Wednesday, January 31, 2018

Going on Ahead

Games that can say something about testing, then. Christmas brought a new one to our house, for my kids, from my Aunty Wendy. It's called Hedbanz and it's a game you probably already know as 20 Questions or Who Am I?

Essentially, each player takes a card and, without looking at its face, places it on a band around their head so that the other players can see it. The player then has a limited time to ask the others yes/no questions about their card, in an attempt to work out what it is. A typical game might go like this:

  • Player: Am I alive?
  • Players: no.
  • Player: Am I manufactured?
  • Players: no.
  • Player: Do you find me in the home?
  • Players: yes.

And so on. A skill in the game is asking questions that together partition the space of all things in different ways to narrow down to a specific item.

I like playing the game with my kids because it allows me to see their logic at work, and allows me to show them different strategies for getting to an answer. For our annual Testing Can Be Fun event at Linguamatics and then later for the participants at CEWT #5, I made a few modifications to the rules.

  • You are playing in teams.
  • Your team's aim is to identify your animal.
  • You may only ask ten yes/no questions about it.
  • And you have to specify them in advance.

What's happening here? I'm cutting the search space down to a specific category to make the task tractable (Hedbanz has three categories: animals, food and things) but I'm also imposing two significant constraints: only a small number of questions, and no way to choose later questions based on the answers to the earlier ones. Oh yes, and a limited amount of time to come up with the questions.

Once the ten questions are written, they can be asked. The teams were paired up to ask each other their questions, and record the answers. Hedbanz is more flexible than some variants of the game in that it permits a few different answers: "yes", "no", "rarely", "sometimes", and "usually".

Once the answers are collected, the teams are given a little time to think about what animal they might be, and then take it in turns to say:

  • What animal they guessed.
  • What strategy they adopted for their questions, and why.
  • What the best question they heard was, any why.

On both occasions we followed that with a short discussion about how this task could be like testing, what kinds of testing it could be like, how it wasn't like testing, why, what kinds of strategies were motivated by the constraints, what kinds of pain were caused by having to enumerate all the questions up front, what that made us think about testing approaches which enumerate all questions up front, and so on.

The whole thing can fit into 30 minutes.
Image: Neil Younger (via Twitter)

No comments:

Post a Comment