Build tools, but do not let toolitis make tool-building the reward

"Let me set the foundations up properly first, then the game will flow by itself" sounds reasonable, and at first glance it is true. The problem is that building tools is far shorter, simpler and more satisfying than building a game. An editor, a pipeline or a save system works the day it is finished and makes you feel you moved forward every day. With the game, there are stretches where you cannot tell for weeks whether it is fun. That is how building tools turns so easily into the most prestigious way of running away from the game, and it has a name: toolitis.

Not building tools is a mistake of its own. Mark Brown of GMTK puts tools at the top of the list of things that went right in his Mind Over Magnet post-mortem. He could drag the drill, the block and the electromagnet into the scene in Unity and have them snap to the grid, change their values with a slider, and by the end of development building or changing a level had become almost effortless. He says he does not regret any of the time he spent on those tools. I did the same thing on mobile, turning hybrid casual games into templates and taking a month of work down to forty hours, and I am still building tools where they are needed to speed my whole process up.

In both cases the tool was written on top of work that was already being done. So the rule is simple: know who is going to use the tool and how, and make the job you do most often the fastest job you have. Toolitis is the exact opposite. You are writing a tool for problems you have not had yet, in a game you have not made yet, with nobody to use it and no repeated work to speed up, only the feeling that it will be needed later. Now that AI is this good and this powerful, picking the right tool matters more than it did, but the invisible cost of sitting down to build one has gone up too, because the tool you give months to today can turn into a day's work a month from now. AI is moving unbelievably fast. At the end of the day a tool's value is decided by whether it was used in a game that works and sells.

The furthest version of this is writing your own engine. Writing an engine is genuinely fun, you touch every part of a game you would normally never touch and you make everything work exactly the way you want. But all of those upsides hold only for as long as you behave as though budget, schedule and pressure do not exist. It means reinventing not one wheel but dozens of them, and knowing everything from the GPU pipeline to how the operating system hands out memory as well as the team that built the off-the-shelf engine does. A job being enjoyable does not mean it added anything to the game. If you do have all of that knowledge, and the time and money to put the game off until the engine is done, then by all means go for it.

The other half of toolitis is solving the technical problem the player is not looking at. On Fallout, Tim Cain, a programmer and at least two artists spent weeks each, months in total, on the character's foot sliding while walking and catching at the end of the loop. A year later Diablo shipped with feet skating like they were on ice, and Cain says you will not find a single review from that year that mentions it. His own feeling is that in that time other problems Fallout shipped with could have been fixed. A tool costs you at the same spot: if the problem it solves is not a problem the player will run into, the time you spent comes out of somewhere else in the game.

Build the first version of the game without tools, by hand and ugly, and write the tool for a job at the point where you catch yourself doing that job by hand over and over. On the day you write it, build a real level with that tool or take a balance pass through it, because you do not know what a tool you have not used is earning anyone.

A game without tools still gets finished, it gets finished much later and much more expensively because you did everything by hand. A tool always has to be a means and never the goal, and if your goal becomes building tools you will never finish a game.

The best tool is a finished game.