Deep Management #3: Aristotle Ruminates on Game Dev, Part I

It’s fun to contemplate how ancient thinkers’ philosophies relate to modern problems. As such, it seems natural to look back on a few of Aristotle’s thoughts, as applied to game development! As we’ll point out, these arguments parallel those inherent to methodologies such as Lean and Agile. This post and the next elaborate two concepts from Aristotle’s lectures in Nicomachean Ethics. Here in this first article, we’ll explore the Aristotelean concept of the ultimate good. This will require a discussion of subordinate goods, which is where we’ll find most of the value. This will provide us with a new lens to analyze value and cut waste (in the Lean sense). In the next article, we’ll explore the golden middle way, which can guide us when managing subordinate goods; we will also analyze the asymmetric extremes that surround the best path (asymmetry motivates a number of interesting critiques as well, such as why we developers leap to crunch so quickly, despite counter-indicating research regarding its effectiveness). As with our previous articles, while the topics discussed here parallel those from our book Deep Management, our intent is not to restate the points made there, but to contextualize arguments in new ways.

Having said that, let’s begin! Aristotle originally desired to establish a foundation for ethics, although we’ll repurpose that. Ethics, he argued, was the pursuit of various goods (by this, we don’t mean objects necessarily, but also honor, health, etc.). Many types of goods exist. Some of these goods we pursue for their own sake — given the norms of Greek society at the time, personal and family honor would have been such goods. As a purer example, most Greek philosophers argued that we should pursue the ideal mental state, which they termed eudaimonia — a state of serene calmness and happiness. This was the only true happiness — all else being corrupt. Such individuals would then naturally perform correct actions. Pyrrho would add ataraxia as a precursor to eudaimonia — where ataraxia represented a state of deep equanimity with respect to life’s outcomes.

Goods that we pursue for their own intrinsic worth, we term ultimate goods.

Other goods we call subordinate goods. We value subordinate goods only because they offer a means to reach either an ultimate good, or another subordinate good that is closer to an ultimate good. Subordinate goods are not primarily valuable in and of themselves; losing sight of this can produce waste.

For example, while people work for money, they shouldn’t do so solely for its acquisition. Money, after all, is an imaginary construct empty of objective worth. We value it as a subordinate good because we can exchange it for other things closer to ultimate goods (which, for some, only means survival). Our ultimate good might be to create an indie game that realizes our full creative capabilities, but to do that, we might need to work another job to accumulate sufficient savings first.

Similarly, most of us exercise not for its own sake, although it does produce endorphins, but because exercise improves our health and strengthens us. Many of us view health as an ultimate good. For some, however, there is a higher ultimate good, such as the desire to complete a marathon, for which they might sacrifice health by driving themselves beyond normal limits. Note, this case shows that some subordinate goods have value even though they exist as means to a greater end for some people. This can trip us up when we are managing resource allocation.

Because we pursue subordinate goods to enable ultimate goods, we must then recognize that the excessive accumulation of subordinate goods undermines our ultimate purpose. After all, to acquire our current desired good, we must spend some other good that we consider to be of lesser value. If we continue to accumulate our current desired good to excess rather than switch over to realizing its proper end, then we won’t maximize our ultimate good. Many people desire to create games, write books, or pursue other endeavors, but hold off doing so until they’ve accumulated “enough” savings. Sometimes that safety net keeps rising until they’ve put off their dreams for so long that they’ve passed their most creative window.

Of course, we can also fail to accumulate sufficient quantities of the subordinate good to enable reaching our true end. If we don’t have enough money to finish off our project, we’ll either be forced to release it early in a sub-par state or forced to shelve it and maybe never get back to it.

Thus, maximizing our ultimate good requires managing subordinate goals. We need sufficient lesser goods to reach the greater goods, without accumulating an excess of lesser goods beyond a sufficient risk buffer. This is the application of a project budget to a directed acyclic graph of subordinate goods. The end result will be scored by the values reached in the ultimate nodes. Note, we don’t know the form of the graph at the start. Part of game development involves discovering this structure as we go.

While there can be multiple ultimate nodes, oftentimes for a project we prefer our team to focus on delivering one key ultimate value. The other ultimate goods usually belong to the studio not the project. It is best to subordinate these extra goals to the project’s ultimate good. Some resources can be diverted to them as long as doing so doesn’t jeopardize the core. For example, fostering research partnerships with local universities often doesn’t benefit the project, but can be important to the studio and the community.

We’ll now consider how the pursuit of subordinate goods can go awry, especially in large organizations.

To start, we’ll consider engineering. Sometimes projects do not connect engineers directly with customer needs, especially in the more technical departments such as infrastructure, build, server, and even the client engineering disciplines. These disciplines tend to negotiate more with intermediaries, who often lack the expertise to efficiently match technical propositions with customer needs. In such situations, the engineering department can generate overdone solutions. Lacking access to the ultimate good of customer value, such teams end up preferring their personal ultimate goods. This usually takes the form of an exciting architecture or bold technology. Rather than build what the customer needs, the team instead produces software for themselves. And why wouldn’t they? They are maximizing the ultimate good that is visible to them.

Some typical examples would be overwrought rendering solutions (art falls preys to this as easily as tech), elaborate netcode for non-competitive games, back-end solutions for single-player games, and overdone release-engineering pipelines. Note, all of these solutions might be appropriate for some customer out there, it’s just that they don’t match our customer.

