This week was calm at work, which, combined with insomnia, made for a good opportunity to work on a game as a hobby.
I like variants of popular games, for instance I made a chess with fog of war:
Variants are quick to iterate, and build on top of know games, so they are usually appealing to players.
This week I found Probabilistic Tic-Tac-Toe in Hacker News, a Web game:
https://www.csun.io/2024/06/08/probabilistic-tic-tac-toe.html
As in Tic-Tac-Toe (TTT), you win by getting three in a row of your mark (X or O). However, instead of selecting a place to mark, you know the probabilities of each cell resulting in your mark, your opponent's mark, or losing your turn.
Then you throw a die, and know what piece is being mark in that cell.
I must say it was an interesting variant!
In Hacker News comments someone pointed to another variant:
https://www.smbc-comics.com/comic/incomplete
Here each player gets a random mission, like "Win", "Lose" or "Tie".
Sounds interesting, and as I didn't find an online version, I was excited with making one. How hard could it be?
With a bit of help of ChatGPT 4o, I got a quick draft to build on.
As expected, GPT made a nice working standard TTT. I was surprised that it added the missions and victory conditions well enough.
Perhaps not so surprisingly, it got stuck creating the AI. After a few tries of indicating that it was not working right, it simply couldn't make a working one.
I had to dust off my knowledge on 'minimax' and understand what GPT made, to fix a few things.
After a few hours and sharing it with some friends for feedback, it's working:
If you like it or want to talk about variants and game development, throw me a message!
⁂
EOT
---
Send me your comments to
text.eapl.mx.mebiu [at] slmail.me
or