How hard is it to make a puzzle game for LD?

Posted by
April 27th, 2013 10:31 pm

At first look, making a puzzle game isn’t hard. Basically…

1.) Write a bunch of simple rules that work for one level.
2.) Write 20 more levels that use the same rules (no more coding necessary!)
3.) ????
4.) Profit!

The problem is that trying to create levels that are difficult (but not impossible) is… well.. difficult.

So far, the tactic has been:

1.) Place a bunch of stuff randomly on the screen.
2.) Try to (myself) solve the puzzle I just randomly created. If it takes me a few tries before I finally get it: awesome!
3.) If I solve it right away, add some stuff to where I went to try to make it less obvious.
4.) Continue this method of tweaking things until I break the level and make it impossible to finish.
5.) Start over from scratch with a different random level.

There has to be a better way (maybe there is some higher-level math that can use graph theory to create levels that have many possible routes, but very few “winning” routes). The worst part about this ad-hoc method is that it’s very possible I’ve made a level that is impossible to beat, and I WOULDN’T KNOW IT until it beats me many times. Also, I might have already thrown away a bunch of levels that I thought were impossible, but might have been really good candidates for some of the final levels.

Part of me wants to build an AI that brute forces the level and gives me an idea of difficulty/possibility of them. If this wasn’t for a 48-hour contest, I’d consider it. For now, I’ll just keep scrapping away.

So yeah, in theory, a puzzle is pretty straightforward. On the other hand, the amount of time it will take to create levels for a puzzle game should NOT be underestimated.

[cache: storing page]