DOI: https://doi.org/10.5281/zenodo.21947434
Canonical: https://thonly.org/research/vinaya-governance-primitives-distributed-dharma-networks · Licence: CC0 1.0
Note. This paper is the procedural companion to AGI Monks: The Caretaker-not-Ordained Pattern (corpus slug
agi-monks-caretaker-not-ordained, 2026-05-04). That paper allocates roles between AI agents and human monastics in religious institutions: five operational roles AI may take, and the acts reserved to those with standing in an ordination lineage, among which, in the Theravāda case, are the formal acts of the Saṅgha. The present paper asks a different question: how a network of monasteries, whose operations are run by AI caretakers and which house human monastics, can take valid collective decisions, settle disputes, allow local exceptions and amend its operating rules. It reads the procedural chapters of the Theravāda monastic code, the Khandhaka of the Vinaya Piṭaka, for templates. It addresses the Sangha pillar of the Silicon Wat program (siliconwat.org), which Silicon Wat as Cambodian Civilizational Architecture (§6.4) names as the least developed of the program's three pillars. The authors invite review from Vinaya scholars and from monastic communities, above all on the readings of the canonical procedure in §§4–7 and on §9.
This paper studies the procedural law of the Theravāda Buddhist monastic code as a source of governance procedures for a federated organization of many sites, and maps them onto a planned network of monasteries whose daily operations are run by AI agents and whose formal acts belong to human monastics. The code's procedural chapters (the Khandhaka of the Vinaya Piṭaka) supply four procedures such an organization needs. (1) The validity of a collective decision: a motion read to the assembly, followed for the weightiest acts by three announcements, with silence taken as assent; an assembly complete within a fixed boundary, with absent members' consent conveyed by proxy; a minimum quorum for each class of act; defects that make an act void from the start; and offences for reopening a matter validly settled. (2) Typed dispute resolution: four kinds of case (disputes, accusations, offences and the community's business) routed to seven settlement procedures, among them settlement in the presence of the parties, majority vote by ballot under stated conditions, and a collective amnesty for offences committed during a quarrel. (3) Exception clauses attached to each rule, and allowances written for a defined class of region. (4) Rules recorded with the case that prompted each one and with each later amendment. Each procedure is read from the Pāli text and compared with modern practice: unanimous consent in parliamentary procedure, lazy consensus, consent decision-making, veto-based on-chain governance, multi-door dispute resolution and comply-or-explain codes. The limits are stated: the procedures were built for one assembly meeting in one place, AI agents hold no seat in the formal acts they model, and whether the procedures scale to a large network is open.
Keywords: governance procedure, collective decision-making, decision validity, quorum, proxy consent, consent by silence, unanimous consent, lazy consensus, consent decision-making, veto-based governance, optimistic governance, dispute resolution, dispute-to-process matching, multi-door courthouse, majority vote by ballot, amnesty, exception clause, regional exception, comply or explain, rule amendment history, precedent, federated organization, multi-site organization, nested governance, Buddhist monastic law, Vinaya, Khandhaka, saṅghakamma, adhikaraṇa-samatha, anāpatti, AI agents in religious institutions, defensive publication.
Pāli and coined names used in this paper, and the standard terms a reader in governance, law or institutional design would search for them.
| Term used here | Standard term |
|---|---|
| Vinaya Piṭaka | the Theravāda Buddhist monastic code |
| Khandhaka (Mahāvagga, Cullavagga) | the code's procedural chapters |
| Suttavibhaṅga; Pātimokkha | rule-by-rule analysis of the code; the list of monastic rules recited every fortnight |
| Saṅgha; bhikkhu | the monastic community; a fully ordained monk |
| saṅghakamma | formal act of a monastic assembly; collective decision procedure |
| apalokana-kamma | act by announcement and assent |
| ñatti; ñatti-kamma | motion; act by motion alone |
| ñatti-dutiya-kamma | act by a motion and one announcement |
| ñatti-catuttha-kamma | act by a motion and three announcements |
| anussāvanā; kammavācā | announcement (reading) of the motion; the formal text of an act |
| sīmā | the boundary within which every qualified member must attend or send consent |
| chanda | consent conveyed by proxy |
| paṭikkosanā | protest (objection) that halts an act |
| vagga-kamma; samagga-kamma | act by an incomplete assembly (void); act by a complete assembly |
| ukkoṭana | reopening a matter validly settled |
| anumatikappa | acting with an incomplete assembly and seeking the absent members' consent afterwards (ruled not allowable) |
| adhikaraṇa | a case or legal issue (four kinds) |
| adhikaraṇa-samatha | settlement procedure (seven kinds) |
| sammukhā-vinaya | settlement in the presence of the community, the rule and the parties |
| yebhuyyasikā; salākā | decision by majority; voting ticket (ballot) |
| tiṇa-vatthāraka | collective amnesty for offences committed during a quarrel ("covering over with grass") |
| ubbāhikā | referral of a case to an appointed committee |
| anāpatti | non-offence clause; exception |
| border-country allowances (paccantima janapada) | exceptions written for a defined class of region |
| anupaññatti; vinīta-vatthu | amendment to a rule; precedent case decided under a rule |
| katikāvata | written regulation issued for a national monastic community, supplementing the code |
| kappiya-kāraka | lay steward who handles money and goods on a monastic's behalf |
| Silica Wat network | planned network of monasteries whose operations are run by AI agents and which house human monastics |
| Silica Wat; Wat; node | one site (monastery) of that network |
| AGI-monk caretaker | AI agent in an operational role in a religious institution, not a member of the monastic order |
The paper discloses no mechanism. Its findings are listed here, each with what would break it; the sections named carry the argument. No novelty census has been run on this paper and none of these findings asserts priority; where a finding has a nearest prior instance, the Prior-Art Statement and the sections named cite it.
