From Citizendium, the Citizens' Compendium
John R. Boyd (January 23, 1927 – March 9, 1997) was an influential strategist, whose theories of the interaction between information and action have been widely used in both military and business. He retired as a United States Air Force colonel, but had a great deal of influence on the United States Marine Corps as well as the Department of Defense. The Marine tie-in was through Alfred M. Gray, Jr., eventually Commandant of the Marine Corps, but, at the time, head of doctrinal development.
|“||"Though an airman, Boyd was not an airpower theorist and in fact was little concerned about how to design an air campaign or a strictly air war. Boyd’s concern was the nature of human conflict, the strategy and "grand tactics" writ large, not particular service approaches per se. He jumped from the tactical experience of air-to-air combat and proceeded to think about how to conduct different types of wars."||”|
Boyd's key concept was that of the decision cycle or OODA Loop, the process by which an entity (either an individual or an organization) reacts to an event. According to this idea, the key to victory is to be able to create situations wherein one can make appropriate decisions more quickly than one's opponent. The approach is entirely consistent with information-theoretic ideas of closed loop control systems. It is a basic element of modern command and control.
Boyd theorized that multilevel contexts, such as the tactical, operational, and strategic levels of war, can be modeled with a hierarchy of OODA loops (see below). He also argued that fast OODA loops require a highly decentralized chain of command using mission-type orders, or directive control, rather than method-driven orders in order to harness the mental capacity and creative abilities of individual commanders at each level. He argued that such a structure would create a flexible "organic whole" that would be quicker to adapt to rapidly changing situations. He noted, however, that any such highly decentralized organization would necessitate a high degree of mutual trust and a common outlook that came from prior shared experiences. Headquarters needs to know that the troops are perfectly capable of forming a good plan for taking a specific objective, and the troops need to know that Headquarters does not direct them to achieve certain objectives without good reason. 
The OODA model of decision and action, originally for air-to-air fighter combat, has four phases. This discussion adds interaction with intelligence and command.
- Observe: become aware of a threat or opportunity
- Orient: put the observation into the context of other information; form one's perspective and situational awareness
- Decide: make the best possible action plan that can be carried out in a timely manner
- Act: carry out the decision.
After the action, the actor observes again, to see the effects of the action. If the cycle works properly, the actor has initiative, and can orient, decide, and act even faster in the second and subsequent iterations of the Boyd loop.
Eventually, if the Boyd process works as intended, the actor will "get inside the opponent's loop". When the actor's Boyd cycle dominates the opponent's, the actor is acting repeatedly, based on reasoned choices, while the opponent is still trying to understand what is happening.
Applications to intelligence
While Boyd treated his cycle as self-contained, it is reasonable to extend it to meeting the intelligence cycle. Observation can be an output of the collection phase, while orientation is an output of analysis.
Eventually, actions taken, and their results, affect the senior commanders. From the commanders, rather than isolated intelligence managers, come guidelines on the preferred decisions and actions.
Relationship to military swarming
When faced with swarming attacks, there is a need, at the least, to observe at multiple simultaneous points. If the defender can observe and decide to counter a synchronizing point, a single action can be adequate. If the swarm truly is distributed, multiple OODA loops must run concurrently to counter it.