I suppose now is a good time to add context to the first question. In my time as a member of Agile teams, we have had two different kinds of retrospectives. The first was where the team got together in the same room, complained about everything which went wrong in the last sprint, praised everything that went right during the same sprint, and created concrete action lists in order to prevent the wrongs from being more wrong and the rights from being less right. Over time, we created measurable improvements in our team’s processes (i.e. velocity went up, estimates became more realistic, fewer issues over petty things).
The other sort of retrospective was when the organizer (never the scrum-master) would make games out of the retrospective, similar to what @simon_tomes linked to in the Retromat. The retrospectives were more fun, but in the end, no action was taken and our processes did not improve.
The cause of the non-improvement was not the “fun” retrospective plan, but rather that in making the retrospective more interesting, the team lost perspective of why we were having the retrospective in the first place. That is, the three questions from Corinna Baldauf, specifically “What will we do differently from now on?”
Which leads me to my context-less advice:
What you do with the cubes is less important in a retrospective than what you do after the cubes are put away.
And my opinion-with-some-context:
The 3-cubes per first two questions is a good idea (providing your team can keep it short, which is why I prefer post-its. The post-its are short by nature), giving more time to discuss the results. The third question, “what will we do differently” can be answered outside of the game.
And what I would do:
It being a new team, I would focus on the team members and use the traditional method of post-its, plusses, minuses and try to focus on the discussion more than the game. Interesting retrospectives can happen later, if there is time or interest.