I asked my daughter to borrow me her UNO cards to play Collapsi… and I think it was a really good idea to add a touch of color and fun to the game! :)

The picture shows the 6x6 version (luckily the standard UNO set has 8 cards for each number), but it works fine with the 4x4 version, too. The arrow cards are used as Jokers and players can choose which color to start with. Here we used a key and a dice for player tokens.

However, thanks for this little and cute game, I am playing it a lot since I discovered it!

PS:I have some ideas for a computer engine to play against (I am a software developer), I want to make it really strong, I will keep you updated! ;)

This is pretty cool! Can’t wait to see what you discover with the computer opponent. 

Hi, thanks for the original game. Here is a link to an updated version in JavaScript, with a solo mode added. I looked at the rules and the video, and the 3x3 version seemed simple to me, so I made it with the classic 4x4 layout.

With 26 cards, it is more complicated, but it can be solved, see the image with the game logo.

 https://sentinel65.github.io/Collapsi_AI_BASE64_en.html

So, after asking AI whether and how to solve a solo mode task in Collapsi with 16 cards and classic settings, I was told that it is theoretically possible, but that it doesn't happen very often. So my successful attempt with 16 cards was probably just a coincidence, because I haven't been able to do it since. So I switched to the original solo game variant with 9 cards, but to make it a little more difficult, I changed their values to 1, 1, 2, 2, 3, 3, 4, and red and black jokers. The solo game has been updated and statistics on successfully solved puzzles in solo mode have been added.

Solo mode (difficulty)

The program offers three levels of difficulty to choose from easy, medium and hard. In difficult mode, it is randomly generated from three sets of cards.

['1', '1', '1', '2', '2', '2', '3', 'R', 'B'] - easy

['1', '1', '2', '2', '3', '3', '4', 'R', 'B'] - medium

['1', '2', '3', '3', '3', '3', '4', 'R', 'B'] - hard 1

['1', '2', '3', '3', '4', '4', '4', 'R', 'B'] - hard 2

['2', '2', '3', '3', '4', '4', '4', 'R', 'B'] - hard 3

https://sentinel65.github.io/Collapsi_AI_BASE64_en.html

Wow, really good implementation, congratulations! :)

I see that you implemented minimax for engine play, did you also add alpha-beta pruning to speed up calculations? In that case, that's also what I wanted to do by myself and you already made up a really good web interface, so my effort is not needed :D

I also saw that you wrote detailed information about the various components making up the heuristic, really nice, I had almost the same ideas for them.

Thank you very much. The truth is, I am not a programmer or a web developer. I am 60 years old, I really enjoy board games, and my older son (40), who programs in Java, introduced me to this game. I am always excited when I see someone come up with a simple game that does not require a lot of materials or money. I have to give credit to Mark S. Ball here. I've already done one project for myself with the help of free chatGPT, so it helped me a lot here too. But it's not easy to express in text what I actually want and how it should look and work. So from that point of view, it was three weeks of quite demanding work and arguing with AI. The initial versions were graphically simple with a 4 x 4 square grid and numbers inside. My son pointed me to images in BASE64, so I moved a little further. AI is a great thing in this regard, but you still have to learn how to use it. JavaScript has its limitations, but my goal was to create something for myself that I could play on my phone alone or with my girlfriend anywhere, regardless of whether I was connected to the internet or not.

This is amazing! I don’t know how I missed all of these messages. Keep up the great work!