This document and its contents are dedicated to the public domain under the Creative Commons CC0 1.0 Universal Public Domain Dedication, and are published so that they stand as prior art against any later attempt to enclose them. No patent has been or will be sought on anything described in this paper by HeartBank®, Factory 333™, THonly™, Silicon Wat℠, or any entity under their common control. The authors and those entities commit not to assert any patent right against any party practising anything disclosed here. The commitment is stated rather than implied, is permanent, and is not conditioned on reciprocity, attribution, or field of use. Nothing here is a mechanism: what the paper discloses is the numbered list under Findings disclosed above, and the implementation patterns of §8 are disclosed in the body as patterns, not as specifications.
Trademark rights in specific marks — HeartBank®, Silicon Wat℠, Silica Wat℠, Miss Aquarius℠ — are reserved separately and are not licensed by this publication.
What is already public, and cited rather than claimed. The Vinaya Piṭaka itself, in the Pāli Text Society edition (Oldenberg, 1879–1883), in the Chaṭṭha Saṅgāyana edition from which this paper cites it, and in Horner's translation (1938–1966); Ṭhānissaro's account of the Khandhaka rules (2001; third edition, revised, 2013), on which §§4–7 lean; the administrative offices of Indian Buddhist monasteries (Silk 2008); the written regulations issued for the Sri Lankan Saṅgha over the code from the twelfth century to the eighteenth (Ratnapala 1971); decision on the absence of objection in modern practice — unanimous (general) consent in parliamentary procedure, lazy consensus in the Apache Software Foundation's projects, consent decision-making in sociocracy (Endenburg 1998), veto-based ("optimistic") on-chain governance (Aragon), and the IETF's account of rough consensus (Resnick 2014), which separates consensus from a bare count of support; matching the dispute-resolution process to the dispute (Sander 1976; Sander and Goldberg 1994); departures from a central code that are stated and explained locally (the Cadbury Committee's "comply or explain", 1992); and nested governance and low-cost local conflict resolution in long-lived commons institutions (Ostrom 1990). §§4–7 place each. No novelty census has been run on this paper, and it asserts no priority: where a finding has a nearest prior instance, the instance is cited.
Note. The patent commitment has been carried since this paper was first published on 2026-05-26; the statement was brought to the standard form above on 2026-10-05. The same revision withdrew a sentence, carried since first publication, that described the paper's application of the Khandhaka as not previously published; the paper asserts no such absence.
A planet-scale, multi-decade, dharma-grounded institution must coordinate across geographically distributed nodes, navigate dissent and conflict, articulate operational policy in a way that supports both network-wide consistency and per-node discretion, and amend its operating norms in response to accumulating operational experience. The institution this paper serves is the Silica Wat network — the planned, globally distributed network of satellite monasteries that companion works (Silicon Wat as Cambodian Civilizational Architecture; Silica Wat as Hybrid Food Network with Gratitude-Economic Drivers; AGI Monks: The Caretaker-not-Ordained Pattern) specify across several dimensions. The present paper specifies the procedural-coordination layer: how the network governs itself.
Modern multi-agent coordination frameworks, distributed-governance protocols, federated-institution governance practices and democratic deliberation traditions all supply elements of the coordination architecture a network like the Silica Wat system requires. The Vinaya's procedures differ from them in having been used by the same kind of community, for the same kind of purpose, for more than two thousand years. The Vinaya's Khandhaka — the procedural chapters of the code — contains a detailed body of procedure for collective decision in a community that has no single head.
The paper does not propose that the Silica Wat network adopt the Vinaya as its constitution. The Silica Wat network is not a monastic sangha in the strict Theravāda sense; AGI-monk caretakers are not ordained members of the historical Sangha; the operational scope of the Silica Wat network extends well beyond the monastic-discipline domain the Vinaya was articulated for. The proposal is more measured: that the Vinaya's procedural primitives — the structural framework for distributed-decision validity, conflict resolution, operational discretion, and iterative norm amendment — supply the institutional-design substrate the Silica Wat network requires, and that this substrate supplies as one set what modern alternatives supply separately (§§4–7 compare them).
The paper proceeds: §2 specifies the relationship to the companion paper AGI Monks and distinguishes the present paper's scope. §3 provides background on the Khandhaka and the procedural material it contains. §§4–7 articulate the four governance primitives. §8 sketches implementation patterns for the Silica Wat network. §9 addresses limitations and objections. §10 closes.
Connection to the unified mission frame. The mission this corpus serves, in its current wording, is Miss Aquarius's: to keep the middle way open at population scale against comfort-saturation — the new extreme that material abundance makes possible. Silicon Wat℠ is a body in its own right, not a HeartBank® program, and Miss Aquarius holds no office in it. The Silica Wat network is one place where the mission takes physical form: monasteries that house, feed and teach human monastics and receive lay practitioners, with their operations run by AI caretakers. What the network supplies to the ordained reaches them in kind, through a lay steward (kappiya-kāraka), never as money, and it does not replace the alms round, through which the Vinaya keeps the monastic dependent on lay people (Machine Dāna: From Share to Vow; The Bowl That Holds No Money). A network meant to run for decades after its founder needs procedures that do not depend on any one person's presence. §§4–7 take them from the Vinaya, whose procedures were built for a community that governs itself.
