Decision-Making: The OODA Loop

This is week three of Molding Moonshots!

I’ve already gotten a few pieces pieces of really great feedback from some early adopters of this newsletter. Interestingly, all are at early-stage startups:

• Mike, a cofounder and COO at reOrbital, had a suggestion about how to make plots from pandas more easily readable.

• Justin, a founder and CEO at Campus Tree, pointed out to me that the comments were disabled. This led to a really interesting conversation about what the practical objectives of this blog are, and the relationship between brand and content in support of those goals.

• Somebody else, at a startup that’s still in stealth mode, made a particular point about last week’s post that I found really insightful. They identified a compelling reason founders in space could be avoiding pre-seed rounds that I had not considered.

Based on what they’ve said, it seems like a good idea to open up comments, starting with this post. I’ll also go back and open comments on the earlier posts.

Having just made that decision, I started thinking about the process of reacting to that feedback. This week, I want to consider a particular approach to decision-making that’s relevant to both entrepreneurs and investors called the “OODA loop”, and point out what people who are great at what they do have in common as it relates to this approach.

---

OODA is an acronym for the four steps in the process: Observe, Orient, Decide, and Act. Developed by John Boyd for operational-level military planning during the Cold War, it has had a huge influence on aerospace engineering; it drove the design of all fourth-generation US fighters. The framework has also been applied to many other decision-making contexts, such as disaster response, litigation, and business. This isn’t some sort of “single-player game”, related to the type of event that game theory studies; rather, this is a way for participants in multi-player games with imperfect information to make the best decision possible. Iterations of the OODA loop can be done as an individual, or as an organization.

Here’s what the OODA loop looks like as a diagram:¹

Importantly, the OODA loop is a closed-loop control system; feedback is essential to the process. What happens in later stages of the loop should impact future instances of this decision-making framework.²

Observe

At the beginning, the goal is to understand the outside environment. In the framework of a feedback loop, which is very much what the OODA loop is, everything we think or do, as entrepreneurs or investors, should be a response to an outside stimulus.

The startups that succeed aren’t founded for their own sake; they’re founded to solve a problem that somebody has, which is a sort of external factor. Something outside the system is the raison d'être for the system. Therefore, the point of departure for an iteration through a business decision-making process should be a frank evaluation of external factors, as well as feedback from prior iterations.

From a very early stage startup’s perspective, observing looks like customer discovery. Later on, it might look like research into competing products or services, looking for possible competitive edges. Beyond entrepreneurship, from a truck driver’s perspective, to observe is to scan the road ahead of the vehicle.

Orient

During this stage of the process, external information is integrated with and contextualized by internal knowledge. Looking solely at external factors is necessary, but not sufficient, to make a great decision. The internal framing of external data that happens in this stage almost has to be unique to the decision-maker, regardless of whether its a group or an individual.

Boyd suggests five key internal factors, but it’s not really clear how essential they are to the orthodox OODA loop.

What does orienting look like? A software developer might look at an error message from a IDE and recall prior instances where they saw that message. A consultant might draw on past engagements as they evaluate client data. A founder might synthesize some combination of experiences nobody else has had that’s going to give their startup’s product an edge.

Decide

Ultimately, the synthesis of internal and external data should enable the user of the loop to make a clear decision. I think it’s really interesting that Boyd frames this as a testable hypothesis.

This creates a link between the OODA loop and the scientific method that’s really curious.³

Framing the decision itself as a testable hypothesis creates a methodological link between business decision-making and the R&D that so often characterizes deep tech startups. Thinking about these two apparently disparate aspects of a startup as utilizing frameworks that aren’t just similar, but have two identical steps (“Hypothesis” and “Test with experiment”) is important because by creating links between them, the line between technical and nontechnical staff weakens. This way of thinking should make it more comfortable for people on both sides of the line to contribute to problems that the other half are solving, which is really important for early stage startups.

It’s worth remembering that while the process is happening, the world does not stop to wait! As a result, by the time somebody’s ready to make a decision, a key observation may be out of date, or orienting perspectives might have shifted. It’s totally acceptable in this framework to say “My information is out of date” and jump back to the Observe step. Similarly, a decision to not make a decision at this point is a valid outcome of this process, and should also lead the decider back to the Observe step.

If time permits, it’s a good idea to record key decisions. Different fields have different standards for how to do this; VCs write investment memos, judges write opinions, etc. The reason this gets done is to create a record not just of the decision, but of the rationale behind it. This makes it possible to clarify the key issues in the decision, have more informed discussions about a decision and its effects, and potentially perform a root cause analysis when something doesn’t work out. That last use is really the most important one, because understanding why things don’t work becomes critical information for the Orient step in future iterations of the OODA loop.

Act

This is the time to test the hypothesis with an experiment. Or, as Nike says, “Just Do It”.

---

I like this approach to decision-making for two reasons:

1. The OODA loop is a closed-loop control scheme. From my observations, it seems relatively rare to get something in business exactly right on the first, or even second, try. A general framework for decision-making should bring the user to a better outcome regardless of it’s their first application of it to a particular problem or their twenty-fifth. The only way I know to do that consistently is to take the outputs from prior iterations and use that as an input to the current iteration.

2. The OODA loop can be applied to qualitative or quantitative decisions. As an engineer, I was trained to used decision-making processes, like certain types of trade studies, that reduce everything to numbers. That’s great when dealing with engineering problems, but I find that it really doesn’t help work through things like moral dilemmas. I’ve seen similar processes for qualitative problems, such as rules for hermeneutics, in the context of religion. There’s not all that many heuristics that I know about that can deliver great solutions to both quantitative and qualitative problems, but this is one.

---

Last week, I met a venture capitalist who said his biggest competitive edge as an investor was how quickly he responds to founders. As an investor, his core competency is supporting founders, and he does it faster than anybody else he knows.

Earlier this week, SpaceX, the most valuable company in the space industry, did a secondary sale that implied a valuation of $180 billion. Their core competency is launching rockets, and they can go through all the OODA loops for that process more quickly than anybody else in the world. They’ve launched about 100 rockets since this time last year, about 10x more than their closest American competitor.

This isn’t limited to the entrepreneurial ecosystem. Football quarterbacks go through this process multiple times every down as they run their plays and react to the dynamic situation on the field; they just may not be conscious of it. Chip Kelley upgraded the Oregon Ducks football team from good to great by applying the concept more explicitly to his team. He shortened his offense’s OODA loop, and kept opposing defenses from going through their OODA loops in the ways they knew how to.

Across applications of the OODA loop, what elite individuals and institutions share is their ability to complete iterations in their core competency very quickly.

1. https://en.wikipedia.org/wiki/File:OODA.Boyd.svg

2. For an introduction to open-loop vs closed loop controls, see this Wikipedia article.

3. By Efbrazil - Own work, CC BY-SA 4.0, https://commons.wikimedia.org/w/index.php?curid=102392470