Consider an overdone release-engineering pipeline. For most studios, their current build and test processes are immature, so we should allocate more resources — the issue is how much more. Many non-technical managers fixate on feature delivery alone, but quality and iteration speed provide value to customer as well. Even so, release-engineering designs can skyrocket to enormous scales, which make sense for some organizations, but not for all. These are the cool proposals, however, that scratch the “ultimate good” itch for some engineers. Redirecting a quarter of engineering staff for two years might not make sense for a project that needs to be in full production in six months. Project managers might push back on such a proposal, but push back how much? What is the proper system? The engineers are best situated to answer that, but without the clarity that arises from understanding the project’s purpose, they’ll fall back to their preference.

To resolve this, engineering teams need to be integrated into the full customer-facing experience. Organizational culture must prioritize pursuit of the project’s ultimate good over personal goods. Engineering leads will then align the customers’ needs with appropriate solutions. The project managers can then trust their staff with allocation. The engineers then do the necessary work, and perhaps a bit more for risk and agility, but no more. That is, incidentally, the essence of being lazy, which is the most important trait of a good engineer, although that is its own article. Laziness is also the heart of the Lean and Agile methodologies.

Now, we’ll consider internal politics as another example of subordinate goods that can go awry. In Deep Management, we assert that politics inside a company is always waste, although it can be necessary up to a point (still, we should never normalize it as anything but waste). Politics becomes necessary when we must negotiate between parties that otherwise lack resolution to a conflict. Within our organization, our decision-making culture should provide solutions to all disputes, but that is an ideal we never fully reach. Where the system doesn’t handle the conflict, politics picks up the slack. Trades, favors, threats, and pay-offs all combine to reach stable decisions that allow us to proceed. This isn’t necessarily bad.

As with technical people, however, those with political talent often enjoy and overvalue its application. As these individuals tend to rise to senior positions, they can shape how resources get allocated, either to produce customer value or to service internal power figures. Bias can blind them to the true ultimate goods. They overweight the importance of their own skill set, and focus on their power base (turf, favors, alliances, prestige projects, etc.). Many such figures believe this fosters both the project and studio — a perceived win-win. They are mistaken, however. The subordinate good of a political bank account has been overweighted. Such projects tend to fail in a particular way. They consistently underdeliver milestones, but still get greenlit. Oftentimes, such projects cycle personnel below the level of the politicians, as continuous delays require sacrifices. This in turn weakens the creator culture of the studio.

When appropriate resources get allocated to this subordinate good, however, the deficiencies in our decision-making culture can be smoothed over, allowing for far less stressful production cycles. As with technical staff, those with more political inclinations need to see how their work fits in the graph of goods.

A strong culture here is our primary remedy. Most staff will have personal ultimate goods that conflict with project and studio ultimate goods. To develop a sound culture, we need to instill the ethic that the ultimate good of the project overrides personal concerns. At times, due to unresolvable conflicts, we’ll have to rely on politics, either allowing our staff to explore a personal good more than required by the customer, or otherwise compensating our employee for their focus on project needs. That, though, is clearly waste. It’s another reason why we shouldn’t demand so much of our employees that they can only meet their personal ultimate goods on project and studio time.

In our next post, we’ll discuss a bit more the conundrum of how we find the right amount of work to apply to a subordinate good.

Given all that, what then should be the ultimate good for the project? We’ve already said customer value, but we need to examine this in more detail.

The ultimate good for a project can be a tricky question. After all, for many companies, their explicit purpose is profit. From the point of view of someone like Milton Friedman, that is the ultimate good. Even for such a figure, however, that is a more complicated claim than it appears. A company that neglects social responsibilities and externalizes costs, or that exploits its customers, will in the long term fail to generate profits. Thus, even for someone who takes this hard line, subtle factors remain.

This subject merits a longer discussion than we’ll give it in this article, but we suggest that customer value provides a consistent guidepost that is easy to employ. We know that if our customers are satisfied, they will continue to engage with our brand, and that is a growing asset. They will spend more and suggest our product to their friends. This also reminds us that we must not exploit our customers — at least not excessively — as that generates backlash that harms long-term profits. On the other hand, when profit is the ultimate good, it is easy to justify exploitative actions that generate customer resistance and sometimes legal repercussions. As we suggest in Deep Management, we must be grateful to our customers and use that as a guide. Our hope should be that they continue to trust us and invite us into their lives. 

We also note that it’s perceived value that matters, not “actual” value. Again, a proper analysis of this requires a post of its own. Oftentimes, we proceed with analysis as if we were dealing with real value for simplicity — just remember research such as Kahneman and Tversky’s cumulative prospect theory and the many other cognitive biases that construct perceived value.

For interested readers, we suggest Clayton Christensen’s work for a deeper, entrepreneurial understanding of value. His book Competing Against Luck offers a good entry point for this specific issue. He has also published articles on this topic in the Harvard Business Review.

We wish to reiterate that these issues aren’t idle concerns. We have seen many projects fall prey to these afflictions, when they were easy enough to avoid. The most common are inversions where the subordinate good replaces the ultimate. If we build ten incredible-in-themselves systems that serve the needs of no one, then we’ve actually built ten useless systems.

John Bible
Twitter: @JohnJadeBible
Main Site: http://johnbiblebooks.com
Game Site: http://artofgeekery.com