The Pattern

Eleven people, five continents, six centuries, and very nearly the same set of decisions.

The institute's closing note

§1 of 5The Brief

Read the eleven briefs together and they sort into three kinds, with almost nothing in between.

The first is absence: the language has no writing at all. Sequoyah, Bukele, Mashtots and — in the sense that mattered to its users — Evans's Cree all begin here.

The second is misfit: writing exists but the available script serves the language badly. Sejong facing Chinese characters, Kanté facing Ajami and Latin, the Barry brothers facing exactly the same pair a generation later, Lookout facing a Latin orthography that was teaching his students English sounds. This is the largest group, and it is growing: every script invented in the last century belongs to it.

The third is alignment, where the technical problem is secondary to the political one. Cyril needed Slavonic to be a liturgical language rather than a dialect; Wulfila needed a Christian scripture that did not read as a pagan one; Thaana's anonymous maker turned the islands' writing towards Mecca and possibly away from outside readers.

What no brief in this series is: aesthetic. Nobody in twelve chapters invented a script because the existing letters were ugly.

Every one of them is an argument about who should be able to read, or about whose institution the reading belongs to. That is the series' founding claim and eleven cases have not produced a counter-example.

§2 of 5The Constraints

Three constraints recur, and the third is the one that is usually left out of the story.

The phonological constraint is the obvious one, and it is nearly always solved well. Every script in this series fits its language more closely than the script it replaced. Not one of the eleven produced a deep orthography — a Sejong or a Kanté writes what is said, and the mismatch that afflicts English and Tibetan spelling is a product of centuries of drift, not of design. Deep orthographies are what happens to writing systems; nobody builds one on purpose.

The political constraint is the one that shapes appearance. Where a designer needed to mark a boundary, the letters look like nothing nearby: Glagolitic, Armenian, Osage. Where a designer needed to claim an affiliation, they lean towards it: Thaana and Adlam and N'Ko all run right to left, towards the Arabic tradition of their readers rather than the Latin of their administrations. Letterform distinctiveness is a political variable, not a design preference.

The manufacturing constraint is the forgotten one, and it governed every case before 1990. Evans cast type from tea-chest lead and made ink from soot and fish oil; that is why Cree syllabics rotate rather than multiply, because a system where one shape serves four syllables needs a quarter of the sorts. Sequoyah's syllabary needed a press, got one at New Echota, and lost it to the removals. Kanté hand-copied for forty years. The constraint that decided whether a good script survived was, for most of this history, metallurgical.

§3 of 5The Decisions

Two decisions recur so consistently that they look like findings about human cognition rather than about design.

The unvarying borrowing of shapes. Not one inventor in this series drew an alphabet from nothing. Sequoyah took Roman letters from a spelling book, Wulfila took Greek uncials, Cyril's successors took them again, Thaana took numerals, Bukele took regional marks, Evans took Pitman shorthand. Even Glagolitic and Armenian, the two that look most invented, are argued to draw on models. What is invented is the system; the marks are almost always found.

The syllable as the naïve unit. The three inventors in this series who worked without any grammatical tradition — Sequoyah, Bukele, and Evans working with Cree collaborators — all arrived at the syllable. The phoneme is not what a speaker hears; it is what an analytic tradition teaches you to hear. Where a tradition was present (Sejong's phonology, Mashtots's Greek, Kanté's schooling) the result is alphabetic. Where it was absent, the result is a syllabary. Eleven cases is a small sample and this is stated as a strong tendency, not a law.

Then there is the case that stands apart, and it is why Yi is the glyph on this chapter. Every other design here was made by a person or a small circle. The Liangshan Standardized Yi Syllabary of 1974 was made by a committee: a state standardisation that took the thousands of variant characters of classical Yi, in use across generations of scribes with no fixed inventory, and reduced them to a fixed set of around 800 syllabic signs, approved for official use in 1980.

The committee produced something no individual in this series did — a genuinely regular, exhaustively specified, teachable inventory — and it did it by discarding the variation that made the classical tradition what it was. Individuals in this series design for a community. The committee designed over one. Both are acts of script design; the difference in what gets lost is the point of including it.

§4 of 5The Adoption Curve

The single strongest finding of the series is about adoption, and it contradicts the intuition the series starts with.

Resistance comes from the literate, not the illiterate. Choe Manri's 1444 memorial argued that abandoning Chinese characters meant abandoning civilisation — filed by a class whose credential was the difficulty of the old script. Frankish clergy fought the Slavonic liturgy. Sequoyah's opposition came from neighbours, but the sustained pressure on every script here came from people who could already read something else. Nobody who cannot read has ever objected to a new alphabet.

A script needs a corpus, not a decree. Mashtots trained scribes and produced a library within a generation, and the alphabet outlived the kingdom by fifteen centuries. Wulfila produced one book tied to one confession, and when the confession lost, the script had nothing left to do. Kanté wrote around 180 works with no state behind him and N'Ko survives; Glagolitic had a corpus and papal sanction and still lost to an easier-to-cut replacement that inherited that same corpus. Survival tracks the breadth of what is written, not the quality of the design and not the force of the authority.

Schooling and literacy are separable. The Vai case is where this was actually measured — Scribner and Cole's fieldwork in the 1970s used a community with script literacy acquired outside schools to separate the two, and found that the sweeping cognitive claims made for literacy belonged largely to schooling. That finding was only available because a nineteenth-century syllabary spread person to person and never got a school system.

And the gate has moved. For every chapter up to N'Ko, the last obstacle was manufacture. From Adlam and Osage onward it is a committee: a Unicode proposal, a code block, a font. Both were encoded in the same 2016 release, one by two men who started as schoolboys with no institution, the other by a tribal government's language department. The queue does not care which. That story is Admission Procedures.

§5 of 5The Verdict

The series was assembled to test one claim: that a writing system is a designed interface, that its decisions are legible and criticisable like any other interface's, and that those decisions have consequences measurable in who gets to write.

Eleven cases support it, with one qualification the institute wants on the record. Design quality is real and it is not decisive. Hangul is the best-designed script in this building and spent four centuries as the vernacular alternative to Chinese. Glagolitic was carefully made and lost to a cheaper copy. Gothic was competent and died with its confession. What predicts survival is whether a community has things it wants to write and the means to write them — and a good design lowers the cost of that, which is worth a great deal, but it does not decide the outcome.

The second finding is about who does this. The eleven are a king, a silversmith, a monk, a bishop, a philosopher, a trader, a missionary, a teacher, two schoolboys, an elder, and one person whose name is gone. Script invention is not the property of institutions, of states, or of the literate class. It has been done by children. It has been done by someone who could not read.

And it is not finished. The two most recent scripts in this registry were encoded in 2016, one of them begun by a fourteen-year-old and a ten-year-old with a notebook. The Inventors' Room is an open collection, and this series is declared complete at twelve chapters rather than concluded.