Today we’re gonna talk about the OODA Loop, a fascinating decision algorithm created by a guy named John Boyd. The OODA Loop is a military tool, developed by Boyd. Boyd was a really fascinating guy. I recommend reading his biography called, “Boyd.” It’s very instructive to look at his career (and the career leaps he made) as a continual process of turning edges into nodes. But today, we’ll talk about one of those huge leaps that Boyd made – the creation of the OODA loop.
The Four Stages of the OODA Loop
The OODA Loop has four stages. The first stage is Observe, in which you’re taking in information from the environment. The second stage is Orient, in which you’re using your mental models and experiences to understand what you see. The third stage is Decide, in which you’re using that analysis to decide what to do. And the fourth stage is Act , in which you act upon that decision, This is how it becomes a loop: once you act, you’re getting feedback from your environment, which will cause you to step back and Observe and you go through the whole thing again.
The OODA Loop as a Decision Making Algorithm
Just using this as a framework for making decisions, when you’re going through and you’re making a big decision for a company, or for an organization, or for your life. I think organizing various tactical techniques like Perceptual Positions (Observe), Wardley Mapping (Orient), Metaphors of Movement (Decide), and Intentionality(Action), into this framework can give you a clear step by step process for taking iterative action. However, the real power of the OODA loop comes when it’s viewed in the context of an adversary.
Getting Inside An Opponents OODA Loop
John Boyd was a military guy, and the This was made for military decision making. And when you’re making decisions in the military, you always have an adversary. The adversary has their own OODA Loop, you have your OODA Loop, and these OODA Loops are playing against each other. What Boyd said was that the way to defeat an adversary, and this echoes some Sun Tzu’s principles, is you actually want to get inside the enemy’s OODA Loop. As I understand it, there are two basic way’s to do that.
Moving Through the OODA Loop Faster
The easiest way to get inside the enemy’s OODA Loop is just go through the OODA Loop faster than them. You want to go from observation to action just a little bit faster than your adversary, but with the same quality of decision. What this does is eventually get thstuck in the loop. By the time they get to the action stage, their observations will be out of date. This means they can either act on the wrong information, or they have to go back to the observe stage. This is how they get stuck in the loop .The faster you go, with the same quality decisions, the earlier in the loop you can get them stuck.
Fooling Their Orient Phase
The other to get inside their OODA Loop, is try to fool them into orienting in the way you want them to. So you know they’re observing, orienting, deciding and acting and you can present information to them, that will make them act in a way that you want. This requires a keen understanding of your opponents mental models, and you also have to be very careful as you can actually give away your own strategy if you do this wrong.
Levels and Game Theory
The really interesting thing here, when you start talking about getting inside their OODA Loop in those two ways, is that they are trying to do the same thing to you. So they’re trying to get in your OODA Loop and you’re trying to get in their OODA Loop, and it can seem a bit recursive. Do they know that you know that they know that you know? How do you know how many levels deep to go in trying to get inside their OODA loop? This is one of the central questions of game theory.
A good rule of thumb here from this book, “The Art of Strategy,” who looked into a lot of the research into this, is that most people think two, or three levels deep. So if you’re dealing with the average person, what you probably want to do, is try and realize that their OODA Loop is that they’re going to think about themselves, your reaction to them, and maybe one level deeper about how they should change that direction. Ideally, with these people, you should be at level 3 or 4.
The other option is to find an advantegous Nash equilibrium – a move you can make where the best option for them is ALWAYS advantageous for you.
OODA Loops and AI
There’s some really interesting tie-ins here with getting inside other’s OODA loops, andAI research, specifically some of the research the Machine Intelligence Research Institute has been doing into reflective oracles and logical inductors.
How Reflective Oracles Get In Eachother’s OODA Loops
Reflective oracles use computation theory to give a basis for mixed strategies. An example of a mixed strategy is doing one strategy 70% of the time, the other strategy 30% of the time, and this allows a Nash equilibrium where my adversary can’t gain an advantage over me by switching strategies. Reflective oracles were arrived at by trying to consider machines who had access to reasoning about other machines’ OODA loops. This naturally resulted in the machine’s coming up with mixed strategies, and provides a clear basis for you to use mixed strategies as well in order to prevent others who are trying to get into your OODA loop from taking advantage of you.
How Logical Inductors Get Into Eachother’s OODA Loops
The interesting thing about Logical Inductors is that they take into account incomplete logical facts – that is, if the speed at which you go through the OODA loop matters (which Boyd argues it does), how does one use limited reasoning to create a strategy that still can’t be taken advantage of. The logical induction answer is to take all the possible facts, find a fixed point average that tells you what do, and use that. I’m not quite sure what this means from a pratical perspective when working with OODA loops, but what is interesting is that there’s a clear strategy that works. I’d love to hear other’s thoughts on how to translate this into practical tools.
OODA loops are an incredibly powerful decision making algorithm that can incorporate lots of tactical decision tools in a logical way. Even more importantly, by taking into account the speed and Orientation, they’re a very powerful way to reason against an adversary and determine how to get an advantage. Start playing around with the concepts and let me know how it goes!