In the mountains of western Sichuan, there is a writing system that fewer than ten people on earth can read. It belongs to the Ersu people — a Qiangic-speaking group of about 13,000, isolated from major population centers, whose priests developed a script used exclusively for religious texts: divination, healing, calendrical ritual. The script is called Ersu Shaba, after the priest class (Shaba) who are its only practitioners.
The script is organized in panels, like a comic book. Each panel contains a discrete thought. The animals depicted are those of the Chinese zodiac — each representing a day on the lunar calendar. But it is not just what appears in the panel that carries meaning: it is where in the panel it appears (left = morning, right = evening) and what color it is (red = fire; a moon-and-star sign in white = bright day; the same sign in black = dim). The full system is encoded in a visual grammar that the Shaba priests carry in their heads and pass to their successors. By any measure, it is an extraordinary achievement of condensed religious knowledge.
And it is dying. The restriction that made it powerful — only shamanic priests, always male, can learn it — is also the mechanism of its extinction. A population small enough, a tradition restricted enough, and the knowledge doesn't survive into the next generation. Fewer than ten readers. Maybe five hundred to one thousand years of continuous use. Ending now.1
This page is about proto-writing — visual symbol systems that encode concepts rather than language — and about the specific pattern that appears across its most significant examples: these systems tend to be born as monopolies, used by elites and priests to hold knowledge others cannot access, and they die when the monopoly collapses.
The distinction matters and is often missed.
Writing is a visual representation of spoken language. What writing does is encode the sounds (or words) of speech so they can be transmitted across time and space without the speaker present. To read writing, you must know the language it represents. Chinese writing encodes Chinese; cuneiform encodes Sumerian and Akkadian; the alphabet encodes phonemes.
Proto-writing is different. It encodes concepts and ideas directly, without going through spoken language. The symbol for "fish" in a proto-writing system does not represent how to say "fish" in any particular language — it represents the concept of fish. This is why the same proto-writing can be understood across language boundaries, and why it cannot record poetry, narrative, argument, or anything requiring the precision of spoken language.1
Writing was independently invented only four times in human history: Mesopotamia, Egypt, China, and Mesoamerica.1 Proto-writing appears far more frequently — because it solves a simpler problem. You don't need to map onto language; you just need stable, recognizable symbols that carry consistent meaning.
The instances:
The pattern across these systems is not incidental. Proto-writing tends to concentrate in specific contexts:1
Administrative: Mesopotamian cuneiform began as an accounting tool — who owed what to whom. Kaidā glyphs are the same thing: tax quotas for households too illiterate to read Japanese. The symbols record information for administrative control, held by the literate party (the taxing authority) and legible to the taxed through pictographic simplicity.
Magical/divinatory: Oracle bone script, Ersu Shaba, Adinkra — all three arose in contexts of ritual practice. The first writing may not have been for commerce or governance but for communicating with gods. The symbols encode sacred information that has power precisely because it is not generally known. A divination system loses its authority if everyone can read it — the priest's knowledge monopoly is the source of the priest's power.
Royal prerogative: Adinkra cloth was once exclusively royal — it lined the Gyaman king's bed, appeared on his ritual objects, was stamped on the cloth worn in his presence. Rongorongo was readable only by Rapa Nui nobility and priesthood. The writing system that belongs to the king marks royal authority every time it is deployed.
In all three contexts, the knowledge restriction is not incidental to the system's function — it is the point. A tax system written in symbols only the taxing authority understands gives that authority information asymmetry over the taxed. A divination system readable only by priests makes the priests the necessary mediators between the population and the sacred. A royal script that only nobles can read marks every document it appears on as an extension of royal power.
This is Hayden's secret society logic applied to literacy: restricted access to knowledge is the core resource that maintains hierarchy. The secret society's most powerful tool is not violence but information asymmetry. The proto-writing system is the written face of the same mechanism.1
Understanding proto-writing as a knowledge monopoly mechanism changes how you read several adjacent phenomena:
The death of writing systems becomes legible. When a script dies, it is almost always because the monopoly that sustained it collapsed. Rongorongo: the monopoly-holders (Rapa Nui nobility and priesthood) were killed by disease and slave raiding before they could transfer the knowledge. Ersu Shaba: the transmission chain is failing as the Shaba priest class is no longer being fully reproduced. Kaidā: the external imposition of literacy (compulsory education) made the taxation function of the proto-writing unnecessary. In each case, the script survives longer than its social context — and then, when the social context is gone, the script becomes an artifact without a reader.