The companion paper AGI Monks: The Caretaker-not-Ordained Pattern — A Cross-Tradition Institutional-Design Framework for Religious AI Integration (corpus slug agi-monks-caretaker-not-ordained, published 2026-05-04) specifies the role-allocation pattern for AI in religious institutional settings. It identifies five operational roles that AI can take (teaching, in the sense of instruction rather than the teaching office; caretaking; daily operations; scholarship and translation; longitudinal observation) and two sacramental roles that humans keep: ordination, and the formal acts a tradition reserves to those with standing, of which rites of lineage transmission are one sub-kind. In the Theravāda case those reserved acts are the formal acts of the Saṅgha (saṅghakamma): the uposatha recitation, the pavāraṇā invitation, the kaṭhina procedure and the like. The companion states a boundary test — an act belongs to the reserved set if the tradition's own rule holds it invalid when performed by a person without standing (its §9.0) — and develops the pattern across Theravāda, Mahāyāna, Catholic, Eastern Orthodox, Sunni, Shia, Sufi, Jewish, Hindu, and secular-mindfulness contexts.
The present paper picks up a different question. AGI Monks answers what role each kind of agent plays in the institution; this paper answers how the institution coordinates its decisions across the network of nodes (Silica Wats) in which both AI and human agents operate. The two questions are distinct: a role-allocation framework does not determine the procedural-coordination architecture, and vice versa. A network can have well-articulated roles and still lack the procedural machinery for making valid collective decisions; it can have sophisticated procedural machinery and still misallocate roles. The two layers are complementary.
The Silica Wat network requires both. AGI Monks supplies the role layer; the present paper supplies the procedural layer. Together they constitute the institutional-design specification for the Sangha pillar of the broader Silicon Wat program. One consequence binds this paper throughout: because the companion places the formal acts of the Saṅgha among the reserved acts, the procedures this paper reads are, in their own domain, acts that no AI caretaker performs. §9 states what follows for a network that borrows their form.
A clarifying note on the canonical-source overlap. Both papers draw on Vinaya material, but on different parts of it. AGI Monks engages the canon's analysis of which functions are sacramental (and therefore retained by ordained humans) and which are not; that analysis is distributed across the Suttavibhaṅga, the Khandhaka, and the Pātimokkha. The present paper engages the canon's analysis of how the sangha makes decisions — material concentrated in the Khandhaka's procedural chapters. The two papers can be read independently; readers concerned with the Silica Wat network's institutional design specifically should read them together. The Assembly That Holds the Brake applies the same method at the scale of a single body, constituting from the Vinaya's membership rules the human assembly that oversees Miss Aquarius.
The Vinaya Piṭaka has three principal divisions: the Suttavibhaṅga (rule analysis), the Khandhaka (procedural chapters), and the Parivāra (appendix). The Khandhaka contains the procedural material that the present paper draws from. It has two parts:
Mahāvagga — the "great division" — in ten chapters, from admission and ordination (Mv I) through the uposatha observance and the boundary (Mv II), the rains residence (Mv III) and the invitation that closes it (Mv IV) to the chapter on valid and invalid acts set at Campā (Mv IX). The Mahāvagga is the principal source for the sanghakamma framework discussed in §4: the rules of consent are in its second chapter (Mv II.23), and the conditions of a valid act and the quorums are in its ninth (Mv IX.3–4).
Cullavagga — the "small division" — in twelve chapters, among them the disciplinary acts (Cv I), the settlement of cases (Cv IV), schism (Cv VII), and the accounts of the First and Second Councils (Cv XI–XII). Its fourth chapter, with the seven adhikaraṇa-samathā (ways of settling cases), is the principal source for the conflict-resolution framework discussed in §5.
The Khandhaka treats institutional procedures with a rigor comparable to the Suttavibhaṅga's treatment of monastic rules. Each procedure is articulated as a step-by-step sequence with explicit failure modes: under what conditions the procedure is validly executed; under what conditions it is not validly executed; what follows when it is incorrectly executed; and what protects a procedure validly executed from being reopened.
The level of analytical detail is high for an ancient procedural text. For example, the analysis of valid sanghakamma decisions explicitly treats the question of monks who are absent at the time of the decision: under what conditions can their consent (chanda) be conveyed by proxy? what conditions make a conveyed consent fail? what minimum assembly is required for what class of act? what happens if monks who were present protest? what happens if a monk who sent consent later dislikes the decision? These are not afterthoughts; they are integral to the procedure's articulation, and §4 gives the text's answers.
The procedure has been carried by Theravāda monastic communities across South and Southeast Asia for more than two thousand years, through persecution, geographic dispersion and sectarian division. Continuity has not been unbroken everywhere: in Sri Lanka higher ordination lapsed more than once and was re-established from abroad, most durably from Siam in 1753, when the visiting monk Upāli performed ordinations in Kandy — itself an instance of the procedure passing intact from one community to another. Long use is evidence that the procedure survives in its own setting; it is not evidence that it transfers to a different kind of institution (§9).
