The Battle of the Retrospectives is an exercise for practicing various ways for conducting retrospectives. Although it is a good starting point to the diverse world of retrospectives, it could be used also as an instrument to roughly assess and eventually select the retrospective format which best suits participants’ needs and individual preferences.
Might significantly vary depending on the number of retrospective formats covered. Still each part of the exercise should be strictly timeboxed.
Decks of playing cards, sleep/night masks, desks, papers, pens and timer for timeboxing. A selection of retrospectives should be also available (each printed on separate papers). Their formats should allow them to be conducted in less than 15 minutes (e.g. the Rocket retrospective, the Start-stop-continue retrospective, the Starfish retrospective, etc.).
Split people into teams of more than or equal to 5 people. Explain the rules of the exercise and make sure that everybody understands them correctly.
I. Set the stage
Pick up 20 random playing cards. Remove these cards from the original decks and distribute the resulted decks to the teams. Ask each team to assign the following roles to its members: one Yes/No role, one Blinded role, one Connector role and N (or unlimited) Dealer roles. These roles are described as follows:
- Yes/No role – Might just say Yes or No and should only interact with the Blinded role.
- Blinded role – Should be blinded (by wearing the sleep/night mask) and shouldn’t be allowed to speak. Should interact with the Yes/No and the Connector role only. However in order to switch interaction between the Yes/No and the Connector role, the Blinded role has to turn 180 degrees. Moreover only the Blinded role is allowed to put cards on the desk.
- Connector role – Shouldn’t be allowed to speak and should interact with the Blinded and the Dealer role only.
- Dealer role – Should initially hold the deck of playing cards and shouldn’t be allowed to speak. As more members of the team are sharing this role, the deck of the playing cards should be equally distributed between them. The Dealer role should only interact with the Connector role.
While teams are assigning roles, take one original deck of playing cards (consisting of all the 52 playing cards) and shuffle it well. This should result in a random arrangement of the playing cards. Write it down on a note and make multiple copies of it (as you’d have to provide one to all the teams in the next phase). This activity might be done offline (before the exercise itself) in order to save some time. Finally ask teams to take their starting position near their own desk. An example of a possible starting position is given below (where the rectangle represents the desk). The staring position (as well as the assignments of roles) could be changed by the teams after the first round (if they’d like so).
II. Get to work
Call the Connectors and give them copies of the random arrangement of the playing cards. Ask them to go back to their teams and take their starting position. Make sure that the random arrangement is not shared with the other team members during the entire phase (only the Connectors should know it). Once all the teams are ready, ask them to arrange their cards on the desk following the order on the note. They should take special care to arrange the cards in a straight line (as they might not gain points otherwise). Also if the next card from the desired order is missing (was removed from the original deck) – they should leave an empty space on the deck (equal to one card size). Start the timer and close the phase after 10 minutes.
III. Calculate the results
Only the cards arranged on the desk and the ones removed in the very beginning (the 20 random cards) are taken into consideration when calculating the results. The table given below shows the points for each type of card.
Gaining or loosing points depends on:
- Whether the card is placed correctly on the desk. This includes the following conditions: (1) the card follows the desired order; and (2) it is on the straight line. If these conditions are fulfilled – the team wins card’s corresponding points.
- Whether an empty space is presented on the desk and is properly reflecting a missing card from the desired order. If this is the case – the team gains card’s corresponding points.
Ask teams to calculate their points using the above rules.
Example: The figure given below provides a simplified example of the game and how the results are calculated. It assumes that (1) the original deck consist of 12 playing cards only – the 2-s, 3-s and 4-s (for simplicity); (2) the desired order is 3D (3 Diamonds), 3S (3 Spades), 4C (4 Clubs), 4H (4 Hearts), 3C, 2H, 2D, 4S, 2C, 3H, 4D and 2S; and (3) the playing cards removed from the original deck are 4C, 3C and 4S. The perfect arrangement represents the arrangement which matches the desired order (and therefore brings the maximum possible points) while the actual arrangement has some deficiencies. The results calculated below refer to the actual arrangement of the playing cards.
Provide teams with a new retrospective format. Give them 5 minutes to read it carefully and raise any questions or concerns they might have. Then ask the teams to retrospect following the given format. They have exactly 15 minutes to do so. They are allowed to change their role assignments, starting position, ways of communicating, strategies, etc. When teams are ready you might start with the next round.
Round Two to N
Repeat round one until all the retrospective formats are covered. If there’s enough time and you’d like to practice more – you might try having two or more subsequent rounds for each retrospective format.
Calculate the results by summing up the points gained by each team from all rounds and announce the winners.
Let the teams define what retrospectives are and briefly discuss how they could benefit software development. Ask them to rank the retrospective formats based on the: (1) attractiveness of their procedure (e.g. simplicity and learnability, ease of use, informality, anonymity and confidentiality, teamwork, etc.); and (2) their usefulness (in terms of how they improve the perceived and actual performance of the team). Give each team some time to present and argument their ranking. If there’s enough time you could ask participants to map the exercise with software development (or Scrum in particular). This includes exercise’s assignment (e.g. what do the cards and their desired order represent, why there are missing cards, etc.), roles and interactions (e.g. who plays the Yes/No role, why the Yes/No role could only interact with the Blinded role, etc.), procedures (e.g. what do rounds and their phases represent, etc.) and assessment (e.g. what do points represent, why different types of cards are associated with different number of points, etc.). Ask teams to identify some common pitfalls demonstrated by the exercise (e.g. the Yes/No role saying only Yes or No, the Blinded role turning 180 degrees in order to interact with the Connector role, etc.). Then, again in the context of retrospectives, discuss when should the teams have retrospectives; how often should they be; who should be invited (e.g. what if only some of the roles in the exercise were allowed to retrospect); what should be retrospected (e.g. should they cover emotions and feelings, various shareholders’ perspectives, etc.); how should they know when the retrospectives are successful; etc.