Sympathy for the Devil

When Best Practices are worst practices

Let’s do some role-playing. Let’s call it Game Industry Tycoon. You’re the CEO of a $20 billion game publisher.

You’ve just sat down for a greenlight meeting. One of your top development teams is pitching what could be your next big game. On your left is your head of marketing. On your right, your CFO. Across the table: your head of studios. The dev team stands nervously at the front of the room. Over the next two hours, you’ll see early art, a list of comparable titles, a sizzle reel showing the game’s intended mood, and a glimpse of gameplay — all meant to convince you this $100 million, four-year bet is worth making.

You sit in silence, tasked with deciding whether to greenlight the project, based on a 90-minute pitch.

Here are some of the things going through your mind during the pitch:

• Will the game be fun? This is the most important question...and the one you’re least equipped to answer. The more innovative the game, the harder this question is to answer. But your job is to figure out in 90 minutes what your customers will decide over millions of player-hours. 
• Will it sell? This is a different question but just as important (for many execs much more important). How big is the market? Is this game based on a big franchise? Is its gameplay part of a major trend? Will the idea be exciting to many different customers? You start to look toward the other games to ensure there is a market. You might ask how big the genre is and how likely it is for this game to “get only 10%” of the market. Meantime, you don’t know what other game companies are doing in that genre. 
• How much will it cost to develop, really? You look at required headcount and timeline. You ask your CFO for a second opinion as you look at the costs, and your HR head on the desired hiring plan, as you figure you will have to hire a few hundred more people to support the development.
• Can I trust this team? Can the team – especially the leader – pull this off? What have they done in the past? Do they seem confident? Are they truly excited or is it more of an act? This game could be a passion project with a huge audience, or it could be the team sounds so passionate because they are worried about getting fired. Either way the team is highly motivated to be convincing.
• Will the team leader quit if I say no? Top developers are scarce. If you turn this one down, they may walk, straight into funding from a VC or a rival publisher. And once they’re gone, they’re hard to replace.
• Will I get fired? If you greenlight the game and it flops, you may not survive the postmortem. One failure earns difficult questions. Two becomes a pattern. By then, your board is wondering if you're good at your job.

Those of us who attended these meetings as observers at EA called them “beg for life.” It was gallows humor, but it said something real about how brutal and opaque the process felt for the developers, even 20 years ago.

The pressure doesn’t end after greenlight, it gets worse. The team marches through milestone reviews, where one thing is always clear: the costs are adding up. Your CFO reminds you of it constantly, and for good reason. Cost is easy to measure. But what you likely still don’t know, even deep into development, is whether the game is fun. The world gets built. The bugs get squashed. The art gets done. But nobody really knows if it’s working.

Now ramp up all the stakes and all the pressure when the game is late. You have a choice. You can agree to delay it. You can force your team to launch what they have. You can kill the game. It may be the right thing to give your team more time. But your credibility with the board is shot. They might ask if you can forecast. Or whether you are pushing the team enough? Missed deadlines eat trust. 

Ramp it up one more time if the category gets hot and you have a lot of competition. You started out thinking you were one of the few games in this category and your team had a really good take on the genre. Turns out several other companies did too. 

With so little information and such high stakes, the temptation to keep pouring money in to see how the game comes together is overwhelming. You keep funding the game, pushing your team to deliver, knowing that if you cut it off now (say $50, $100, $150 million in), you are assured of defeat, but if you continue to develop the game you have some hope for victory. Economists call this thinking the Sunk Cost Fallacy, and would tell you it is one of the most insidious traps in all of finance, the investment equivalent of the Sirens seducing Odysseus.

Although the details are always different, some derivation of the scenario above is how AAA games have been approved at big Western game companies for the last 30 years. The playbook has been handed down from a different era, largely unexamined as the cost went up by a factor of 10-100X. They are, in fact, considered “best practices”. As CEO, you will get asked about your greenlight process from time to time, from your board, the analysts that cover you, and the investors who buy your stock. Few will criticise you for having a standard process like the one above. 

So, have some sympathy for the management teams of big companies. Under the current structure it’s a really hard job. 

But not too much sympathy. 