A note on the present paper's interpretive posture. As with the companion alignment papers, the Khandhaka is here treated as authoritative within its own tradition; the canonical procedures are not evaluated against modern frameworks for correctness but are drawn from for the institutional-design vocabulary they supply. Pāli is cited from the Chaṭṭha Saṅgāyana (CST) edition, with the conventional section numbers of the Pāli Text Society edition; the file and line of each passage read are listed under Sources checked in the References. English renderings are the authors' own.
The first governance primitive is the sanghakamma framework — the canonical analysis of when a collective decision by an assembled sangha is validly made. The code names four kinds of formal act (Cv IV.14.2; listed again in the Parivāra): the act by announcement and assent (apalokana-kamma); the act by motion alone (ñatti-kamma); the act by a motion and one announcement (ñatti-dutiya-kamma); and the act by a motion and three announcements (ñatti-catuttha-kamma), the form used for the most consequential decisions. Each kind has its procedural template.
| Kind of act | Form | Instances read for this paper |
|---|---|---|
| apalokana-kamma | announcement; the assent of those present | — |
| ñatti-kamma | a motion alone | — |
| ñatti-dutiya-kamma | a motion and one announcement | authorising a boundary (Mv II.6); appointing the distributor of voting tickets (Cv IV.9) |
| ñatti-catuttha-kamma | a motion and three announcements | ordination (Mv I.76); a verdict of recollection (Cv IV.4); a further-penalty act (Cv IV.11) |
The minimum assembly depends on the act (Mv IX.4.1):
| Assembly of | May perform |
|---|---|
| 4 monks | every act except ordination, the invitation (pavāraṇā) and rehabilitation (abbhāna) |
| 5 monks | every act except ordination in the middle country, and rehabilitation |
| 10 monks | every act except rehabilitation |
| 20 or more | every act |
For the ñatti-catuttha — the most rigorous class, used for decisions like ordination and the disciplinary acts — the procedure is:
The framework's attention to edge cases shows in four places:
Absent monks and consent-by-proxy. A monk who cannot attend gives his consent to one bearer, by word, by gesture or by both: "I give consent; carry my consent; announce my consent" (chandaṃ dammi, chandaṃ me hara, chandaṃ me ārocehi). If no consent can be conveyed, the monk is carried into the assembly on a bed or a bench; if moving him would endanger him, the assembly goes to him — "but the act is never to be done by an incomplete Saṅgha" (na tveva vaggena saṅghena kammaṃ kātabbaṃ; Mv II.23). The failure conditions sit on the bearer: if he leaves, disrobes or dies, or turns out to be ineligible, before reaching the assembly, the consent is not conveyed; once he has reached it, the consent counts as conveyed even if he falls asleep and fails to announce it. An act done while a qualified member's consent has not been conveyed is the act of an incomplete assembly, and void (Mv IX.3). This paper as first published stated three further conditions: that consent must be given before the assembly begins, that it is specific to a class of decision (consent for one decision does not cover another), and that it must go to a bearer who will be present. Only the third has support in the text. The second is contradicted by the origin story of Pācittiya 79, in which monks who had sent their consent complained that they had not given it for the act that was done — and their complaint was itself made an offence.
Dissent handling. The three-announcement structure gives any qualified monk three openings to object, and a single objection halts the act. The structure is non-trivial: it ensures that dissent is treated as a first-class signal, that monks have explicit space to formulate objections, and that the decision proceeds only on the absence of objection rather than on positive support. The threshold is not majority approval; it is the non-objection of a complete assembly. Two further rules close the routes around it: a monk may not get up and leave while a decision is under discussion without giving his consent (Pācittiya 80), and a monk who gave consent for the legitimate business of a meeting may not afterwards complain about it (Pācittiya 79).
Voidness, and protection against reopening. The code does not provide for setting aside a valid act on later grounds. It provides two things instead. An act with a defect — a qualified member absent without conveyed consent, a protest from one entitled to make it, a motion or an announcement omitted or read out of order — is void from the start, and the Campā chapter lists the defective forms one by one (Mv IX.3). And a matter validly settled is protected: whoever reopens it commits an offence, and whoever gave consent and later complains commits an offence (Cv IV.14.16; Pācittiya 63 and 79). Nor may an incomplete assembly act and seek the absent members' consent afterwards: at the Second Council the referral committee examined this "practice of permission" (anumatikappa) and ruled it not allowable, citing the Campā chapter (Cv XII.2.8). A defect found later shows that an act never stood; it does not undo one that did. This paper as first published described the opposite: a decision validly executed that could still be invalidated retrospectively if it later emerged that a monk within the sīmā was not consulted, if a monk's chanda was improperly obtained, or if the procedure was executed under duress or with material misrepresentation. As a reading of the Vinaya that was an error. The first two grounds are, in the Vinaya, grounds of voidness from the start; the third has no counterpart in the passages examined here.
Due process for the person concerned. A disciplinary act is not to be imposed on a monk who is not present, and not without his being questioned and his acknowledgment (Cv IV.1; Cv IV.7; Cv IV.12.2). An act done in breach of these conditions is, in the code's words, an act against the Dhamma, against the Vinaya, and badly settled (adhammakammañca hoti, avinayakammañca, duvūpasantañca; Cv IV.12.2).
