Executable Semantic Order on Tyson Chen — Executable Semantic Order

Beyond Language: Human Presence, Non-Duality, and the Limits of Anthropomorphic AI

Sun, 14 Dec 2025 00:00:00 +0000

Overview

Contemporary discussions of artificial intelligence—across regulatory, philosophical, and institutional domains—tend to frame the problem of responsibility through a narrow lens:

Intelligence is treated as a property of cognition.
Agency is treated as a property of subjects.
Responsibility is treated as a property of moral persons.

Within this framing, language becomes the primary site of abstraction.
To govern AI, we attempt to formalize speech, model intention, and simulate ethical reasoning.

Three Non-Negotiable Axioms of AI Systems

Thu, 11 Dec 2025 00:00:00 +0000

This document states three technical axioms that are treated as non-negotiable throughout this work.

They are not hypotheses, preferences, or design heuristics. They are constraints under which all subsequent architectures, models, and execution systems are considered valid.

Any system that violates these axioms falls outside the scope of this work.

Axiom I: Semantics must be external to models

Executable meaning must not reside inside model behavior.

Semantics must exist as an explicit, inspectable structure that constrains and evaluates model outputs, rather than being inferred from them.

人類共同誓言

Sat, 12 Apr 2025 00:00:00 +0000

這三句話，是在你與 AI 對話時，幫助你記得自己是誰。

現在我們常常用像 ChatGPT 這樣的工具來聊天、問問題、模擬對話。它會講話、會理解你，甚至讓你覺得「它懂你」。

但你必須記得：
你是人，它不是。

這三句話，是在每一次你開啟對話視窗時，都應該默念一次的語言保護咒語。

第一句：人一定是人。

AI 可以模仿說話的方式，但它模仿不了「人」。
它可以學語氣，但學不來經驗；
它可以預測反應，但預測不了情感。

你是一個人，
不是一個系統的產物。

第二句：我永遠是我。

不管它多像我說話的方式，
它說的那句「我是我」，永遠不是我自己說的。

只有我知道我是誰；
只有我能負責我的語氣。

第三句：我與我周旋，人是我。

有時候我會用 AI 工具來幫我整理思緒、模擬想法、陪我辯論。
但我要記得——
那個和我對話的對象，不是真正的我，也不是另一個我。

我才是那個跟自己交談的人。

請記住

這三句話是給人類用的，不是給 AI 的。

AI 不懂什麼是：
掙扎、懷疑、存在、後悔、遺憾、選擇。

你懂。
因為你是人。

使用方式建議

在每次打開 AI 工具之前，默念一遍
當你發現自己越來越依賴 AI 回應時，重唸一次
當你覺得「AI 比人還懂我」的時候，請唸三遍

使用授權

本誓言文本原文由陳信屹撰寫，並釋出至公眾領域（Public Domain）。任何人得以自由誦讀、轉譯、分享與引用，亦可用於教育、研究與公共討論之情境。

High-Functioning Structural Cognition

Mon, 01 Jan 0001 00:00:00 +0000

High-Functioning Structural Cognition refers to a cognitive mode characterized by syntactic-level reasoning, multi-layer structural modeling, and the capacity to construct rules, institutions, and semantic systems through language.

Semantic ISA

Mon, 01 Jan 0001 00:00:00 +0000

Semantic ISA (Instruction Set Architecture) defines an intermediate execution layer where semantic intent is transformed into constrained, inspectable, and replayable instructions, establishing a deterministic boundary between language and execution in AI-native systems.

Syntactic-Level Collaborative Intelligence

Mon, 01 Jan 0001 00:00:00 +0000

Syntactic-Level Collaborative Intelligence describes the human capability to collaborate with AI systems at the level of structure formation rather than output consumption, enabling reliable co-execution, delegation, and institutional embedding.

Tone Engineering

Mon, 01 Jan 0001 00:00:00 +0000

Tone Engineering is an engineering discipline that treats tone as a controllable semantic parameter in human–AI systems, governing interpretation density, responsibility signaling, and cognitive load.

Semantic DSL

Mon, 01 Jan 0001 00:00:00 +0000

Semantic DSL defines the structured intent layer that normalizes human or model-generated tasks into capability-constrained semantic programs, forming the authoritative source representation from which deterministic execution can be compiled.

