Colophon: how Phi is made

Phi is my language. I designed its sound system, fourteen consonants and five pure vowels that never glide into each other, chosen to make breath and full articulation easy to notice. I designed its grammar around a single exceptionless principle, announce then deliver: every modifier, every particle, every dependent clause declares itself before the thing it affects. I decided what the language refuses to name: money, the counted minute, units of measure, gendered person-words, the vocabulary of domination. I made the rulings that settle how time is told, how names are made, how the dead are spoken of. Those decisions live in a canon file with my name on every one of them. They were argued, sometimes for days, and then I decided.

Phi began as a folder of Markdown notes in Obsidian, and the first commits are dated April 2021. Two hundred and eighty-nine commits later, at the end of 2022, those files held the four fricative digraphs, the five pure vowels, the syllable that opens on a consonant and closes on a vowel, the rule that lets a digraph be the only cluster a word may carry, and the refusal of diphthongs that still governs every word in the lexicon. wa already asked a question. lo already made a plural. There were Tengwar fonts in the manual directory, waiting for a second script. Nothing then in existence could have written a line of it, and nothing did.

Not all of it survived the years that followed. There were nine plain consonants then and there are ten now, and I have rebuilt the particle system from the ground up since, so most of those early forms now do different work or none at all. What held is the sound of the language. That is the part you meet first, and it is the part I would have found hardest to change later.

I did not always build it this way. I build the language with AI assistance now, and I want to say plainly what that means, because it matters here and because I would rather you hear it from me than find it in the commit history.

The best way I can describe the process is a workshop. William Morris designed patterns; the hands at Morris & Co. wove and printed them under his eye, to his standard. Nobody says the weavers, rather than Morris, made those textiles. The argument historians actually have is over which patterns were his, because late in the firm's life some of the designing quietly passed to other hands, and it took a century of scholarship to relabel the work. I take that as the model's fine print, and as a warning: the arrangement stays honest only while the designing stays at the designer's desk. The designer-and-workshop model is the Arts and Crafts mode of production, and it is, more or less exactly, how Phi is made. My workshop is stranger than his: the drafting hands are an instrument rather than apprentices. But the patterns are mine, the standard is mine, and nothing enters the language unless I have held it up to the light.

Concretely, the instrument drafts against my designs: candidate words built to my phonology and my sound-symbolism principles, translations built to my grammar, documentation built to my rulings. It holds consistency at a scale no single memory can, checking a lexicon of over a thousand words and every example sentence in every document against the grammar, at zero tolerance: the whole corpus validates with no errors before anything is committed. And it does the accounting that makes honesty checkable, so that when I translate a chapter of Morris, every Phi sentence carries the exact source clause it renders, and anyone can audit what was kept, what was changed, and what was lost.

What the instrument does not do is decide. Words are proposed to me and I reject them. I have thrown out coinages for sounding wrong, re-glossed them for meaning the wrong thing, recoined them when a ruling changed underneath them. Every merge is my hand. If you want to know whether a human is in this loop, read the project's own history: it is largely a record of an instrument being corrected.

I also want to be honest about why I work this way, because the reason is not convenience. A complete language with a real literature has essentially never been finished by one person. Tolkien spent sixty years on his languages, revised them endlessly, and left them incomplete, with no long connected texts; he wrote the novels to give the languages somewhere to live, and the languages still outlived his lifetime unfinished. Phi's natural size, a full grammar, a living lexicon, a novel's worth of literature with interlinear glosses, is larger than my remaining decades. Done by hand alone, this work is decades long, and the honest version of that sentence is that it would never have been finished at all; it would have joined the long tradition of beautiful languages that died in notebooks. I decided the idea deserved to exist at its full size more than my process deserved to be pure.

That decision has costs, and I hold them knowingly rather than pretending them away. These tools carry real critiques: the unconsented work in their training data, the energy they burn. I do not control either, and I will not tell you those problems don't apply to me because my project is small or my intentions are good. I hold the tension the same way we hold organizing mutual aid over corporate platforms: eyes open, weighing what the compromise buys against what it costs, ready to be argued with. What I can say is narrower and true. This work displaces no one's livelihood; nobody was ever going to be commissioned to hand-gloss eighty thousand words of Morris into a personal language. And it is the opposite of slop, whose defining feature is volume without a human who can be questioned. Phi has exactly one human, everything routes through him, and this document is where he answers questions.

One last thing, for whatever it is worth. The language itself is my position on thoughtless automation. Phi refuses the counted minute and the priced hour. It has no word for wage, and it does not treat speed as proof of worth. Every sentence gives the speaker somewhere to notice what is known, what is intended, and what the words may do between people, because I believe the constraints are the teaching. I used a fast tool to build an artifact of attention, a language that keeps asking me to be here for the sentence in front of me, and I am at peace with that trade. If you aren't, I understand, and I'd rather have that conversation in the open than win it by omission.