Now that the contest is over I'd like to get some feedback on the randomized maze idea with an eye towards turning it into a full fledged map. I'm also fishing to see if there is any interest in adding bonus challenges to the existing rmaze1, plotting more solutions to the second half, and putting a BTS control room at the start that allows you to select specific mazes.
First off, a bit about the map.
original post
http://forums.thinking.withportals.com/dload.php?action=file&file_id=207
Or Hober's map pack:
http://forums.thinking.withportals.com/dload.php?action=file&file_id=209Random elements can be added to a map generically by using a logic_case entity and sending it the target input "pickrandom."
This random maze was made using a number of chambers that had multiple connections. Various connections were blocked off using entities that could be killed/toggled/opened/etc. Each entity was assigned a different name (I like to use a simple coordinate system), and just those entities required to navigate the maze and create a few loops and dead ends were opened (I chose func_brush entities and set their target inputs to "kill" since they weren't going to be reused.)
Although there is probably a way to generate truly random solutions, I went with the easier method of plotting by hand a number of different mazes and programming them into a series of logic_relays that the logic_case can choose from. A logic_case can pick between 16 choices, but you can choose from more with multiple logic_case entities.
The most difficult part of the process was creating the architecture of the maze itself. Hedge style mazes are fairly easy, but portal mazes have to be designed carefully to prevent the player from thinking outside the box, so to speak. The line of sight nature of the portal gun makes it hard to let the player know where the ending is or even view the map without using lots of transparent materials, and those can quickly lead to high poly counts. You also have to keep the map simple enough for the player to navigate and for the designer to plot solutions. I made a test version that worked in three dimensions, but found it almost impossible to orient myself (I'm still working on it), and for the 2D release I read a number of complaints from players who became frustrated because they got lost.
Incidentally, the secret of navigating the second part of the maze is to scout out the solution in the easier first part before making the attempt. That's why I think a time challenge might be interesting to play. A portal challenge for this map seems dull, and steps challenges in mostly-metal maps just annoy me on general principle.So, where to go from here.
First, I read a general dislike for this maze-only style of map. Definitely add some other gameplay elements. I have a hazy idea for a maze made of portal cleansers that you have to carry a cube through, for instance. We'll see how the alpha looks.
Second, more visibility. Maybe simpler architecture, maybe smaller maps, but the player has to be able to easily figure out where they are and where they are going. An earlier version had light strips, but they weren't much help; maybe little street signs. 
Third, less repetition. Easily done with a little time and forethought.
)