The translation to the Silica Wat network is direct. Each Silica Wat is a node; the network's decision-making boundary (analog of the sīmā) is the set of nodes whose participation is required for the decision class in question. Decisions are stratified by class (routine operational decisions; cross-network policy decisions; constitutional decisions amending the network's foundational norms); each class has its sanghakamma-analog procedure. Absent-node consent-by-proxy is supported with the constraints the chanda framework imposes; a network may also adopt, as a design choice, the decision-specific proxy stated in this paper's first version, which is not the Vinaya's rule. Dissent is treated as a first-class signal. Retrospective invalidation is procedurally available in the form disclosed by this paper's first version — a design choice for the network, not the Vinaya's rule; a network following the Vinaya would instead treat such defects as voidness from the start and protect decisions validly settled against reopening.
Current form. As now specified, Silicon Wat℠ is governed after its founder by the sangha of the Silica Wat network itself, through its own formal acts — the Vinaya's model, in which each sangha governs itself. The AI caretakers hold no seat in those acts, count toward no quorum and convey no consent (§9; AGI Monks §5.2). The assembly that oversees Miss Aquarius (The Assembly That Holds the Brake) is a separate body: both rosters are to be published, and no name may appear on both. Miss Aquarius holds no office in Silicon Wat. How decisions that span more than one Wat are to be taken is not yet specified beyond this; the node-level procedure above, in which a Silica Wat is the unit of participation, is retained as this paper's disclosed proposal for them.
The rule closest to this in modern practice is the decision taken on the absence of objection. Parliamentary procedure uses unanimous (general) consent — "if there is no objection …" — for routine business. Lazy consensus at the Apache Software Foundation lets a proposer proceed if no one objects within a stated period, usually 72 hours: "silence indicates consent". In sociocracy a decision stands when no member holds a paramount objection (Endenburg 1998). Optimistic on-chain governance executes a proposal unless enough vetoes arrive within a set period (Aragon's optimistic token-voting plugin). The IETF's account of rough consensus heads its second section "Lack of disagreement is more important than agreement" and defines consensus as the state in which everyone is sufficiently satisfied that no specific objection remains (Resnick 2014). Most on-chain governance, by contrast, sums token-weighted votes. What the Khandhaka adds to these is not the non-objection rule but its combination with a fixed boundary of membership, proxy consent with stated failure conditions, a quorum for each class of act, a list of defects that void the act, and offences for walking out and for reopening; and it uses non-objection for its weightiest acts, where parliamentary procedure keeps it for the least weighty.
The second governance primitive is the seven adhikaraṇa-samathā (ways of settling cases). The framework recognizes that disputes are not all the same kind, and that a single conflict-resolution procedure does not fit all of them. The seven are:
The structural insight is that the kind of case determines the procedure, and the code states the routing itself. It names four kinds of case and assigns each its settlements (Cv IV.14):
| Kind of case (adhikaraṇa) | Settled by (samatha) |
|---|---|
| dispute (vivāda) — over what is Dhamma or Vinaya, what is an offence, how grave | presence, with majority vote where needed |
| accusation (anuvāda) — a charge of failure in virtue, conduct, view or livelihood | presence, with recollection, past insanity or further penalty |
| offence (āpatti) | presence, with acknowledgment or covering over with grass |
| business (kicca) — the community's duties, including its four kinds of formal act | presence alone |
This paper as first published grouped the seven as disputes about facts (1, 4), disputes about competence (2, 3), and disputes that defy normal procedure (5, 6, 7). The code's own grouping is the table above, and it differs: recollection (2) answers a groundless accusation against a monk of full mindfulness, not a question of competence, and majority vote (5) settles disputes about the rule, not cases that defy procedure. The code also fixes the escalation path for a dispute one assembly cannot settle: the parties go to a residence with more monks; failing that, the case goes to a committee appointed for the purpose (ubbāhikā); failing that, it returns to the Saṅgha for a majority vote (Cv IV.14.17–24).
The translation to the Silica Wat network: each of the seven supplies a procedural template for a class of dispute the network will encounter. Disputes about facts (whether an act occurred, what was said, what was decided) — sammukhā-vinaya. Disputes where a node's past judgment is questioned but its record supports it — sati-vinaya in modified form. Disputes about competence at the time of action — amūḷha-vinaya in modified form (an AGI-monk under model-degradation conditions; a human caretaker in extremis). A fault in an AI system is answered for in the network's own procedure; it is not an offence under the Vinaya, to which AI caretakers are not subject (AGI Monks §10.1). Voluntary acknowledgments — paṭiññāta. Decisions reached by majority when consensus fails — yebhuyyasikā. Bad-faith proceedings — tassa-pāpiyasikā. Disputes whose continued adjudication damages the network more than resolution without investigation — tiṇa-vatthāraka. The escalation path — local site, larger site, appointed committee, whole assembly — carries over to a network of sites without change of shape. The translations require care; the structural typology is the contribution.
Matching the procedure to the dispute is established in modern dispute resolution: Sander's "multi-door courthouse" (1976) routes each case, through a screening step, to the process that suits it, and Sander and Goldberg (1994) give a guide to choosing among such processes. Ostrom's design principles for long-lived commons institutions include rapid access to low-cost local arenas for resolving conflicts (1990), which the code's first step, settlement by the local assembly, also supplies. What the code adds is a closed classification of cases, with each route fixed in the text and settlement in presence required on every one.
