Four Years Old, on a 4×8 Sheet of Plywood
5 min read
My father gave me my first train set when I was four.
HO scale, on a single 4×8 sheet of plywood. He didn't just put track on it — he painted it. The figure-8 went down on the board by hand, and around it, scenery and a river. No station. Just the ground the trains ran through, made by him, on a Saturday I don't remember.
An over-under figure-8, if you haven't seen one: the track leaves, loops, and comes back to cross over itself, one leg climbing to a bridge while the other passes underneath. It's the most track you can fit in the least space. It's what you build when a sheet of plywood is the whole world you've got.
We were in a small house on top of Queen Anne Hill in Seattle. From up there you looked down at the grain silos by the railroad tracks — Pier 86, the Port of Seattle grain terminal, with better than two miles of rail running along the water underneath it — and past that, Puget Sound. Real trains below the window. A painted one on the plywood. Nobody pointed that out to me and it didn't need pointing out.
What it did
Here's where I'd normally tell you what that moment meant.
I can't. Not honestly. Something started there that turned into an engineering mindset, and I have never been able to describe the mechanism. It's not a lesson I can extract and hand to you. It's not a metaphor that resolves. It's just that a four-year-old watched his father make a world out of paint and plywood, and something about the world being makeable went in and stayed in, and thirty-odd years of a career came out the other side.
The way I'd put it now is that the moment sent out tendrils — strings of pattern and memory that keep touching each other in the dark, and where they overlap you get an interference pattern. That's as close as I get. I'm wary of pushing it further, because the honest answer is that the line from that plywood to what I do now is real and I cannot draw it.
Thirty-six years later I shipped my first application. It's for model trains.
I'm not going to pretend I planned that.
Three years back in it
For about three years now I've been properly back in the hobby: model trains, and DCC — Digital Command Control — and all of its electrical signals. Instead of controlling the track voltage, you send addressed digital packets down the rails, and every locomotive carries a decoder listening for its own address and ignoring the rest.
Which makes a layout a small distributed system with a bus and an addressing scheme. That is not the pitch anyone makes for model trains, and it's a large part of why I find it absorbing.
Somewhere in there the engineer noticed he was doing the same work he does for clients, except the client was him.
That turns out to matter. The craft doesn't change and the standards don't change. What changes is that nobody is waiting on it, so the part of your attention permanently running a stakeholder simulation gets to switch off. What's left is the meditative state — where you look up and three hours are gone and you're not tired, you're refilled. Most people need a version of that, and a lot of us have quietly stopped making room for one. Mine runs on 124mm modules.
Why the app exists
The practical problem is that Kato Unitrack is sectional. Every piece is a fixed geometry with a molded roadbed and Unijoiners on each end. You can't cut it to fit. A curve is 315mm radius at 45 degrees or it's a different piece. So a plan either closes or it doesn't, and you find out by kneeling on the floor with a box of track.
RailMind is an iPad app where you drag Unitrack pieces onto a canvas and they snap end-to-end only where real track can actually connect — at the Unijoiner endpoints, at the real radii, in N and HO. If it snaps on screen it snaps on the plywood. Everything else follows from enforcing that: the solver that tells you which piece would close a loop that's almost closed, the buildability check, the parts list that prices out what you'd have to go buy.
Two decisions worth stating plainly, because both are the kind that get quietly reversed later.
There's no AI in it. An early version of the concept had an engine that proposed layouts for you to accept or reject. I cut it. Not out of principle about the technology — I build with it every day and I've written plenty here about that — but because proposing the layout removes the part I do this for. Kneeling on the floor with a box of track is the hobby. I'll automate the arithmetic of whether a loop closes. I won't automate the deciding.
It runs entirely offline. No network, no account, no telemetry. It's a drafting tool for something you do at a table in your house, and it has no business phoning anywhere.
There's a version of this post that ends with a link and a price. This isn't that — it's in App Review as I write. I'd rather tell you why it exists.
My father was not making a point. He was fitting the most track into the space he had, and painting a river on it, for a four-year-old, in a small house on a hill.
I've been building versions of that ever since, and I still can't tell you exactly how the one became the other.