One of the craziest parts about working on Creatures of Aether was how we landed on the scoring system.

A Thread

Triple Triad is the game that inspired Creatures of Aether and it uses a 3x3 Board and the score counts the cards in player hands plus the board to determine the winner.

You can tie 5-5 in Triple Triad.

When I made Elementimals in 2011, I didn't want ties and only used cards on the board to count the score.

The issue with this change was the player who goes first also goes last so they have a huge advantage. This was fine since Elementimals was just against the CPU.

In Creatures of Aether, I knew I wanted to go up to a bigger board of 4x4. A lot of Creatures' design is based around the number 4 (elements) and a 4x4 board would give us more room for cool abilities.

I was excited because an even number of tiles meant that both players would take the same number of turns. But it would also mean we would have ties like Triple Triad had.

I was okay with that. I decided that on a tie, we would go to a 3x3 overtime match.

The 3x3 match like in Elementimals is inherently skewed toward one player.

But that's good that's what I wanted.

I knew even with a 4x4, the player going 2nd would have an advantage because the game is about flipping cards and the 2nd player gets the final chance to flip.

So in my head, we would use the 3x3 overtime to balance out the win rates. We would track the win data and hopefully flipping the player order in overtime would get us close to a 50:50 win rate.

Well we tracked the win rates and we found the difference between the win rates to be bigger than I expected and the draw rate to be lower. On the 4x4 board, the game data ended up around:

42% chance player 1 wins.
50% chance player 2 wins.
8% chance of a draw.

But this was actually amazing for us. The draw rate was exactly the difference between the two win rates.

Early beta players already disliked the overtime since it extended matches and felt unwinnable as the disadvantaged player.

So all we had to do was eliminate ties and give them to Player 1 to get close to 50:50.

How did we do that? At the start of the match we flip a coin. The winner of the coin flip goes first and gets 1 extra point for the rest of the game.

This allowed us to eliminate overtime because games could no longer end in ties. But we did it in a way that doesn't heavily skew toward one player like in Elementimals.

Our win rate for player 1 now floats around 48%-52% depending on the meta but has always remained close to 50:50.

We really lucked into this change because I went into the game assuming we would have ties and need a way to deal with them. But now we don't and you know why :D