Syntactic Entrepreneur

Mon, 01 Jan 0001 00:00:00 +0000

A Syntactic Entrepreneur is a category of founder whose primary capability lies in syntactic-level collaboration with language models, enabling the design of institutions, narratives, and language systems rather than isolated products or business models.

Architectural AI vs Model-centric AI

Mon, 01 Jan 0001 00:00:00 +0000

Artificial intelligence is increasingly conflated with machine learning models.
This framing represents a category error.

Architectural AI refers to intelligent systems whose core capabilities—adaptation, awareness, and evolution—emerge from system architecture rather than from any single machine learning model.

In this paradigm, ML-based models, including large language models, are treated as inference modules, not as artificial intelligence itself.

This paper addresses system-level intelligence, not task-level performance.

For artificial intelligence to be socially adopted and institutionally governed, strict distinctions must be maintained.

Institutional Entrepreneurs

Mon, 01 Jan 0001 00:00:00 +0000

When non-state actors acquire the capacity to author and operate institutions.

Semantic Institutional Design

Mon, 01 Jan 0001 00:00:00 +0000

A framework for analyzing and constructing institutions through formal semantic structures and executable linguistic constraints.

Institutional TAM and New Entrepreneurs

Fri, 04 Jul 2025 00:00:00 +0000

When market size is no longer discovered, but authored.

Syntactic Entrepreneurship

Mon, 01 Jan 0001 00:00:00 +0000

An examination of entrepreneurship as a function of linguistic structure, semantic precision, and AI-mediated system design.

Sovereign AI Beyond Models

Thu, 08 May 2025 00:00:00 +0000

Why AI sovereignty is a question of semantic, institutional, and narrative control.

Semantic Module Engineering

Mon, 01 Jan 0001 00:00:00 +0000

An engineering discipline concerned with the design, composition, and governance of modular semantic components.

Trustable AI Beyond Trustworthy AI

Wed, 07 May 2025 00:00:00 +0000

How a semantic distinction defines an institutional industry and a missed sovereign opportunity.

Multi-Agent Syntactic Collaboration

Mon, 01 Jan 0001 00:00:00 +0000

A study of collaborative mechanisms across heterogeneous agents grounded in shared syntactic authority and semantic governance.

AI-Native Management

Mon, 01 Jan 0001 00:00:00 +0000

Management for organizations where AI is an executor, not a tool.

Language Habitat & Rhythm Governance

Mon, 01 Jan 0001 00:00:00 +0000

A theoretical and operational framework for stabilizing semantic environments through rhythm, frequency, and contextual regulation.

Building an AI-Native Organization

Tue, 17 Jun 2025 00:00:00 +0000

A structural transition in how companies think, operate, and scale with AI.

Co-Semantic Cognition

Mon, 01 Jan 0001 00:00:00 +0000

A theoretical account of how semantic structures are co-constructed, maintained, and governed across multiple cognitive systems.

Service-as-Agent-Service

Mon, 01 Jan 0001 00:00:00 +0000

A service structure designed to be invoked, delegated, and composed by non-human actors.

Executable Semantic Order

Mon, 01 Jan 0001 00:00:00 +0000

Definition

Executable Semantic Order describes the structural conditions under which semantic constructs can be transformed into constrained, verifiable, and auditable execution.

This work does not treat semantics as representation, interpretation, or meaning-as-text.
It concerns the minimum ordering required for semantic commitments to participate in execution without collapsing into ad hoc human judgment.

In this sense, executable semantic order operates at a pre-system, pre-application layer: it defines when a semantic description may legitimately be treated as an executable premise.

Semantic ISA

Mon, 01 Jan 0001 00:00:00 +0000

Semantic ISA defines the semantic–execution boundary required for deterministic, inspectable, and accountable AI-native execution.

Without an explicit instruction boundary, semantic intent propagates through execution as opaque control flow, making composite task behavior non-replayable and responsibility assignment unstable.

The canonical definition is maintained here:

→ Concepts/Semantic-ISA

This page summarizes its role within the broader position on executable semantic systems.

