Where the icon comes from.

I copied it

I did not design this composition. I copied it.

I had Ghostty and iTerm2 in my Dock, and I liked how they sat there. Both put their mark up in the top-left quarter of the tile rather than centred in it, and I took that arrangement more or less directly. When the three of them lined up together in the Dock it looked deliberate — like a set, like a family of tools that belonged next to each other. It looked designed. That was the whole reason, and for a long time it was the only reason I had.

The mark itself came from somewhere else: an earlier name, a C beside a seven-dot matrix. When the project was renamed I kept the composition to the unit and swapped the letter. So by then the placement was two removes from wherever it started — I got it from Ghostty and iTerm2, and I carried it across a rename without ever asking where they had got it.

The eureka

One day I noticed the composition was very close to Terminal.app — Apple's own. Dark field, mark tucked into the upper-left, the rest of the tile left alone.

That bothered me. Three terminal apps and Apple all landing on the same arrangement is not a coincidence, and "everyone copies Apple" felt too easy an answer. So I went looking for the reason, and it turned out to be much better than imitation:

Terminal.app puts the prompt in the top-left because that is where a cursor actually is.

Row 1, column 1 — the home position of any left-to-right text screen. It is not a composition choice at all. It is literal depiction, inherited from physical terminals: a real screen has its cursor waiting in that corner before you type anything.

So the lineage I had imagined — Apple invents a layout, everyone copies it — is wrong. What is really happening is that everyone is drawing the same object, and the object has its cursor in that corner. The arrangement is convergent, not derivative. Ghostty and iTerm2 did not copy Apple any more than two photographers copy each other for pointing a camera at the same building.

The empty space is the point

The part that matters is what this implies about the three quarters that look empty.

On a terminal icon they are not decoration and not padding. They read as screen that has not been written to yet. The mark is not floating in a corner — it is sitting at the beginning of a surface, with the whole surface still ahead of it. That is real meaning, carried entirely by placement.

Centre the mark and you destroy the meaning. Centring makes it a logo. Top-left makes it a place in a document.

Where the reason stops being mine

Here is the part I would rather not write, which is exactly why it is worth writing.

That reason only transfers if the tile reads as a screen. Terminal.app's tile is a depiction of a window — a dark rectangular field standing in for a display. Mine is a soft radial-gradient squircle. It is a modern app tile, a brand surface. Nothing about it says screen.

What is actually mine

Two arguments survive without borrowing anything from terminals.

The mark is directional. It is a letter followed by a dot trail running rightward — the trail was drawn as a cursor moving away from the letter, and that is still how it reads. A directional mark placed dead centre goes static. Placed top-left it has somewhere to go: the space to its right is the space it is heading into.

And asymmetry is rare in a Dock. Nearly every icon centres its mark and fills its optical area. Not doing that is a cheap, permanent way to be recognisable.

There is a cost, and it is measured rather than felt. When I built the site's favicon I had to abandon the quarter placement entirely and centre the mark at 76% of the tile — because at 16 pixels the quarter left it about seven pixels wide and unreadable.

The lesson I kept

I copied a composition because it looked good — and it was good, for a reason I did not know, which turned out to be a reason that only partly applies to me. That is not a failure. Borrowing a solution that many people arrived at independently is usually smart.

The failure would have been discovering the reason and never checking whether it was my reason too. Sometimes the honest answer is "no — but here is a different one that is."