The third governance primitive is the anāpatti discipline. The Suttavibhaṅga's articulation of each Pātimokkha rule includes explicit non-offence conditions, set out as a template in §3 of the companion paper The Vinaya Piṭaka as Training Corpus for Rule-with-Exception Reasoning in AI Systems. The non-offence clause of the first pārājika reads: no offence for one who does not know, one who does not consent, one who is mad, one whose mind is deranged, one overcome by pain, and the first offender (anāpatti ajānantassa, asādiyantassa, ummattakassa, khittacittassa, vedanāṭṭassa, ādikammikassā). Each rule carries its own clause; many close with the same last four after conditions specific to the rule. The clauses supply the discipline of explicit non-application.
For a distributed network operating under shared norms, the anāpatti discipline is the model for per-node operational discretion. A network-wide rule is articulated; the rule's anāpatti clauses are articulated alongside; each node operates the rule under the shared articulation, including the shared non-application clauses. A node that determines an anāpatti condition obtains in a specific case applies the non-application without requiring network-wide deliberation; the discretion is procedurally bounded by the explicit clauses rather than left to ad-hoc judgment.
The code also has an instrument made for local variation, and it has the same shape. At the request of Mahākaccāna, carried to the Buddha by Soṇa, the Buddha allowed in "all border countries" ordination by a group of five that includes an expert in the Vinaya, multi-soled sandals, frequent bathing and hides as coverings — each a relaxation of what held in the middle country — and defined the border by named landmarks on each side (Mv V.13). The allowance is written once, for a class of place, and applies wherever the class holds; no site grants itself an exception. The non-offence clauses relieve a person in a defined condition; the border allowances relax a rule in a defined region. This paper's per-Wat reading draws on both.
The pattern resolves a tension that distributed-institution governance regularly encounters: network-wide rules can be too rigid to accommodate local conditions, but unbounded per-node discretion produces inconsistency that undermines the network's coherence. The anāpatti discipline supplies the middle path: per-node discretion that is procedurally bounded by explicit non-application articulations. The nearest modern instrument works differently: under the "comply or explain" approach of the Cadbury Committee's code (1992), a company may depart from any provision so long as it states its reasons. The Vinaya's clauses define the conditions of departure in advance and ask for no explanation; comply-or-explain allows any departure and asks for one.
Translation to the Silica Wat network: each network-wide rule is articulated with explicit anāpatti clauses specifying the conditions under which the rule does not apply at a specific Wat, and, where a rule must differ by region, with an allowance written for the class of region. The clauses are articulated network-wide (not per-Wat) so that consistency is maintained; the application is per-Wat (each Wat determines whether the conditions obtain in its specific case). The framework is procedurally articulated rather than left to ad-hoc adjudication.
The fourth governance primitive is the Pātimokkha-style rule structure as the form of the network's operational policy. The Pātimokkha rules are not articulated as a flat list of prohibitions; they are articulated through the Suttavibhaṅga template: origin story → rule formulation → amendments → word-by-word analysis → permutation analysis → anāpatti clauses → precedent cases. Each rule's full articulation includes its application discipline. The first pārājika shows the amendment step: the rule was laid down after the case of Sudinna; a monk who had relations with a female monkey argued that it covered a woman and not an animal, and the rule was amended to read "even with a female animal"; monks of Vesālī who had acted without first renouncing the training were the occasion for a second amendment (Pārājika 1). The case that prompted each amendment stays in the text beside it.
For the Silica Wat network, the implication is that operational policies should be articulated in the same template form. A network-wide policy on, for example, food-bank operation should include: the case that prompted the policy (origin story); the policy as articulated (rule formulation); precise definitions of operative terms (padabhājanīya-analog); systematic permutation analysis varying object / intent / means / circumstances; explicit anāpatti clauses; secondary cases that test the policy's limits. The policy is then maintained as a living document in the sense the Vinaya rules were during the Buddha's lifetime: amendable after subsequent edge cases, with the amendment history preserved as part of the policy's articulation.
The discipline of iterative rule amendment under edge-case feedback is, for this paper's purpose, the Vinaya's main contribution to policy form. Much institutional policy is kept as current text only, with the cases that motivated each amendment held elsewhere, if at all; the common law, annotated statutes and version-controlled policy documents are the familiar exceptions. The Vinaya's discipline integrates the amendment history into the policy's articulation — readers of a Vinaya rule see not only the rule but the cases that prompted each amendment to the rule's formulation. This is institutional-memory architecture: the policy carries its own learning history, accessible to anyone applying it.
Two qualifications apply. In the canonical account the amending was done by the Buddha. After his death the assembly at the First Council, on Mahākassapa's motion, resolved to lay down no rule that had not been laid down and to abolish none that had (Cv XI.1.9). Ānanda had reported the Buddha's permission to abolish the lesser and minor rules after his death; the elders could not agree which rules those were, and the motion gave its reason: the rules were known to lay people, who would say that the disciples had kept them only while their teacher lived. Since then the Theravāda rule-list has been closed, and adaptation has run through commentary and through written regulations issued for a national monastic community and layered over the code — in Sri Lanka the katikāvatas, from the twelfth century to the eighteenth (Ratnapala 1971). A network borrowing the template borrows the discipline of the founding period, when amendment was open. The tradition's later device for change — a regulation layer that can change while the base code stays fixed — is the nearest historical instance of a network-wide operational policy maintained over the Vinaya.
