On the morning of 18 February I gave the tool a game design document, version 3, written over the previous weeks: a roguelike card game based on Scopa, eight acts, a shop, bosses. By evening I had two hundred and fifty files in a repository, an eight-module architecture, thirteen recorded architecture decisions, fifteen identified risks, and a project plan: nineteen sprints, nineteen weeks.
On 24 February the game was on itch.io.
What I believed
That the plan was wrong. Nineteen weeks for a project that took less than one is an error by a factor of twenty. I spent years estimating projects; a gap like that, in my old job, is a fault. My first reaction was to tell myself the tool estimates like a 2019 team, and that everything it announces should be divided.
What it was
The plan was not wrong. It was made for something else.
The nineteen sprints described an order: first the rules engine, then the game loop, then the economy and the shop, then the depth systems, talents, potions, bosses, saving. That order was right. I followed it. The first four sprints were delivered on 18 February, the next seven on the 19th, with two hundred and eight tests and ninety-six percent coverage. Thirty-one commits that day.
What no longer held was the duration attached to each step. The plan assumed writing the rules engine takes a week. It took an hour. But it also assumed you know what order to build in, and that you do not build the shop before you have cards to sell. That stayed true, and that is what the plan gave me: not a calendar, a sequence.
The next two days, the 20th and 21st, were not in the plan. That was balancing: play, find that act 1 is too easy, raise the score thresholds, play again. The plan cannot foresee that, because it depends on what you feel with the game in your hands.
What I take from it
A plan has always had two functions: say in what order, and say how long. The second has just disappeared for the part of the work that consists of writing code. The first is intact, and it is worth even more than before, because when you can build everything in a day, the temptation to build everything at once is strong.
So I still ask for a plan on day one. I read the order, ignore the durations, and keep an empty line at the end for what the plan cannot know: the moment when you play.
The plan was not off by twenty on duration. It answered a question I no longer ask.
Get the journal by email
About one entry a week. What I build, what breaks, what I learn. No promotion, one-click unsubscribe.
You can also follow the RSS feed.