AI [Alpha 13]

This update changes how AI works.

In the original Ending all directional execution occurred in the following order: up, right, down, left. This allowed puzzles to have finite solutions. But it also added a certain bias - enemies always considered up first and left last.

For Unending I chose to rotate where the AI started its search over 4 turns - starting at up, then starting at right, and so on. This encouraged more natural looking behaviour from the enemies. One can push this behaviour further by leaving breadcrumbs where the player walks (as I have done in other games).

After making some levels for the Player Turner I found that I rarely had 4 turns to shake out desirable behaviour from enemies. I had at most 1 or 2. I wanted the more predictable biased behaviour from Ending where it didn't matter what turn it was. I think this also made for better puzzles as symmetrical levels presented different solutions for what appeared to be identical paths.

Unending no longer rotates its execution order. I believe that this will allow players to develop better instincts about enemy behaviour and lead to better puzzle designs. It also means that the Ending dungeon is more faithful to the original.