management, technical, team and corporate topics as well as Swiss and International speakers. What questions do we think will be important to the participants. We also wanted to ensure a balance of the various consulting companies. It wouldn't do to have one or two companies dominate the proceedings.

Next we set up the 10 Slots (4 Tutorials and 3 x 2 Tracks ) and used the same process to put stories to tracks, looking to create the right balance. Again, moving cards around on the table was an easy way to identify which topics complimented each other, which were overlapping, etc.

Eventually we were happy and the cards stopped moving around. We wrote down the results and also took a picture of the final assignments.

That left four cards left over. We decided that these would be our alternates (and we think we have enough good material for an evening event in October or November).

That was it. Four was a good number of participants in the meeting. We managed to stay in our timebox of 2 hours without really thinking about it.

And looking at the results, I think we have a good balance of interesting talks for the Conference this fall (which I think is really a statement about the speakers and their submissions!)

P.S. What happens next for the speakers? I will contact them later this week, so we can start to prepare the program.