It’s known that the efficiency of heat engines is limited. So a portion of energy will be wasted. The same holds true in collective intelligence. A complex system contains several components. The process of gathering the info that is distributed among the system into a coherent knowledge entity is comparable to that process. Inevitably a portion of info will be lost in the process and the aim is to minimize that so that the collective intelligence gathered from the system provides the best possible description of the system.
As an example, A painting might contain several shapes, etc. To produce that painting those elements should be placed alongside each other in a particular fashion. A better example would be a Puzzle that contains several pieces. Pieces are dispersed among a complex system and the task in collective intelligence is to retrieve those pieces and place them together in a coherent entity. The top-down approach works for very simple systems but in hugely complex systems info is so dispersed among the system that the top-down approach (someone who knows every detail of everything) is not doable.
The same ideas can be used in the wave probability. It was said yesterday that protective lines should be waves. But those waves are produced by different basic sine waves as specified by its Fourier transform. Every basic sine captures one characteristic of the scene to be observed (think of it as different layers in a neural network). So the task in probabilistic observation is to define such a wave proj line that minimizes the lost info so that the best possible description of the scene is provided.