Writing's origins become more legible. If accounting and divination are the two most common contexts for proto-writing development, and if writing developed from proto-writing in Mesopotamia (accounting) and possibly China (oracle bone divination), then writing may have been born from the same knowledge-monopoly logic. The first scribes were priests and accountants. Writing may be, at its origin, a tool for maintaining information asymmetry at scale — for creating a permanent record that only the initiated can read.
The modern equivalent is visible. Legal Latin, medical terminology, financial jargon, academic disciplinary vocabulary — all of these function as knowledge monopoly systems within a world that is technically literate. You can read the words but not decode the meaning without training that is itself controlled by an institution. The information asymmetry is maintained not by restricted access to the writing system but by restricted access to the interpretive framework.
The Ersu Shaba case is the most extreme example of the knowledge monopoly script and the clearest demonstration of its failure mode.
The Ersu are a small ethnic group in western Sichuan — around 13,000 speakers of a Qiangic language that is itself endangered. Their shamanic priests (Shaba) developed a script for religious texts: divination, the curing of the ill, calendrical ritual. The script is used exclusively by these priests. Its origins are poorly understood — possibly related to the Dongba symbols of the neighboring Naxi people, also used for religious purposes, possibly in use for 500 to 1,000 years.1
The Ersu Shaba operates through visual grammar rather than phonetic encoding. Each panel represents a discrete thought. The animals are the Chinese zodiac animals (representing days of the lunar calendar), but their position within the panel (morning/left, evening/right) and their color (red = fire, white moon-star = bright, black moon-star = dim) carry additional meaning that cannot be decoded from the image alone — you need to know the grammar.1
This grammar is not written down. It lives in the Shaba priests' heads and is transmitted by direct apprenticeship. There is no reference guide, no external documentation that would allow someone to reconstruct the system from the artifacts alone. The script and its interpretation are inseparable from the transmission chain.
The transmission chain is breaking. Fewer than ten people can now read the Ersu Shaba with any competence. The reasons compound each other: the Ersu community is small and geographically isolated; the Shaba priest class is not being fully reproduced because the social context that made shamanic priests necessary has been disrupted by modernization; the script is not taught in schools because it is not Chinese; it is not documented at sufficient depth because the linguists who might document it need access to the remaining priests, who are few and aging.
The Ersu Shaba will likely be fully unreadable within a generation. The tablets and texts will survive. The grammar in which they are written will not. The monopoly was so complete that no external record exists to reconstruct the system when the monopoly-holders are gone.1
Compare Rongorongo: the Rapa Nui script was lost in the same way — rapid collapse of the monopoly-holding class (disease, slave raiding, population collapse to under 200 individuals) left the texts intact and unreadable. We have hundreds of Rongorongo tablets. We cannot read them. The script may have been the fifth independent invention of writing in human history. We will likely never know what it says.1
Both cases demonstrate the failure mode of the knowledge monopoly script: too successful at restriction, not enough redundancy, fragile to demographic catastrophe.
The proto-writing analysis provides a template for identifying knowledge monopoly mechanisms wherever they appear:
Ask who controls the interpretive layer. A symbol is only useful to whoever can read it. The Ersu Shaba character means nothing without the priest's grammar. Legal jargon means nothing without legal training. Financial derivatives documentation means nothing without financial training. The question is not "who can see the symbols?" but "who can decode them?" The gap between these two populations is the monopoly.
Track what happens to the knowledge when the holders die. The most revealing test of a knowledge monopoly is what happens to it when the monopoly-holders are removed. Rongorongo: catastrophic depopulation → total loss. Kaidā: external literacy imposition → folklorization (now found on tourist merchandise). Adinkra: Ashanti conquest → dissemination (royal prerogative became general, which weakened its power but preserved the knowledge). The durability of the knowledge depends not on the symbols but on the social structure that gives those symbols meaning.
