Author Topic: Maze research from LIACS  (Read 10239 times)


  • Newcomer
  • Posts: 1
  • Karma: +0/-0
    • View Profile
Maze research from LIACS
« on: July 31, 2015, 03:30:12 PM »
Hello fellow gamers and developers,

A roguelike can only be as good as the maze algorithm it is built on, so I thought I would start with some proper research into what makes mazes difficult. I am most interested to find out whether the hypothesis is true that any and all loops will always make a maze easier, i.e. is it possible to place loops long enough to truly confuse the player, or is it always better to just have a dead-end so that the player must go back?

As a computer science student at Leiden Institute of Advanced Computer Science, I am researching this topic and have made a website where people can complete mazes generated by depth-first and hunt-and-kill algorithms with varying amounts of loops. I would be most grateful if you would finish some mazes - in return, all my data (except the visitor ip-addresses) will be made available along with my conclusions on the same website, so you can finally get some statistics on how people solve mazes. Please, don't be shy to post this around your friendgroup - the more data, the better!