Game design works like method acting

Most of the teams I talk to are trying to design a game in a genre they do not play. The genre was picked from data, the competitors were scanned, a table was drawn up, and then three people who have not played 100 hours of that genre in their entire lives sat down and started designing. The game that comes out is right on everything measurable and wrong on everything that cannot be measured but makes the difference.

Game design is a lot like method acting. In the method the actor does not perform the character from the outside, they live that character’s conditions and produce the behaviour from the inside, and a more believable character comes out. It is the same in development, the deeper you are inside the genre you are making and the more niche knowledge you carry in that genre, the better you design. Do not take that as whoever plays a lot knows, though. Just as someone who watches a lot of films does not become a good actor, someone who plays a lot of games does not automatically become a good designer.

The best example is De Niro’s Taxi Driver. Before playing Travis Bickle he got a real New York taxi licence and carried real passengers on 12 hour shifts for a month. The licence expired four months after the film came out. He did not do it to suffer, he did it to learn what the man doing that job knows. How a person sits in the back seat on the night shift, what the driver watches in the mirror. None of it is written in the script.

The opposite is far too common here in Turkey. Look at the data, find the pattern that works, copy it, ship. It is not a wrong method either, it genuinely works on the decisions you can measure, because there every decision you make comes back with a number and that number is what tells you that you were wrong.

But most design decisions cannot be measured. How many minutes a run should take, how long the death screen should hold, which minute the first upgrade should arrive at, what the player should feel the second time they die. You have to make all of them before you ship and the answers are neither in the data nor in the competitor scan. The answer sits only in the intuition of someone who has played that genre enough, because that person knows from experience how the wrong answer feels.

While making Metal Thunder it was not realistic for me to reach an AC-130 crew from Turkey, so I did part of the research from documentaries. But the part that really helped was constantly exposing myself to military content. I played almost every CoD from MW4 onwards from start to finish, I was opening War Thunder on my breaks, I was constantly researching aircraft and weapons. Most of the decisions inside the game came out of staying inside that material, not out of a design document.

While making DriveWave I had started watching the Fast and Furious series again, got fired up and made the game. Around then I set up a sim rig at home again and I was constantly going back and forth between Assetto Corsa and Forza.

Taste develops the same way, and it is the thing every developer needs in the age of AI, because it is the only thing that can separate you from the rest. Have AI do as much research as you like, let it give you pages of PDF, it cannot give you the niche and visceral knowledge that is specific to a genre or to a profession, specific to the places a human has touched. The only way to get it is to spend hundreds of hours in the genre you love, playing the bad examples too. There is no shortcut and it cannot be bought, and that is why that person stays the only one on the team.

In our country this works even harder, because most teams here learned to pick a genre with data. On mobile, picking a genre was a matter of looking at a table and there it was the right job, because you could measure it. When you move to the premium side with the same reflex the genre is still picked off a table, but this time no number comes back to confirm the table. The team designs a genre it does not like for two years and only learns that at launch.

Do not take from this that anyone coming from mobile is doomed to lose, because the clearest example of the opposite is right in front of us. Luca Galante, the developer of Vampire Survivors, spent most of his career at a company making mobile slot machines. The game’s chest opening animations, that showy layer of sound and visuals, and that one more run feeling you get when you die just short of thirty minutes, all of it was learned there. Galante says this about slot machines, the player is really spending money on every press, so there is an incredible sensitivity to detail in the sounds, the animations and the order they come in. The man did not pick the genre off a table, he brought the feeling of the thing he had stood inside for years, and the game won a BAFTA.

Ask yourself one question. How many hours have you played in the genre you are making in the last year, and how many of those hours were bad examples. If the number is low, you are not making your game’s design, the game you are taking as a reference is making it, and you do not know what that game does or why.

Do not take from this that you should only make the genre you love either. The genre you love may not make money and that is a real problem. What I am saying is this, you can pick the genre with data, but once you have picked it you have to live there, otherwise what you picked is not a genre, it is just a label.

Your player works out whether you love that genre in the first ten minutes, and tells you so by hitting Alt-F4.