Reverse Engineering Dot The Game
alan
Every so often, I reverse-engineer puzzle games. I start playing them when I’m stressed. I play them like I’m trying to get an answer to a question. That’s when I get interested in what’s underneath the game. This time, I reverse-engineered Dot The Game by Emmanuel Roux.
Background
This habit probably started when I wanted Optimus Prime as a kid but couldn't have it, so I built an ~interpretation~ of him with paper supplies. When I finally got to play with a real Optimus, I was crushed to find out I had the logic all wrong. (The paper Optimus used the smoke stacks as arms. Very skinny arms.) Paper Optimus was later crushed. I still remember exactly how the real Optimus transforms. I wouldn't have noticed that without trying to build it myself.
It’s a simple game, a 4x4 matrix of sums of a 5x5 matrix of dots. I worked it out on paper and then asked GPT to code it in JavaScript. I had a game up and running in less than an hour. I was stunned at easy this was for GPT. I encourage everyone to try to reverse-engineer games with GPT. You get a sense of how different this design material is.
Here’s a snapshot of the game, built from simple prompts:
Here’s a storyboard of the gameplay:
Aesthetically, the game feels like a 4x4 grid of numbers. But the game is one level down, stored in a 5x5 grid of dots (like a holographic universe). Once I felt I had grasped the game, I followed my curiosity. For example, could you solve the puzzle with shapes instead of numbers?
A 3 cell means a corner shape is required, so the only question was which corner it fit into. All the numbers worked this way, a shape and a number of options.
| Type | Options |
|---|---|
| 0 | 0 |
| 1 | 4 |
| 2 | 6 |
| 3 | 4 |
| 4 | 0 |
I was playing a different imaginary/heuristic game above the surface game, but the logic on the lowest level still defined it.
Little games like this are fun because the solution is set, which lets me focus on the journey to the solution. The numbers reveal something about their neighbors. Rules bind neighbors. Applying a rule reveals more of the board by limiting the options a shape in a cell can take.
An algorithm could solve this almost instantly, but I typically follow a path. I follow the most certain rules, but there are many different possible paths in a given game.
If I think about trying to pick the most optimal path, this starts to resemble a traveling salesman problem.
(cont…)