Postmortems
Decisions that cost something — time, a team's patience, a project — and what changed because of them. Each one is written on the work it came out of; this is all of them in one place.
-
Nine months, and no approval
Nine months of design work, a team of three, and a proposal that was not taken forward. What it cost was the assumption that a good design and a real problem would be enough on their own. The part I would do differently is the sequencing: I built the full system design and then went looking for the decision, when the decision was the thing to go looking for first.
From ADU URIC Portal
-
A month before I called it
A month of work went in before I accepted that nothing was cohering into a game, and the honest cost was the weeks between knowing and saying so. Between 2020 and 2022 I built a lot of prototypes that went nowhere, and the thing that period actually taught me was to recognise it sooner. This is the entry where I did, and still not quickly enough.
-
I had made myself the camera bottleneck
Every camera tweak an artist or designer wanted came through me, because configuring a zone meant editing components and knowing which fields mattered. On a three-person team with four months, that is an expensive place to put a queue, and the cost was not my time but everyone else waiting on it. Building the editor tool ended the queue. Being the only person who can change something is not seniority, it is a single point of failure.
-
Build times were eating the team
Every change had to survive a build before anyone could judge it, so the slow builds were not costing me time, they were costing three people theirs, and the cost came out of how often we were willing to try something. I stopped treating build time as an inconvenience and started treating it as a feature of the project, and profiled and cut it like one. What changed was not the frame rate; it was how cheap it became to be wrong.