Deep Management #1: Decision Making and the Seven Principles

Why do we make strange decisions? And how can we make better ones? In this blog, we'll explore how to critique our decision-making culture in games. This entry starts with seven key principles.

Why do we make strange decisions? And how can we make better ones?

In this blog, we seek to understand how to produce a superior decision-making culture. The art of deconstructing decisions enables that. This requires breaking down the logic that generated our choices. Post-mortems often evaluate decisions from an after-the-fact perspective. To refine our process, we must evaluate past choices in the context in which they existed at the time. While we do require an outside view to see the failures of our entrenched practices, to critique ourselves, we must do so from an accurate representation of the past. Only then can we solidify the rules that underpin how our culture engenders success.

Our greatest threat here is hindsight bias. In later posts, we will consider this further, but we should begin from a state of skepticism and with a desire to seek deeper truths. If we naively analyze our history, our evaluation can be flawed. When we consider a past choice already knowing the outcome, we weight factors that support the realized result, and suppress alternative possibilities. Success, in fact, can arise from bad decisions and vice versa, so we must be cautious. We must situate ourselves in the mindset of people at the time to correctly consider how effectively our policies functioned. Moreover, we must operate from the perspective of individual, unique agents, not from the view of a standardized corporate consciousness. We make decisions based on perceived local conditions, incentives, and punishments, all of which generates our collective behavior. We desire for agent behaviors to coalesce into actions that solve overarching goals, but we do not function as a centralized intellect.

In the end, though, empirical results matter more than hypothetical ones. While we must consider counter-factual possibilities to make our analysis resilient, actual events — especially when repeated — are hard to dispute.

As a case example of the right perspective to hold, consider this interview in Destructoid: Rod Fergusson, a former Epic producer, now at The Coalition, noted that he would have canceled Fortnite in its earlier Save the World phase; he also would not have greenlit the Battle Royale mode. Given what we know now, that seems absurd! Was it though? He knew his own evaluation criteria, then actively disregarded his knowledge of the outcome, and forthrightly asserted that this would have been his call both times. Given this, we should then consider this call in situ. While that would have been his call, was it the correct call for Epic’s culture at the time, and which decision would have been the right call regardless?  We need to articulate the principles that drove the choice. For Rod, the risks and prospects as he would have perceived them at the time resulted in his response. It’s possible, for example, that the real reason the project continued was a fear of sunk cost; we’re not asserting this is the case, but it does happen. That would have been an error. Or perhaps the risk/reward profile at Epic differs from Rod’s, leading to a different calculation. If Fortnite was greenlit for the wrong reasons, we should correct our criteria, despite its success. Perhaps Epic finds that the decision obeyed their rules, and that Rod is the outlier, which is also a reasonable outcome. It’s important to extract what our stance was and what it should be.

Bear in mind our key dilemma: usually the decision to buy a lottery ticket is irrational, but it is hard to tell that to the current winner!

Given examples such as this, we can postulate some key principles that our culture must embrace to produce a self-improving decision-making process. First, we require humility, as we must be able to publicly admit that we would have decided in a way that most people would consider a gross error. We must care more about improvement than our projected image. Second, our culture must be founded on trust: even a humble person would be a fool to give ammunition to others who will use it as a weapon against us for their own advancement. And third, we must be bold. It is easy to go along with success and the popular explanation. Doing so rarely offers the great rewards that innovative cultures demand.

Even if we possess these three character traits, though, that is insufficient.

Analysis is only as sound as the facts that compose it. Individuals often construct narratives around events — whether innocently due to our human cognitive faults, or as a strategy to advance themselves. Hindsight is itself a narrative bias, where we formulate the plot to match the outcome and theme, obscuring the actual events.

Thus, as much as possible, we must rely on measurement. Measurement-driven decision-making can face hazards, as the data is rarely total and often biased away from some critical areas, but this cannot deter us. In later posts, we’ll address these threats so that we can minimize the distortions they produce. Ultimately, there is no replacement for hard facts. Moreover, we need a decision history that provides context as part of our measurement system. It should be possible for people other than the decision makers to use this history to understand the context. Even for ourselves, this is necessary, as the many systemic flaws of our memory are well documented. Frequently, however, other people will be judging the quality of these decisions, many of whom might not even have been on the project at the time.