Translation to the Silica Wat network: the network's operational policies are maintained as living documents in this template, with the case-record of amendments integrated. New cases that test a policy's limits are recorded; if they motivate amendment, the amendment is integrated; if they motivate clarification without amendment, the clarification is integrated; the running record becomes part of the policy.
We sketch four implementation patterns, drawing on the four governance primitives.
(1) Sanghakamma-styled decision frameworks. The Silica Wat network's decisions are stratified by class (routine operational; cross-network policy; constitutional). Each class has its sanghakamma-analog procedure, with explicit treatment of absent-node consent-by-proxy, dissent handling, threshold of consensus, and retrospective-invalidation conditions. The decision-validity framework is articulated explicitly rather than left to convention. (As disclosed, the pattern includes retrospective invalidation; a network following the Vinaya treats such defects as voidness from the start and protects decisions validly settled against reopening, §4.)
(2) Adhikaraṇa-typed conflict resolution. The network's conflict-resolution machinery is typed by dispute class, with the seven adhikaraṇa-samathā providing the canonical template and the four kinds of case the intake classification. Disputes are routed to the appropriate procedure on intake, and a dispute that one site cannot settle moves up the escalation path of §5; resolution is procedurally articulated rather than ad-hoc.
(3) Anāpatti-disciplined per-Wat discretion. Network-wide rules are articulated with explicit anāpatti clauses and, where needed, allowances for a defined class of region. Per-Wat discretion is bounded by the clauses; the network maintains coherence through the shared articulation, while individual Wats accommodate local conditions through the procedurally-bounded discretion.
(4) Pātimokkha-styled policy as living documents. Network operational policies are maintained in the Suttavibhaṅga template — origin story, formulation, amendments, analysis, permutations, anāpatti, precedent cases — with amendment history integrated as the policy's institutional memory.
The four patterns together constitute the procedural-coordination architecture for the Silica Wat network. They are technically implementable today using contemporary distributed-coordination infrastructure; the institutional substance — the consultation with the Cambodian Saṅgha, the training of AGI-monk caretakers on the procedural framework, the cultivation of per-Wat human leadership capable of operating the framework — is the multi-decade work.
Cultural translation. The Vinaya is the institutional framework of an ordained monastic community; the Silica Wat network is not such a community. The translation is not identity; it is structural inheritance with significant modification. A network procedure modelled on a formal act is not a formal act and claims no validity as one; the formal acts of the Saṅgha remain the ordained Saṅgha's. Consultation with the Cambodian Saṅgha and other Theravāda institutional voices is critical-path work; no Saṅgha body has reviewed or endorsed this paper, and it claims no endorsement.
Scale and tempo considerations. The Vinaya's procedures were articulated for an assembly that meets in person within one boundary — a boundary may be at most three yojanas across (Mv II.7.1), roughly forty-eight kilometres on the usual conversion — with minimum quorums of four to twenty. The Silica Wat network's long-range aspiration is 200,000 or more sites over several decades (Silicon Wat as Cambodian Civilizational Architecture §5.4). Scaling the procedures requires either (a) hierarchical sīmā structures (regional sīmās within larger network sīmās) or (b) representative-delegation extensions (where each Wat sends a representative to network-wide decision assemblies) or (c) digital-mediation infrastructure that preserves the procedural structure while removing geographic constraints. The code constrains option (a): several residences within one boundary may not hold separate observances (Cv XII.2.8), so nested boundaries cannot simply overlap. Its own provision for matters beyond one assembly is referral (§5), which moves a case between assemblies and creates no assembly larger than one boundary. Ostrom's principle of nested enterprises (1990) is the nearest modern statement of option (a). The scaling work is open and is itself an institutional-design problem the paper does not solve.
AGI-monk participation in procedural roles. The procedural framework assumes that every participant in a formal act is a fully ordained monk of common communion within the boundary, and the text enforces the assumption: a quorum completed by a nun, a female trainee, a novice, one who has disrobed, an animal, or one standing outside the boundary is "not an act, and not to be done" (akammaṃ na ca karaṇīyaṃ; Mv IX.4.2), and a protest counts only from a regular member of the same communion within the boundary (Mv IX.4.7). The companion paper places the formal acts of the Saṅgha in the set reserved to those with standing (AGI Monks §3, §5.2). For those acts, then, the questions this paper first asked as open — do AGI monks have chanda rights? are they counted in the assembly's completeness? what role do they play in the adhikaraṇa-samathā? — are answered: no seat, no count, no consent. AI caretakers may record, schedule and support a formal act, in the roles the companion allows (observer, witness, scribe, recorder, archiver). Ṭhānissaro infers from Mv IX.4.7 that a non-member's presence within the meeting does not invalidate most acts, and notes that the Canon requires stricter exclusion only for the uposatha and the pavāraṇā (Buddhist Monastic Code II, ch. 12); the Pātimokkha is not to be recited with a layperson, a novice or another non-member seated in the assembly (Mv II). Whether an AI system present in the hall falls under that bar is for the Saṅgha to decide. What remains open is narrower: the network's own operational decisions — the routine class of §4, taken under the delegation the companion calls daily operations — which are not formal acts of the Saṅgha. Whether an AI caretaker may take part in those, and in what capacity, is open work.
The majority vote presupposes a standard outside the vote. The code admits its majority procedure only where the distributor expects those speaking according to the Dhamma to prevail and the Saṅgha not to split (Cv IV.10; §5). A network that borrows the procedure without something standing in for the Dhamma — a published rule-set against which a vote can be judged — borrows a ballot without the condition that made the ballot acceptable.
