Rip out a system and nothing changes, it was never part of the game

Third in the series, and the subject is system design. The first two were on the money models side, this is the part where you roll your sleeves up.

The sentence I hear most is this, “we have five systems running at once, of course I know system design.” But the question is not how many systems you built, it is whether they talk to each other. Lining five systems up side by side is a list more than it is a design decision.

We can split it into three categories. Mobile F2P, PC F2P and premium. On the money side mobile F2P and PC F2P run the same machine, I covered that in the previous post. On the design side all three separate very clearly.

The most basic building block of game design is setting systems up, overseeing how they connect to each other and drawing the frame of the game. Mobile F2P, on the other hand, can carry systems that work independently of each other, which is why you can split a mobile F2P game in two, gameplay and systems. On mobile you can fit whatever system you want on top, regardless of what the gameplay is. And the person who grows up there usually comes up short at building systems, because they never had to build one that interlocks with another.

Designing things independently always feels easier anyway. Every system gets its own currency, its own loop and its own screen, and when one breaks you repair it without touching the others. But under normal conditions the systems are supposed to be related to each other inside the context of the game. The test is simple, if you rip a system out and nothing else in the game changes, that system never became part of the game.

In Dark Souls, for example, souls are both money and experience points, and on top of that they stay where you died. A single resource ties three systems to the same decision, and the question of whether to attack one more enemy or turn back and spend becomes an economy question, a progression question and a risk question at the same time. Take the economy away and the death penalty stops meaning anything, take the death penalty away and there is no economy decision left.

In mobile F2P the battle pass does not need to talk to the merge board, and the ad reward does not need to talk to meta progression. You can pull one out, plug another one in, and the game stays the same game. The rewards handed out are usually supplemental to the game and exist to save the player time in it.

What gets done is clicking systems together like lego. Building a game with that assembly logic was tried on the premium side too, and it did not take.

Marvel’s Avengers is the clearest example. It came out of Crystal Dynamics, the studio that made Tomb Raider. Loot, gear, a battle pass and a cosmetics shop were each bolted onto the game separately and none of them touched the others. If you ripped them all out you were left with a barebones working action game, and that is the real problem. It sold 3 million copies but wrote Square Enix a $63M loss, its CCU on Steam fell from 28k to 3.3k within a month, and in 2023 the game was pulled from stores completely.

It is not the only case. Anthem came out in 2019 and its servers shut down in January this year. Suicide Squad was built on melee in the first pitch, was turned into gunplay on the way to live service, its Steam CCU stopped at 13k and Warner Bros wrote a $200M loss. I am not even going into the monstrosity called Concord.

There is the reverse too. In 2012 no publisher would fund Warframe, and one Korean publisher said they would fail because Western developers do not know F2P. They had one thing in their hands, movement and combat. The open world came four and a half years later, space combat six and a half years later, and both of them rode on top of that core. Last month’s Steam CCU was 157k, which is 13 years after release.

War Thunder is even clearer. When it opened in 2012 all it had was planes, around 200 planes and a flight model. Tanks came in 2014, helicopters in 2018, and the navy later still. Whole new vehicle classes were bolted onto the game and all of them took, because the damage and physics model underneath could carry all of them. It broke its Steam all time record in April this year as well, 125k.

What happened when they touched that core is a separate lesson. In May 2023 they shipped a rank based economy update and pushed repair and ammunition costs up. Recent reviews on Steam turned overwhelmingly negative and most of the people writing them had over 2,000 hours in the game. Gaijin rolled back most of the changes within three days and published an apology four days after that.

One of the reasons for this is the length of development cycles. In mobile F2P the cycle is much shorter, so pre-production is almost nonexistent. You start from a one sentence idea and a one page document, and the rest gets filled in later. It carries on emergent and systems get dressed on top of it, and you build the thing by borrowing designs that are already proven to work in somebody else’s game.

In premium, the place where you start to build the game is pre-production itself. You have to break your head over the design and set the systems up there, you have to decide what you are putting into the game at the very beginning.

And if you do not think about it, it sits there awkwardly. Systems whose reason for being there is not clear either never get used or hand the player a clear negative in their experience. Because premium games are built more on giving an experience, even the smallest immersion break can be enough for the player to throw the game away. In mobile F2P there is not much of an aim like immersion. The aim there is to give dopamine shots in short bursts, and to keep resting the player while doing it. Some games even have softlocks so that the player’s session does not go past 15-20 minutes, because what is being aimed at is the player coming back 2 hours later and building a habit by spreading the 2-3 hours a day across sessions. The more the player turns the game into a routine, the more attached they get, and after that sunk cost makes it harder to quit. This is a playbook taken from MMORPGs like WoW.

Mobile F2P runs more on sessions, and the cognitive load of picking the game up and putting it down does not matter much. You spend 18 minutes to play 15 minutes, you pick the phone up, open the game and you are right where you left off. Not much carries over, and when you close the game you do not think about it much either. Spending gets much easier because of this, because the time invested in the game in one sitting is very low, but since it gets opened 5-6 times a day it feels like part of your day. And since everything in these games is time gated, the feeling of progress arrives and you do not begrudge the money you hand over in the moment. Downloads mostly come from 15 second ads too, you run into the game rather than going looking for it.

On the premium side the game is built entirely on systems and on those systems talking to moment to moment gameplay. The player’s investment is much larger, they hand over money upfront and even the decision to buy the game is an investment of time. If you are going to play for 1 hour it takes about 2 hours out of you, going somewhere and sitting down, turning the devices on, getting into the game, the breaks, and the critique you run in your head after you close it. So the game sits much further forward in the player’s mind. That is what pushes them to think about the game during the day, and that is where IP value comes from too. The player gets attached to the world and the characters of the game.

If you are working in a solved genre like mobile F2P, you have to borrow the frame and paint your own canvas inside it. Stepping outside the frame in the wrong places can ruin the whole picture, so going with A/B instead of hunches matters more there. The data on mobile is not always right either, but product people are always in a data is king headspace. If you are trying something new, you have to learn to take only the fundamental data and throw the rest away.

I have designed in both. In mobile F2P we put out close to 250 prototypes in two years, took 9 of them to soft launch and only one of them could turn into a global launch. In most of those prototypes the work we did was not finding a new core, it was arranging known working pieces in a different order and on top of different gameplay mechanics. Everything I have done on PC had to start from 0. Every feature added on top without thinking about the context of the game dragged the dynamics of the game down.

So write out the list of systems in the game you have and cross them off one by one. If crossing one off changes nothing on any other screen, in any other decision and in any other number, that system is bolted next to the game, not into it. Ask the same question when you are hiring, can the person in front of you tell you which two systems they connected to each other rather than how many systems they built.

In mobile F2P good design is separating the systems cleanly, in premium it is interlocking them. PC F2P has to do both at once, and that is the genuinely hard part.

If your systems work independently of each other, you have not made a mistake, you have just not done any design.