Game twik Friday

So I’ve got three twiks for you all today that hopefully will see at least a prototype level development sometime in the near future. The theme of the day is: Color! I love colors. The twiks are

  • Sheepdog–use your fingers to guide sheep into the pens of their color while chasing off wolves!
  • Lighthouse–position lighthouses with different colors to mix their lights and fill up colored beacons
  • ColorLyne–basically just Lyne but with colored paths and color changing vertices (it would be able to embed both Lyne and Flow Free, that popular and routine iOS game)

I’m only going to talk about Sheepdog because I think it’s the most interesting and I’ll probably be prototyping it soon (so consider this a pre-proto warning for the beta testers out there reading this). The neat thing about Sheepdog is the whole concept is built around flocking, algorithms for which have existed for a while (e.g. BOIDS). There are a few core mechanics/parameters that I think could make for a very interesting game. They are

  • terrain and placement of units
  • species/types of sheep and their flocking parameters
  • wolves and their speed/intelligence/etc. parameters
  • number and speed of sheepdogs

Difficulty increases with topologically complex terrain, slow/few dogs, fast/intelligent wolves, and hard to understand or mixed flocking dynamics, and conversely the game is easy with simple terrain, single type of sheep with predictable flocking, slow dumb wolves, and fast/numerous sheep. The interesting variable to me is the flocking behavior which is highly parametrizable, and it would be possible to make it fairly complex or even genetic in response to the player’s play style (given some type of level set challenge).

This gives rise to another game entirely, which is something of a get all sheep in the pen as fast as possible, with the sheep having strategies that are genetically selected so as to make it as difficult for the player as possible. Completely custom levels that no one else has likely seen. Seems like a neat idea. Or you could have automated sheepdogs that learn your strategy via a set of parameters in your behavior (i.e. you are training them) and then you’re scored by how well they perform.

What kinds of AI games would you like to play?

Leave a Reply

Your email address will not be published. Required fields are marked *