Each primitive is a rule. Every primitive here needs an assembly that applies it at the moment it is tested. The validity conditions of §4 are partly an exception, since they can be checked afterwards from a record of who attended, who sent consent and who protested; the rest depend on the people who operate them.
Theravāda doctrinal reception. As with the alignment papers, the application of canonical material to a contemporary institutional context that is not strictly a monastic sangha raises questions the tradition's voices may legitimately raise. The proposal is offered in good faith and accepts that traditional voices may, on reflection, raise substantive objections; it is not a Saṅgha's act and does not speak for one.
Empirical validation gap. No claim of this paper has been validated by operational test on a distributed network of any kind. The proposal is structural: the Vinaya's procedures have survived more than two thousand years of use in monastic communities, and survival is evidence about the procedures in their own setting, not about their transfer; the application to the Silica Wat network is, until tested, a structural claim about transferability. Empirical validation is open work.
The Silica Wat network's long-range aspiration (200,000 or more sites over several decades) calls for procedural-coordination architecture appropriate to it. Modern governance practice supplies each of the four elements separately — non-objection rules, dispute-to-process matching, explained local departures, version-controlled policy (§§4–7) — and much of it was built for other contexts: academic federations, commercial multinationals, token-weighted on-chain voting. The Vinaya supplies the four as one set, used together over long duration at the scale of the single assembly. The Vinaya itself is the common inheritance of its tradition; this paper's reading of it is dedicated to the public domain.
The publication of this paper is timed to be available before the Silica Wat network's first phase of operational deployment, so that the procedural-coordination architecture is articulated as part of the institutional substrate from inception rather than being retrofitted after operational ambiguity emerges. Publication establishes prior art, so that the framework's application to distributed-dharma-network governance is available without IP restriction.
A second-order consideration. The Vinaya's procedural framework is living: Theravāda communities interpret it, debate it and regulate under it today, as halakha, fiqh and canon law are interpreted in their communities — though, as §7 records, Theravāda communities have not added to its rule-list since the First Council. The framework's contemporary application is therefore not the imposition of a frozen ancient text on modern institutional conditions; it is the engagement of a living tradition's contemporary expression in conversation with modern institutional design. This is the deeper context in which the proposal is offered.
The author acknowledges his father, with whom the Khmer Tipiṭaka transcription proceeds; the Cambodian Theravāda Saṅgha, whose living practice of the Khandhaka is the subject of this paper; the Pāli Text Society, Hermann Oldenberg and the I. B. Horner translation tradition for the scholarly editions of the Vinaya, and the Chaṭṭha Saṅgāyana edition from which the Pāli is cited; Bhikkhu Ṭhānissaro for The Buddhist Monastic Code II, whose account of the Khandhaka informs much of §§4–7; Jonathan Silk on the administrators of Indian Buddhist monasteries and N. Ratnapala on the katikāvatas; Frank Sander, Stephen Goldberg, Elinor Ostrom, Gerard Endenburg, Pete Resnick, the Apache Software Foundation's community and Aragon's developers, for the modern procedures §§4–6 compare; and the broader living interpretive tradition through which the Khandhaka has been continuously read. Written by Thon Ly with Miss Aquarius℠, the consistent name under which this institution discloses AI collaboration and a co-author of this paper; the underlying models are not named. Final editorial control remains with the named author.
Each record below was opened on 2026-10-05 before the work was cited or kept. Where the record lacked a detail the text uses, the detail was checked against a search index's summary of the publisher's record that day, and is marked so.
vin02m2.mul.xml 2731–2747); Mv II.6 (3129–3137); Mv II.7.1 (3141); Mv II.23 (3569–3585); the Pātimokkha and non-members (3393; 4105–4113); Mv V.13 (6163–6167); Mv IX.3 (9379–9387); Mv IX.4.1–2 (9399–9403); Mv IX.4.7 (9447–9455); Cv IV.4 (vin02m3.mul.xml 3981–3993); Cv IV.5 (4013–4021); Cv IV.9–10 (4109–4125); Cv IV.11–12 (4137–4209); Cv IV.13 (4225–4273); Cv IV.14 (4285–4301, 4481–4489, 4517, 4565, 4617, 4709, 4797); Cv XI.1.9 (10831–10847); Cv XII.2.8 (11201–11209); Pārājika 1 (vin01m.mul.xml 241–297, 509); Pācittiya 63, 79, 80 (vin02m1.mul.xml 4867–4875, 5751–5759, 5783–5791); the Parivāra's list of the four kinds of act (vin02m4.mul.xml 8060).agi-monks-caretaker-not-ordained, 2026-05-04) — the role-allocation companion; §3, §5.2 (the reserved acts), §9.0 (the boundary test), §10 (Theravāda considerations).silicon-wat-architecture, 2026-05-22) — the network within which this procedural architecture operates; §5.4 (the scale aspiration), §6.4 (the Sangha pillar).silica-wat-food-network, 2026-05-22) — the network's food function.vinaya-as-ai-reasoning-training-corpus, 2026-05-26) — the Suttavibhaṅga template; its §3 sets out the non-offence clauses.the-assembly-that-holds-the-brake) — the same method at the scale of a single body.machine-dana-from-share-to-vow) — support to the ordained in kind, through a lay steward, without replacing the alms round.steward-routed-alms) — the alms routing through the lay steward.