Both data and a decision history are useless, however, if they’re hidden. We must also be transparent. Most cultures require some limits to transparency, but we will be better off to the degree that we push it as far as we can. Frequently, people make decisions based on poor understandings of the world in which they operate. If they possessed sound data, their decisions would be better. If their choices and the justifications are public knowledge, it is also easier for peers to recognize the error at its inception. If they can supply the missing information, the decision can be corrected before it harms us.

As a common example, many projects maintain internal delusional release dates. In later posts, we will discuss why management does this, but for this thread the issue is the impact of these optimistic deadlines. Because everyone realizes the official target dates are fraudulent, everyone creates their own estimate, and then makes decisions given those assumptions. These assumed release dates, of course, must never be stated aloud. This leads some people to make unsustainable short-term decisions, believing that project release is imminent, while in the adjacent office, people make decisions based on distant targets. We know that this might sound crazy, but we’ve known teams where the disparity has been as large as two years!

Analysis, however, is only as good as its depth. Many post-mortems are one-off hour-long affairs that primarily offer employees the opportunity to vent. We all know the deja vu of post-mortems that list the same problems we saw in the previous release. To some, this indicates that poor management — a leadership team that never addresses issues — often, though, the problem is that only the symptoms were addressed. If we do not address the cause of the symptoms, the rot will worsen and emerge in new ways.

Thus, our next core principle is root-cause analysis. We must commit to getting to the root cause of problems when we note a symptom. The Five Whys is a good Lean process for this, where for each symptom, we first locate its immediate causes, and then we treat each cause as a symptom itself! We strive to push five levels deep before we quit. Usually, by the third, we’ve dug up enough, but we should seek to the deepest root causes we can reach.

The six principles we’ve discussed so far are well and good, but they cannot aid us if we don’t know where we are going!

Thus, our final principle must be the ultimate purpose of our particular project and company.

For some, the goal is pure profit. As we’ll argue in later posts, that leads to short-term thinking. Long-term sustainable profit arises from a deep understanding and relationship with our customers. That is how we produce a valuable brand. As such, we suggest that our preeminent principle must be an obsession with delivering customer value. We must still be profitable! We aren’t a charity. But if we place profit above all, we will often sacrifice value, which is the root cause of many duds on the market.

An obsession with customer value links the work of our developers directly to the customers’ problems. It slashes out intermediaries. A parallel idea led to the rise of deep learning as well. Deep neural networks were hard to train — we didn’t know how to efficiently improve the weights to increase our scores. Hinton, Osindaro, and Teh broke this impasse with their paper “A Fast Learning Algorithm for Deep Belief Nets.” This triggered the wave that now underpins one pillar of modern machine learning. It is our view that many of the problems of modern management arise from equivalent quandaries. It was because of this parallel that I titled my book on this subject Deep Management. This is also the core mentality behind Agile and Lean approaches.

These seven principles provide a sound foundation, but all organizations possess unique visions. Each of us needs to evaluate what supplementary principles might be required. A company that focuses on medical robotics, for example, will have different requirements than one that does game development.

In all cases, our principles are not absolute and are subject to criticism. Even the seven foundational principles. Our purpose is to continually improve.

The purpose of this blog is to address these and similar concerns. We do not intend to reiterate our book Deep Management, although that is the foundation. We seek instead to continue the journey with you our readers. We wish to consider new angles, and to critique arguments that we previously put forth. It is our hope that you will join us, so that we can all make better choices tomorrow!

John Bible
Twitter: @JohnJadeBible
Main Site:
Game Site:

Latest Jobs


Playa Vista, Los Angeles, CA, USA
Senior Level Designer (Zombies)

PlayStation Studios Creative Arts

Petaling Jaya, Selangor, Malaysia
Lead Concept Artist

Digital Extremes

Lead AI Programmer
More Jobs   


Explore the
Advertise with
Follow us

Game Developer Job Board

Game Developer


Explore the

Game Developer Job Board

Browse open positions across the game industry or recruit new talent for your studio

Advertise with

Game Developer

Engage game professionals and drive sales using an array of Game Developer media solutions to meet your objectives.

Learn More
Follow us


Follow us @gamedevdotcom to stay up-to-date with the latest news & insider information about events & more