- Thoughts
Projects need a higher floor, not just a higher ceiling
Quality that is maintained even in busy and exhausted moments is more important than moments of brilliance. Raising the floor of a project through automation, reviews, and documentation.
When working on a project, there are moments when things go exceptionally well. A skilled developer might solve a difficult problem in one go, or a problem that remained unsolved for days might be fixed by a single person's idea. In those moments, the team's level seems incredibly high.
However, I do not believe that these peaks determine the level of a project.
Rather, what is important is the floor. How things operate during the busiest periods, whether someone else can take over when the person in charge is away, whether a certain level of quality is produced even when someone with less experience does the work, and how far a mistake can escalate once it occurs.
The true level of a project is revealed in these moments.
Depending on exceptional people lowers the floor
In development, depending on exceptional people can sometimes seem like the fastest way. When a difficult problem arises, the skilled person solves it; when something urgent comes up, the experienced person handles it. In the short term, it is efficient.
But when this pattern repeats, a strange structure is created within the project. "Only that person knows this." "Don't touch this part." "If a problem occurs, ask that person." The skilled person continues to create peaks for the project, but the moment that person is gone, the level of the project drops precipitously.
That is why I believe a good project is less about how exceptional results a brilliant person can produce, and more about how consistently decent results can be achieved even in ordinary circumstances.
The tedious things exist to raise the floor
When you think about it, many of the things we find tedious in development actually exist to raise the floor. We perform code reviews. We write test codes. We build and automate CI. We establish coding conventions. We leave documentation. We leave logs and build monitoring systems.
Some might consider these things cumbersome. For a single highly skilled developer, that might actually be the case. A skilled person can understand code without documentation, modify it carefully even if tests are lacking, and avoid mistakes even if the deployment process is manual.
But a project does not always operate in its best state. People can get tired, the person in charge might go on vacation, new people might join, and right before a deadline, mistakes can be made that wouldn't normally happen.
Therefore, a good process must first ensure that the project does not fall below a certain level even when people are not in their best state, rather than trying to make people perfect.
The value of automation lies in reducing variance
I believe this is one of the critical roles of automation. When people think of automation, they usually first think of saving time. Reducing a ten-minute task to one minute, or having a program replace a human task.
Of course, that is important. But I believe the greater value of automation lies in reducing variance. It is about ensuring that a task—one that succeeds when an expert does it but fails when a novice does, one that was done correctly yesterday but could be botched today, or one where results vary slightly from person to person—produces similar results regardless of who executes it.
The point is not to make the best results even better, but to make the worst outcomes less likely.
Build structures rather than demanding perfection from people
The same applies to teams. It is difficult to make everyone a top-tier developer. It is also impossible for everyone to always work in their peak condition.
If so, what we can do is not demand perfection from people, but create a structure that catches mistakes when they occur.
Ensuring that if one person doesn't know, another can; ensuring that even if someone doesn't remember, they can look it up in the documentation; ensuring that even if incorrect code is written, it is discovered through reviews and tests; and ensuring that even if a wrong change is made by mistake, the system blocks it.
As a result, the ability of the most exceptional person on the team might not suddenly increase. Instead, the results of the team's worst moments gradually improve. I believe that is an important signal that a project is growing.
Long-lasting projects do not collapse easily
In sports, a strong team is not the one that shows peak performance in every single game. It is the team that does not collapse easily even on days when their condition is poor. Projects are similar. It is natural for a skilled person to perform well. It is natural to perform well when there is plenty of time. It is natural for things to run well when there are no problems.
However, there are not many projects that can produce results above a certain level even when people change, time is lacking, and unexpected problems arise.
That is why I believe it is important to ask these questions when improving a project.
Rather than "How can we raise our peak level?", first ask, "When is our project at its absolute worst?" and "How can we make that moment slightly less terrible?" Peaks can be created by exceptional people.
But the floor is created together by processes, culture, and systems.
Ultimately, a project’s lasting strength may lie less in how high it can rise than in how hard it is to bring down.
A Trip to Suncheon and Yeosu, Day 3
The last day, ending with some studying and a run.

A Trip to Suncheon and Yeosu, Day 2
The second day of the trip, from a clear morning by the sea to sunset and dusk on Baekyado Island.

A Trip to Suncheon and Yeosu
A trip to Suncheon and Yeosu.


Comments 0
Share your thoughts on this article.
Loading comments…