When I decided to build the J-Marie, the first problem was not epoxy.
It was not marine plywood or fiberglass. It was not the plans. It was not the engine, the fuel tank, the wiring, the cabin, or any of the normal boatbuilding problems people think about when they look at a finished hull.
The first problem was the garage.
I had a finite amount of space. That was it. I could not negotiate with it. I could not wish it bigger. I could not get excited about a boat design and then expect the building to cooperate later because I had fallen in love with the project.
The garage was the constraint.
So before I bought the plans, I studied them. I spent a ridiculous amount of time figuring out the largest boat I could build in that space. Not just whether the finished boat would technically fit inside the garage. That would have been the easy version of the question.
The real question was harder.
Could I build it on a cradle, which would make it taller? Could I work around it? Could I move it catty-corner when I needed to? Could I still store tools, parts, and normal garage junk? Could I still pull the race car in and work on that when I needed to? Could the kids' bikes and the rest of life still happen in the same space?
Because that was the real job.
I was not building a boat in an empty room. I was building a boat inside a working garage that still had to stay a working garage.
That is a different problem.
The biggest constraint was the garage door and the overhang. Removing the garage door would buy me about another 6 inches of vertical space. Side to side, I had enough room. Depth was tighter. The cradle mattered. The door opening mattered. The angle of the boat mattered. Even the lip at the edge of the garage mattered.
There was about a one-inch drop coming out of the garage. I knew that if I placed the dollies correctly, I could use that drop at the exact moment the cabin roof came through the opening. The dolly would drop off the lip, the boat would settle just enough, and I would gain a little bit of clearance right when I needed it.
That was not luck.
That was in the plan from the beginning.
A lot of people thought I was crazy. Friends, neighbors, people who came over and looked at the boat growing in the garage. They saw this 21-foot sportfishing boat filling the space and thought there was no way it was ever coming out of there.
I was never worried.
Not because I am fearless. I am not. I have broken enough parts and paid for enough mistakes to know better than that.
I was confident because I had already done the math.
I had it drawn out. I had it measured. I knew the constraints. I knew what I could change and what I could not. As long as I stayed inside the plan, it would fit.
That last part matters.
Plans do not save you if you ignore them halfway through the build.
When you are building something over four years, little mistakes want to sneak in. A quarter inch here. A quarter inch there. A cut that is close enough. A panel that is just a little off. A bracket moved because it was easier that day. None of that feels like much in the moment.
But a bunch of quarter-inch mistakes can turn into several inches of problem.
That is how drift gets you.
I did make design changes. I wanted more headroom, so I adjusted the plans to get it. But I did not just add height and hope. I went back to the constraint. I recalculated the space. I made sure the change still fit inside the plan before I built it into the boat.
That is the difference between a design change and drift.
A design change respects the constraints.
Drift forgets they exist.
That lesson is not limited to boats.
In automotive work, the constraints get even less forgiving. Now you are not talking about inches. You are talking about thousandths. Bearing clearances. Piston-to-wall. Ring gaps. Oil viscosity. Heat expansion. Parts measured at 70 degrees that are going to live at 190 or 200 degrees under load.
The engine does not care what you meant.
It only cares what clearance you actually built.
Racing adds another layer. Class rules. Engine size. Component limits. Tire size. Gear ratios. Final drive. RPM. Trap speed. Boost. Fuel. All of it has math behind it. If you ignore the spec, you do not just lose performance. You can get disqualified. Maybe worse, you can break expensive parts and spend the next month fixing what one bad assumption caused.
Boost control is a good example.
If you want 12 pounds of boost, you do not blindly configure 12 pounds on top of the dome. You have to understand what is already built in. If the wastegate spring and static control are already giving you 3 pounds, then your adjustment is not 12. It is 9.
Otherwise you overshoot.
And overshooting in that world gets expensive fast.
That same exact reasoning applies in infrastructure.
A new project comes along and says it needs more bandwidth, more database space, more backup capacity, more storage, more compute, more anything. The lazy move is to just add what the request says.
Project needs five more gigs? Add five gigs.
But maybe you already have headroom. Maybe you have three missing, not five. Maybe the bandwidth is already there but the backup window is not. Maybe the storage exists, but retention, replication, or access patterns change the real number. Maybe the capacity is fine and the bottleneck is somewhere else entirely.
You have to understand the system before you add to it.
That is the same as the wastegate.
That is the same as the garage.
You do not add your desired outcome on top of a system without knowing what the system is already contributing.
That sounds simple, but it is one of the places people get themselves in trouble. They treat the new request like it exists in isolation. It never does.
Nothing worth building exists in isolation.
The boat had to fit in the garage while the garage still functioned. The race car had to make power while staying inside mechanical limits. The infrastructure has to support the project while still fitting inside the existing environment, budget, backup model, compliance requirements, and operational reality.
Same problem. Different material.
When the J-Marie finally came out of the garage, it barely cleared. Less than an inch. People were standing around watching because plenty of them still did not believe it was going to fit.
But it fit exactly the way it was supposed to.
The dollies hit the lip. The boat dropped just enough. The roof cleared the opening. Four years of work rolled out into the daylight.
That was a spectacular moment.
Not because I got lucky.
Because the math held.
That is the part I keep coming back to as a builder. Confidence is not the same thing as hope. Hope says, "We'll figure it out later." Confidence says, "I already figured it out, and now I have to respect the plan."
The older I get, the more I trust that distinction.
Whether it is a hull, a race car, a cloud migration, or an AI-assisted workflow, the rule is the same:
Know the constraint. Measure the space. Account for what is already built in. Watch for drift. Do not confuse close enough with correct.
Because math does not care how excited you are.
Math does not care how good the idea sounds.
Math does not care that everyone is standing around watching.
Math does not lie.
And if you did the work right, that is a beautiful thing.