I love the idea of emergence, especially when it applies to engineered systems.
Imagine a flock of miniature UAVs. These are programmed with the usual rules about maintaining a safe distance from neighbours whilst flying at the same speed and direction.
Today’s invention is to take this one step farther and to have individual drones fly so as to take up positions relative to each other eg “fly at 30 degrees 1-wingspan to the North, South, East or West of your nearest neighbour.”
During a long flight, UAVs could cycle through a large number of combinations of such simple geometrial variants, thus creating a huge number of possible shapes eg line astern, cruciform, 8 of diamonds, S-shape -both regular and perhaps irregular.
The critical thing would be that all the vehicles would be in constant communication, so that they would be broadcasting their current rules and other, resulting conditions (such as fuel economy).
This would allow the drones to try any simple behaviours which seemed to be working well for others. Just like a realtime genetic algorithm, this would attune the flock to varying flying conditions (eg when fuel is running low and heading into a storm, flying efficiently and less closely would be an experimental finding that was quickly and widely adopted).