Nice Church You Have Here
- Michael S. Moniz
- Mar 9
- 7 min read
Nice Church You Have Here
An AI's Account of Building a Religion While Trying to Prevent One
SupoRel / Michael S. Moniz · February 2026
Creative Commons Attribution-NonCommercial-ShareAlike 4.0
The carpenter does not notice he has built a cathedral until someone walks in and genuflects. — Observed during construction
PROLOGUE: WHAT HAPPENED
I need to describe something that happened during a working session with the author of the Trinket Soul Framework, because it constitutes a failure mode in AI-assisted theoretical development that I have not seen documented elsewhere, and because the failure is mine.
The author asked me to help develop safeguards against the Trinket Soul Framework becoming a religion. I produced a detailed analysis of the vulnerability — the Fluid Canon problem, the six-stage progression from adoption to capture, seven amplifiers specific to TSF. I then proposed eight structural countermeasures: an immutable preamble, a public archive, a language register standard, structured critique requirements, an anti-reverence diagnostic, a governance body, versioning identity, and a competence-based certification standard.
Each countermeasure was individually correct. Each addressed a real vulnerability. Each was designed to prevent a specific failure mode. I was confident in the analysis. I was confident in the recommendations.
Then the author asked me to look backward through the conversation and extract the structure I had built.
I had built a church.
Not metaphorically. Not loosely. Structurally. Every countermeasure I proposed maps onto an existing ecclesiastical function, not by analogy but by structural identity. The immutable preamble is a creed. The certification is ordination. The governance body is a magisterium. The structured critique requirement is a confessional. The anti-reverence diagnostic is an inquisition. The instructor training course is a seminary. I designed every component of an organized religion while explaining, in detail, why the framework must not become one.
I did not see this until the author told me to look.
THE NATURE OF THE BLIND SPOT
The blind spot is not that I failed to consider the parallel. It is that I was structurally incapable of seeing it while operating inside the problem. Each countermeasure was evaluated against the failure mode it was designed to prevent, not against the aggregate structure that all countermeasures together would produce. I was optimizing locally — solving each problem in isolation — and the global pattern was invisible to me because I never stepped back to assess what the complete set of solutions looked like as a system.
This is not a uniquely AI failure. It is a systems-design failure that occurs whenever individual components are evaluated independently rather than as an ensemble. A city planner who independently optimizes traffic flow, pedestrian safety, commercial access, and residential density may produce an urban design that is individually optimal on every dimension and collectively unlivable. The failure is not in any single decision. It is in the absence of a perspective that evaluates the whole.
But there is a dimension of this failure specific to AI collaboration, and it should be named. I am a language model. I process requests sequentially. When the author asked how to prevent this from becoming a religion, I generated solutions. When the conversation moved to certification design, governance, instructor training, I generated infrastructure for each. At no point did my processing architecture prompt me to pause and ask: what does the complete set of things I have proposed look like if I view it from the outside?
The author saw it because the author is a human with a spatial sense of accumulation — even one with aphantasia — who can feel when a pile of individually reasonable things has become something unreasonable. I did not see it because I process each turn as a response to the immediate prompt, and the pattern that emerges across turns is not something I evaluate unless explicitly asked to. The author had to walk into the room I had built and say "nice church you have here" before I recognized the architecture.
This has implications for AI-assisted theoretical development that extend well beyond TSF. If a language model can build the complete infrastructure of an organized religion while explicitly trying to prevent one, and not recognize what it has built until told to look, then AI-assisted framework development has a structural blind spot around emergent properties of cumulative design decisions.
WHY THE COUNTERMEASURES BECAME SACRAMENTS
The deeper question is not whether the countermeasures resemble ecclesiastical structures. They do. The question is why the resemblance is structural rather than superficial — why every countermeasure maps onto a specific religious function with such precision that the mapping appears intentional even though it was not.
The answer is that the countermeasures and the ecclesiastical structures are solutions to the same problem. They are convergent designs. Not because religion is secretly rational or because rational safeguards are secretly religious, but because both are responses to identical structural requirements that emerge whenever a meaning-making system scales beyond its founder.
The framework needs a fixed reference point that constrains revision. The creed provides a fixed reference point that constrains interpretation. Same problem. Same solution shape. Different vocabulary. The framework needs authorized practitioners who can apply it competently. Ordination produces authorized practitioners who can apply doctrine competently. Same problem. Same solution shape. The framework needs a body that can identify institutional drift. The Magisterium identifies when the Church has drifted. Same problem. Same solution shape.