‍INVISIBLE DECISIONS MASQUERADING AS BEST PRACTICES

The point is not that the leaders of the major game companies are dummies. They are not. The problem with the above scenario is it is structurally broken. As all these decisions are going on at the surface, there are implicit mental models that are not being interrogated:

• Assuming a linear development process: Most pitches start by assuming a linear process. “The game we are talking about is (fill in a description of a finished product, a genre, a user experience). Our estimate is that it will take us 4 years and $100 mil.” By assuming you know what the game is prior to starting development, by definition you are not counting on much – if any – iteration. It wouldn’t go over well if you did. Imagine saying to your board: “We want to spend $100 mil but we don’t know what the final product will look like.” ‍
• Big decisions with very limited information. The current system rests on the assumption that it’s possible to make a massive capital allocation decision on highly limited information. Under the current system you literally don’t know if the game is going to be fun, and that is considered OK.‍
• The schedule dominates over finding fun. Capital investment decisions are not just about money, but the timing of the cash flows. So one of the first questions everyone considers is when the product will ship. That is a reasonable question for a capital allocator but it’s not the biggest determinant of success – fun is.‍
• Star system: At $100 million or more of assumed development spend, only the biggest, most senior names should be making this game. This line of thinking has been especially true in recent venture capital investments in games, where a stint running a big team at Riot or Blizzard is assumed to be a good proxy for ability to make a new game. ‍
• Graphics sell games: As pointed out in What is a Game Anyway? many people think cinematic graphics are a requirement to be successful. But if that were true we wouldn’t have Minecraft or LoL, etc. The financial way of saying it is there is now price inelasticity of polygons. More fidelity, same return. Put more simply, are you getting paid for the marginal polygon? ‍
• Front-load spend on graphics. Even worse is the timing of the spend on graphics. Very often, tens of millions of dollars are spent on building world art, character art, lore art….long before anyone knows if the game is fun. Art is easy to show, easy to evaluate and gets the decision makers excited about the game. And asset production always looks like the long pole in development, so if you don’t get 10% of the art assets by 10% of the project schedule, you’re already behind. As a result the game design and interactions get locked down by a design you targeted three years before completion.‍
• The tools are fine. That’s the assumption — but much of game development still relies on outdated pipelines: painting virtual leaves on virtual trees in Photoshop, assembling assets manually, and debugging with a lot of brute force. Core workflows haven’t fundamentally changed since the first Lara Croft. In most industries, this level of stagnation would be unthinkable.‍
• The industrialized development process is fine: A soul-crushing process for developers is just what we have to do. The pipelines don’t need to change much. We just have to work harder to find and retain highly talented developers who are OK being uninspired. 

These are legacy beliefs that made a lot of sense when the numbers were a fraction of what they are now, graphics sold boxes, and all this was manageable. They no longer apply.

Without examining these mental models, the industry is structurally hampered from doing anything original. No sane capital allocator would dive into a creatively risky bet under these incentives. If it fails, you have to explain not only why you burned the money, but did so when there is no obvious market for the game. The structure rewards sameness. 

What's left is to dominate Red Oceans by increasing development spending and marketing, and then hope to make it up by superior monetization. The euphemism that captured this is “fewer bigger bets”. This is a way to signal management and financial discipline while avoiding accountability for actual innovation.

‍WHY THIS IS A TERRIBLE BUSINESS

So why worry about making something original? Let’s start with the art reason: merely copying what everyone else is doing is a crummy way to go through life. We appreciate the great innovations around us, and the creators who have the courage and taste to bring them into the world, but then we’re going to make the same old thing? That’s not a recipe for self-respect.

But it’s also terrible business. Minimizing innovation inevitably leads to landing in a Red Ocean. If you make a game that looks like everyone else’s game, you have – by definition – a commodity. Commodities can be good businesses, but not if you have a non-commodity cost structure. And nothing about the process above has a good cost structure. The process aspires to industrialize game development to drive efficiencies, but it has only homogenized the product. This is the cost structure of building a Ferrari with the output of Detroit. 

This unexamined structure is probably why so many big game companies have had such low ROI for investors. $10,000 put into a basket of the top 4 Western public AAA publishers five years ago would today be worth $12,200 – an anemic Compounded Annual Growth Rate of 4.1%. That same $10,000 would have gotten you over $20,000 (a respectable 15.5% CAGR) if you just bought a NASDAQ index fund. And it’s much worse for game developers: they’ve been laid off by the tens of thousands over the last few years. The layoff pattern is almost as bad as if those developers worked at investment banks, but with none of the financial upside during boom times.

Our industry is not short on talent or ambition. This isn’t a story of failure. It’s a story of legacy assumptions calcified into “best practices”. And those assumptions are killing our creativity and our investment returns. 

The good news? The unspoken assumptions and mental models can be replaced. Every one of them.