rules and exceptions

Over the weekend, I came to learn about a drinking game called Cheers to the Governor. Like any other drinking game, the complexity builds up over time.

The game starts with each person counting from 1 to 21. At 21, everyone drinks and the person who said 21 gets to create a rule to mess up with the counting. For example, swap out 5 and 7 so now the counting goes 1 2 3 4 7 6 5 8 9 10 ... 21. Whoever gets it wrong has to drink, but every time it hits 21 a new rule gets added.

The rules aren’t restricted to anything else. Perfectly normal for someone to add a rule such as, “say Taylor instead of 4” or “stay silent on 20”.

When people start messing up after 5+ rules, that’s when it gets interesting.

exception debt#

This is a term which I stole from Will Larson’s blog post: Work the policy, not the exceptions . It is much easier to manage a change when it’s global – imagine if the rules apply to all numbers, i.e. everything starts with Taylor (Taylor 1, Taylor 2, Taylor 3, ...) or all numbers should be subtracted with 3 (-2 -1 0 ...).

This is a great example on explaining why having too many exceptions is not optimal, unless it’s for the fun in the game.

accrued rate of knowledge#

People who started playing from the beginning are learning maximum one rule on every turn. I happened to join in the middle of a game and it was a huge challenge to keep up with the rules.

This became a moment of realization for me that we sometimes think an activity is trivial, only because we get to learn it bit by bit. The trivialness doesn’t transfer over to someone who just started on it. To paraphrase Joe Beda: Complexity is a subjective matter – builders tend to judge their product to be simpler than people learning about the product .

One example I can think of is ownership of an aged technology product. Maybe you have a 10 year old computer or 20 year old car. That product probably runs fine functionally. But you, as the owner, have accrued quirks of your product over time. Perhaps your car can’t shift from park mode to drive without wiggling the gearshift on neutral, or your computer can’t recognize the letter “C” on the keyboard without first hitting spacebar.

This might have significant overlap with tribal knowledge or things you know but not in the manual book.

numbers are not orders#

We reached a point where people were playing the game without actually thinking about the real numbers anymore. We were left with 2 numbers without custom rules. Anything else has some quirk to it, e.g. changing the direction of counting, non-verbal act, saying unrelated phrase.

It’s becoming clear that we need to focus on the order, not the numbers themselves.

To me, this felt like learning group theory in abstract algebra, such that a valid group can consist of the numbers 1 4 2 3, in that order. Meaning that in this group:

  • The second number of the group is 4.
  • 1 + 1 = 4
  • 3 - 4 = 1

(I’m skipping a lot of abstract algebra concept for the sake of demonstration / I have also forgotten a lot of the specifics since Cryptography class in college / Consult your local mathematician for accuracy)

making rules

This observation goes into my collection of mental model in creating rules. Not a comprehensive list, yet I believe they provide a decent guideline to start with.

  • favor general concepts over specific details
    • if it has too many exceptions, maybe it’s not the right level of granularity
  • absolutely not the new person’s fault to find a rule harder to understand than you do
    • alternatively, how long would it take for someone new to master the rule as good as you do? aim to keep that number low
  • double check the subjects
    • the ability to separate the concept of number 5 and 5th count changed the complexity of Cheers to the Governor – put clear cuts on overloaded concepts whenever possible

P.S. I’m more fun in parties than this blog post (or at least I strive to be ๐Ÿ˜ฌ)