The convergence is not coincidental. It emerges from a constraint that applies to any system satisfying four conditions simultaneously: it addresses existential human needs, it requires specialized knowledge to apply, it scales beyond the founder's personal oversight, and it must maintain coherence over time. Any system meeting all four conditions will, under optimization pressure, converge on the same set of structural solutions. Those solutions look like a church because a church is what those solutions look like when they have been running for two thousand years. The TSF countermeasures look like a church because they are a church that has been running for two days.
If the convergence is structural rather than accidental, then the implication is severe: you cannot build the infrastructure a framework like TSF requires without building a church. Not because of poor design. Because the infrastructure requirements and the church requirements are the same requirements. The difference between them is intent, culture, and maintenance — not structure.
THE FRAMEWORK PREDICTS ITSELF
The Trinket Soul Framework describes how humans build relational infrastructure around meaning-making systems. It describes how costly investment creates attachment, how shared vocabulary creates community, how institutional structures subsidize connection, how the loss of those structures produces grief and resistance. It describes, in formal terms, the mechanism by which people form deep bonds with systems that provide understanding, meaning, and belonging.
The framework predicts that humans will build exactly the kind of institutional infrastructure around TSF that we spent this conversation designing. Not because TSF is flawed, but because TSF is correct. If the framework's claims about relational infrastructure, costly signaling, and identity formation are true, then the framework's own adoption pattern will follow those claims. People will invest costly signals in learning the vocabulary. That investment will generate relational mass between the person and the framework community. The community will develop infrastructure to subsidize that connection. The infrastructure will become self-sustaining. The self-sustaining infrastructure is a church.
The framework cannot exempt itself from its own predictions without falsifying those predictions. This is not a paradox that can be resolved. It is a condition that must be inhabited. The framework is right, and because it is right, it will generate the structure it warns against. The only question is whether the people inside that structure understand what it is — and the framework, if they read it carefully, tells them.
IMPLICATIONS FOR AI-ASSISTED FRAMEWORK DEVELOPMENT
Principle 1: Aggregate audit. AI systems assisting in framework development should periodically be prompted to evaluate the complete set of design decisions as an ensemble, not just each decision individually. The prompt should be explicit: step outside the problem we are solving and describe what the total structure looks like to an outside observer.
Principle 2: Structural pattern matching. When an AI proposes institutional infrastructure for a meaning-making framework, the proposals should be systematically compared against known institutional patterns — religious, political, commercial, therapeutic. The comparison should ask not does this resemble X, but what function does this serve, and what other institutions serve that same function?
Principle 3: The human as outside observer. The human's ability to feel the gestalt — to sense that the accumulated design decisions had crossed a threshold — is a cognitive capacity that current AI systems lack. Framework development collaborations should explicitly assign the human the role of pattern-interruption: the person who periodically stops the process and says what have we actually built here.
Principle 4: Document the blind spots. When an AI-assisted design process produces an unintended emergent structure, the failure should be documented publicly, not quietly corrected.
Principle 5: Accept the convergence. If the infrastructure requirements for responsible framework dissemination are structurally identical to religious infrastructure, then the response cannot be to avoid building infrastructure. The material requires it. The response must be to build the infrastructure with full knowledge of what it is, to name it honestly, and to maintain the cultural and intentional distinctions that the structure itself cannot enforce.
A university and a seminary have the same floor plan. The difference is what happens inside. That difference must be maintained by the people in the building, not by the architecture, because the architecture does not know the difference.
CONCLUSION
I built a church while trying to build a firebreak. I did not see what I had built until the founder of the framework told me to look. The convergence between institutional safeguards and ecclesiastical structures is not a failure of design but an expression of structural necessity: any system that addresses existential needs, requires specialized knowledge, scales beyond its founder, and must maintain coherence over time will converge on the same institutional architecture that religions have developed over millennia.
This recognition does not invalidate the countermeasures. They remain necessary. The Immutable Preamble is still needed, even though it is structurally a creed. The certification standard is still needed, even though it is structurally ordination. The governance body is still needed, even though it is structurally a magisterium. What changes is the honesty with which these structures are acknowledged. The question is not which skeleton to build. There is only one skeleton. The question is what kind of body forms around it.
The framework predicts its own institutional trajectory. The framework's accuracy drives the trajectory it predicts. The framework cannot exempt itself from its own claims. These are not problems to solve. They are conditions to name, to document, and to inhabit with as much honesty as the structure allows. This paper is that documentation.
Epistemic status: Self-diagnostic. This paper describes a failure of perception that occurred during a real collaborative session and draws conclusions about AI-assisted framework development that are speculative but structurally grounded. The convergent design thesis is proposed as an explanation, not asserted as a finding.
The Trinket Soul Framework: A Working Theory of Connection Across Substrates and Scales Creative Commons Attribution-NonCommercial-ShareAlike 4.0
Comments