There is a small diagram on the front page of notepat.com. It says that the QWERTY key A plays one note, that S plays the next, that the letters under your fingers are also pitches. It contains no code. It is not a binary, not a service, not a runtime. It is a table — a list of agreements about what a key means. And it is software in every sense that matters: it is versioned, it is invented, it is forked, people demand it from one application to the next and complain when it is missing.

I wrote a paper about that table. This is the short version of why.

A thing without a name

A keymap is the layer everyone uses and no one looks at. The mapping from QWERTY to piano notes that ships inside every digital audio workstation. The Vim commands that travel from one editor to the next. The Rubik’s cube notation that lets a stranger in another country hand you a twenty-move solution on a napkin. These are not programs and they are not data. They are conventions that behave like software — and the categories we usually reach for can’t see them. The open-source-versus-proprietary axis has nothing to say about a thing that has no source. Format theory wants a specification; most keymaps don’t have one.

A thing without a name is hard to see, harder to critique, and impossible to cite. So the first move in the paper is to name the category. I call it social software — and I mean three things at once. I mean Clay Shirky’s 2003 sense, software made for sociality. I mean this initiative, the Social Software program at UCLA Design Media Arts that I’m writing inside of. And I mean a third reading that the other two make room for: software whose existence is itself a social agreement. A keymap exists because enough people decided to act as if it does. Stop agreeing and it’s gone — not deprecated, not end-of-lifed, just no longer true.

The case in the middle

The clearest example is also the worst-designed one. Open almost any DAW and you can play the keyboard like a piano: A is C, W is C-sharp, S is D, and so on up a staircase of letters that bear no relation to the notes they make. It is unowned — no company invented it, no patent covers it, every program ships it because every other program ships it. It is a near-perfect specimen of social software. And it is bad. It spends two hands’ width on a single octave. It is illegible without a tutorial. Nobody chose it; it accreted.

I gave it a name too. I call it AWSED, after the keys A-W-S-E-D that spell its first five notes, rhyming on purpose with gaming’s WASD. The rhyme is the argument: once the thing has a name, you can say “AWSED is bad” in three words, and three words is a length at which a critique can travel. notepat is my counter-proposal — a keymap I designed on purpose instead of letting one happen to me. The letter keys play the notes they name: C plays C, D plays D, E plays E, and the sharps sit on the keys around them. Underneath runs what I call the contract line: the tonal mapping is identical across every place notepat lives, and above that line each device is free to improvise its own extended layer. Below the line, deviation is breakage. Above it, deviation is idiom. A keymap that requires a manual is a keymap that has lost its social-software character.

Play it below — click a key or just type. The gold keys are the ones that are their own note.

Why this belongs in a room full of scores

Cycle 2 of Social Software is called Scores for Social Software, facilitated by Casey Reas, and it asked each of us for a score — a way of choreographing events in time through text, diagrams, or instruction. The lineage is the Fluxus event score, Cage and Knowles’ Notations, Hans Ulrich Obrist’s do it, the work Natilee Harren brought us through The Scores Project. I didn’t have to leave my paper to answer the prompt. A keymap is a score. It is an instruction for a body — these fingers, this key, this note — that means nothing until somebody performs it, and that only becomes social once it’s written down in a form that can be handed to the next person.

That last part is the half the paper cares about most. A keymap has two bodies: the binding (A plays C) and the notation (the textual form the binding takes when it travels). Vim has both — :wq is a phrase you can say out loud, type, and teach. The DAW staircase has the first and not the second; there’s no portable way to write it down, which is exactly why it never improved. Where there is no transmissible notation, there is no social mediation to mediate. A designer who ships only the binding-table has shipped half of the object. The ask the paper ends on is small and specific: ship the keymap as a first-class thing — readable, exportable, forkable — and ship the notation alongside it.

The edition

This page accompanies the first numbered print edition of the paper, cut in June 2026 as my contribution to Cycle 2. The edition mechanic is the part I love most, because it is itself a piece of social software: every member of the cohort produced sixty-four copies of their own score, and on June 13, 2026, at Fuser — 1811 Blake Ave, in Frogtown — we brought them all into one room and collated them into a single publication. One assembled set per copy, sixty-four times over. A paper about conventions that exist because people agree to act on them, bound by hand into an object that exists because a room full of people showed up and agreed to make it.

The cohort each arrived with a score: Jordan Silver’s Sonic Architecture, Em Lugo’s Ambling Los Angeles, Æther Cavendish’s Vigil Score, Thomas Noya’s Biophonía, Mavyn Vu’s the radio is an altar : portal, Alex Espinosa’s Music for World Computers, Darlyn Phan’s Meridians, Chelly Jin’s Software as a Choreography, Banyi Huang’s Conjuring Notness, and mine — the keymaps paper, under the banner of Aesthetic.Computer. Convened by Casey Reas and Lauren Lee McCarthy, produced with Wylie Kasai. Reading those scores next to each other is the best argument I can make for the category: none of us shipped a binary, and all of us shipped software.

The keymap on the front of notepat is still there. It still contains no code. It is still just a table of agreements — which, it turns out, is one of the oldest and most durable kinds of software we have. It only needed a name.

—Jeffrey Alan Scudder, June 2026