This paper introduces how to simulate flocking behavior. This is an interesting topic. Some key points...

 1. Clearly, the amount of processing one bird does when flying in a flock is constand. Since flocks can grow in size, and birds don't really have problems with that. This means that optimally, we can simulate flocks in linear time (constant processing time for each bird).

 2. The simulation proposed in their paper takes a quadratic time. One reason behind this is that their implementation is not optimal (bah!). Another reason is that sometimes we need special properties in the flocks to make them more beautiful than real flocks (for example, making the flocks form a certain letter while flying). This is a more-beautiful-than-real-life type of simulation.

 3. The paper introduces the major 3 forces governing each bird in the flock. The 3 major forces are collision avoidance forces (between the flock members), velocity matching, and flock centering.

 4. Detecting collisions with environment obstacles is another problem. The paper describes how we can handle this. One way is to use force fields near the obstacles, but this is bad because objects will fly away even when they are flying parallel to the obstacle. Another solution is to look through the eyes of each bird and make it go for the fastest way around the obstacle.

Overall, that was a nice paper to read, but left more to be desired. I had a feeling that wasn't technical enough!