I Spent Five Years Designing Systems I Thought Were Brilliant

I spent five years inside rooms where the primary evidence of success was a spreadsheet that stopped arguing back. The stamina curve looked elegant. The aggro radii nested cleanly. The animation priorities resolved without obvious conflict. Playtest notes grew quieter. I believed the systems were brilliant because the internal metrics said so.

Then I left those rooms and played the games as a stranger.

The Comfort of Internal Metrics

When you live inside a design document long enough, the document begins to feel like reality. Numbers that balance against other numbers acquire the weight of truth. A resource regenerates at a rate that produces the intended number of encounters per hour. A probability table delivers the intended distribution of rewards. An enemy’s engagement distance creates the intended frequency of contact.

All of these outcomes can be true at the same time that the player experience feels opaque, arbitrary, or strangely effortful. The metrics measure what the system does. They do not automatically measure what the system teaches the player’s body and attention to expect.

What the Spreadsheet Cannot See

The spreadsheet cannot see the moment a player begins to over-correct a mount because the recovery curve was never internalized. It cannot see the quiet decision to open a wiki because a rule was enforced without ever being shown. It cannot see the low-level fatigue that accumulates when every transition severs the thread of attention. Those effects live in the gap between intended output and lived input.

I mistook the absence of loud complaints for the presence of clarity.

The Moment the Gap Became Visible

The clearest break arrived years after I had stopped defending the old work. I was playing a game I had no professional connection to, and I watched myself perform the same compensatory behaviors I had once dismissed in playtest videos. I was managing a stamina system that felt just slightly out of phase with the combat rhythm. I was estimating distances that the enemy did not treat as equivalent. I was waiting through transitions that reset my sense of place.

None of these frictions were catastrophic. Each one was small enough to be endured. Together they produced a low, persistent drag that I recognized too late. I had helped build versions of these systems. I had signed off on their internal elegance. Sitting on the other side of the controller, the elegance was not the most noticeable feature.

The Habit of Defending the Document

Inside the studio, the instinct is to protect the system you can see. The document is legible. The player’s internal model is not. When friction appears in playtests, the first response is often to look for player error or insufficient tutorialization rather than to question whether the underlying variables are communicating what the team believes they are communicating. I did this more times than I care to count.

Distance makes the defensive reflex harder to maintain. Once you are no longer responsible for the document, you are free to notice what the document never contained.

What I Now Treat as Evidence

I no longer begin with the intended behavior. I begin with the sensation a player actually reports and work backward. If a weapon feels heavy before it lands, the startup frames, camera shift, and audio onset are the evidence. If a boss feels oppressive, the closing speed and recovery movement are the evidence. If a day cycle feels productively anxious, the length of the clock and the cost of failure are the evidence.

The old internal metrics still matter. They are necessary. They are no longer sufficient. A system can be numerically balanced and still teach the player the wrong expectations. When that happens, the brilliance was only internal.

The Cost of Leaving

Leaving the industry did not make me wiser in any automatic way. It only removed the requirement that I defend the previous answers. The gray cat still walks across the notes. The noisy coffee shop still supplies strangers navigating systems they do not realize are systems. The quieter café still hosts the actual writing. The work is slower. The standard of proof is higher. I now treat every elegant curve with suspicion until I have felt its consequences from the outside.

Closing the Entry

I spent five years designing systems I thought were brilliant. Some of them probably were. Others were only coherent from the side of the table that held the design document. The difference is not always visible while you still have to ship the feature.

This notebook exists to stay on the player’s side of that difference. The numbers remain useful. They are no longer allowed to have the final word.