Identify what the monopoly is protecting. Administrative scripts protect information asymmetry in resource allocation. Divinatory scripts protect access to the sacred (and to the social authority that comes from being the community's mediator with the divine). Royal scripts protect the legibility of authority. Each protection serves a different hierarchy — but the mechanism is the same: knowledge restriction as power.
The demographics problem. A knowledge monopoly that depends on a small, specialized population is fragile. The Ersu Shaba needs Shaba priests. The Rapa Nui Rongorongo needed Rapa Nui nobility and priesthood. When either is disrupted faster than the transmission chain can be reproduced, the script dies with the holders. This is not unusual — it is the normal failure mode of restricted-knowledge systems.
The colonial interruption. Both Kaidā and Adinkra were obsoleted not by demographic catastrophe but by the imposition of external literacy systems. When you can read and write Japanese or English, the specialized proto-writing that only worked when literacy was absent becomes either folkloric (Adinkra now appears on modern Ghanaian fashion and cultural objects) or archival (Kaidā is studied but not used). The colonial literacy intervention destroyed the monopoly without destroying the knowledge — but without the monopoly, the knowledge has no function.
The documentation failure. The deepest failure mode is when the monopoly-holders disappear before anyone documents the interpretive grammar. This is Ersu Shaba's trajectory and Rongorongo's fate. The physical artifacts survive; the keys to their decipherment do not. We know the Jiahu symbols (6000 BC) and we know their similarity to oracle bone script 3,000 years later — but we do not know what connects them, because no one wrote down the grammar.
Is proto-writing preceding writing, or a parallel development? The standard account treats proto-writing as a precursor stage to full writing. But Adinkra, Ersu Shaba, and Rongorongo exist in cultures that have access to writing and chose not to develop it in this direction. The choice to maintain proto-writing rather than develop full writing may not be a limitation but a preference — the proto-writing serves the function (knowledge restriction, ritual encoding) better than full writing would. Full writing, once developed, is democratizing — it is much harder to maintain as a monopoly. [POPULAR SOURCE — this inference goes beyond the essay's explicit claims] [SPECULATIVE]
The Jiahu-oracle bone similarity. Sixteen symbols from 6000 BC share recognizable similarity with the oracle bone script of 3000 BC. The gap is 3,000 years. The similarity is striking enough that scholars have noted it. Whether this represents continuous transmission (an extraordinary claim), parallel development from common cognitive structures, or coincidence is genuinely unresolved.1 [UNVERIFIED — requires specialist Sinological assessment]
The Rongorongo question. Was Rongorongo full writing (the fifth independent invention), proto-writing (mnemonic for genealogical records), or a navigational aid? Barthel's calendar interpretation is the only accepted specific purpose; everything else is contested.1 [UNVERIFIED]
The writing essays were guest-written by "Pygmy Glottochronologist" (anonymous, @GreatValueArhat on Twitter), not by Stone Age Herbalist. Same [POPULAR SOURCE] classification applies — anonymous, unaccountable, cannot verify specialist credentials.1
The guest author writes as a linguistics enthusiast ("linguistics sperg") with genuine curiosity and apparent competence — references to Butinov and Knorozov (who actually deciphered the Maya script), awareness of the distinction between writing and proto-writing, acknowledgment of scholarly debates. This is not charlatanism; it is competent popular synthesis.
The key tension: the knowledge monopoly analysis developed on this page is an inference from the guest essay's examples, not the essay's own thesis. The essay is focused on the nature and variety of proto-writing systems. The connection to secret societies and knowledge restriction as hierarchy maintenance mechanism (the Hayden framework) is a synthesis generated by placing the writing essay alongside the secret society essay — not a claim the writing essay makes. This synthesis is plausible and coherent, but it should be held as a generated insight rather than a sourced claim.
Cross-Domain ↔ History: Knowledge Monopolies That Persist Require Initiatic Transmission Structures; Those That Don't Collapse Into Extinction
Kelly's research on knowledge transmission systems reveals a critical insight about the knowledge monopoly problem this page articulates: knowledge monopolies survive only when they are embedded in initiatic transmission structures. The Ersu Shaba, Rongorongo, and Adinkra fail not because restricted knowledge is inherently fragile, but because they lack the formal transmission containers (master-apprentice relationships, graduated initiation phases, relational mentorship under competent elders) that Kelly documents in surviving knowledge systems.
Compare two failure modes: Ersu Shaba ends because the Shaba priests die before finding successors to initiate. Rongorongo ends because the Rapa Nui population collapses before the knowledge can be transmitted. But also compare: Aboriginal Law, maintained by Australian peoples for 65,000+ years, survives because it is embedded in systematic initiatic structures where elders transmit knowledge to carefully selected initiates across graduated phases. The Polynesian genealogical systems (Whakapapa) survive because they are maintained through formal transmission lineages where genealogies are recited and taught in containers with competent keepers.
The distinction Kelly reveals: a knowledge monopoly that depends on isolated knowledge-holders is inherently fragile. One death, one demographic disruption, one loss of the keeper ends it. A knowledge monopoly that is embedded in a structure of transmission — where the knowledge is held in containers (ceremonial contexts, apprenticeships, initiatic phases) and transmitted from elder to carefully prepared successor across generations — can persist indefinitely.
The handshake reveals: proto-writing systems fail to persist not because of the nature of restricted knowledge, but because they are maintained as individual specializations rather than as embedded in initiatic transmission structures. The Shaba priest dies; the script dies with him, because there is no formal apprenticeship structure preparing the next generation. The Rapa Nui nobility collapses; the Rongorongo dies with them, because there is no mechanism for transmitting the knowledge outside the restricted elite to a backup population. In contrast, knowledge monopolies that embed themselves in formal initiatic structures (restricted access, graded transmission, competent elder authority, graduated challenge) persist for centuries or millennia. Kelly's cross-cultural documentation shows this principle: sustainable knowledge monopolies are not those that restrict access most severely, but those that combine restriction with carefully designed transmission structures. The knowledge that cannot be transmitted is knowledge that is already dead, waiting only for the last keeper to die.2
History — Secret Societies and the Biology of Hierarchy: Secret Societies and the Biology of Hierarchy — Hayden's 12-characteristic model of secret society formation places restricted access to esoteric knowledge at the center of the institution's power. The Ersu Shaba, Rongorongo, and Adinkra are all instances of this same mechanism expressed as writing systems: restricted literacy used to maintain religious, royal, or priestly authority. The insight: secrecy and hierarchy are not independent features — secrecy IS the mechanism of hierarchy maintenance, whether the secret is a ritual procedure, an initiation experience, or a script only priests can read. The writing system and the secret society are the same social technology in different material forms.
Cross-Domain — Deep Time and Animist Horror: Deep Time and Animist Horror — The Magdalenian trilobite pendant and carved beetle raise the question of whether proto-writing precursors existed in the Upper Palaeolithic — whether the Magdalenian use of symbols on bones and cave walls represents an early form of the same knowledge-encoding impulse that produces Ersu Shaba 13,000 years later. The animist framework that may have governed Magdalenian object use (form is temporary, material carries history) is the same framework that governs the magical proto-writing of Ersu shamanism (symbols carry power that must be handled by the properly initiated). The insight: the chain from Magdalenian symbolic objects to Ersu Shaba to modern professional jargon is not a history of progress but a continuous negotiation between the desire to encode knowledge and the desire to restrict access to it.
The Sharpest Implication
Rongorongo may be the fifth independent invention of writing in human history. We will almost certainly never know what it says. The tablets survive; the readers do not; the interpretive grammar was never written down because the whole point was to keep it from being written down. The monopoly worked — so completely that it destroyed the knowledge along with the monopoly when demographic catastrophe hit. This is not a story of primitive failure. It is a story of a protection mechanism functioning exactly as designed until the protected party was gone. The implication: every knowledge monopoly builds in its own extinction event. The more successful it is at restriction, the more fragile it becomes to the loss of the monopoly-holders. Modern equivalents abound: legal precedent systems that only licensed attorneys can navigate; financial instruments so complex that only a handful of specialists understand their interaction effects; medical protocols encoded in terminology that patients cannot access. These are Rongorongo in slow motion.
Generative Questions