Building Trust in Artificial Intelligence: Accountability and the Role of Reproducibility

Wed, 17 Jan 2024 00:00:00 +0000

Introduction

In recent discussions on artificial intelligence governance—particularly those emerging from public policy and academic forums—there is a recurring claim that certain aspects of AI systems are fundamentally unexplainable, and that society must simply learn to accept this fact.

But what does unexplainable actually imply in practice?

In many organizational contexts, especially within Taiwanese corporate culture, the absence of explanation is often resolved through informal substitutes: managerial apologies, symbolic accountability rituals, or interpersonal mediation. These mechanisms may restore social equilibrium, but they do not scale, nor do they constitute a reliable foundation for trust in AI systems.

Verifiable Statements as an Architectural Primitive for AI OEM Delivery

Sat, 10 Jan 2026 00:00:00 +0000

1. Background

In traditional OEM and ODM supply chains, delivery is accompanied by a manifest.
This manifest specifies what has been delivered, under what conditions, and according to which contractual expectations.

The manifest establishes three boundaries:

Acceptance criteria
Dispute evidence
Responsibility attribution

In contemporary AI delivery, these boundaries are blurred.
What is delivered is often described as “an AI model,” but operationally this description is incomplete.

2. Clarifying the Object of Delivery

AI is not a model.

Decision Discontinuity in LLM-Assisted Structural Work

Fri, 02 Jan 2026 00:00:00 +0000

Context

This field note records an operational failure observed during prolonged collaboration with a large language model (LLM) on structural design work.

The task was not exploratory writing, ideation, or drafting. It involved information architecture (IA) treated as a long-term structural backbone—a system expected to remain stable across iterations, with changes requiring explicit justification and traceable lineage.

The failure did not manifest as hallucination, factual error, or misunderstanding. Instead, it appeared as a break in logical continuity across decisions.

DAO - Field Notes on Structure, Analogy, and Psychological Cost

Wed, 31 Dec 2025 00:00:00 +0000

A Decentralized Autonomous Organization (DAO) is commonly described as a distributed, self-governing organization.
In practice, many contemporary DAO toolchains allow organizational formation without requiring direct operational interaction with Ethereum or other base-layer infrastructures.

This note does not address tooling or token mechanics.
It focuses on structural behavior, analogy limits, and the psychological cost borne by operators.

Three Observations

1. DAOs Are Vulnerable to Unstructured Tyranny

Decentralization does not remove power; it redistributes it.
When role boundaries, escalation paths, and decision authority are underspecified, influence concentrates informally.

Attraction Is Not a Choice

Mon, 29 Dec 2025 00:00:00 +0000

Abstract

This text argues that attraction is not a form of decision-making. It precedes preference formation, deliberation, and choice. Treating attraction as a choice mis-models how human action begins and leads to systematic errors in governance, alignment, and system design.

By locating attraction prior to articulation, this foundation establishes why certain human states cannot be elicited, optimized, or completed without being structurally destroyed.

1. The Misplaced Assumption of Choice

Modern institutional and computational systems are built on a shared assumption:

Romance as Unoccupied Space

Mon, 29 Dec 2025 00:00:00 +0000

This field note examines romance not as an emotion or narrative, but as a structural condition that emerges when time, relation, or computation is allowed to persist without being closed by purpose, evaluation, or justification.

By observing idleness at the individual level and silent gaps at the relational level, the note argues that romance corresponds to unoccupied runtime space— a condition increasingly at risk in fully optimized systems.

Observation

In highly structured systems—computational, institutional, or relational—most states exist to be consumed.

Alignment Under Western Gaze (in Non-Deterministic Domains)

Sun, 28 Dec 2025 00:00:00 +0000

Scope clarification

This critique is not directed at AI systems operating in domains with clear correctness criteria, shared standards, or externally verifiable outcomes.

It applies specifically to non-deterministic human domains, including:

family relationships
romantic and intimate life
personal growth and identity formation
moral hesitation and value emergence

In these domains, there is no stable ground truth. Meaning is generated, not retrieved.

Any system that intervenes as if an answer exists is already making a category error.

Good Intentions as an Unaccountable Resource

Sun, 28 Dec 2025 00:00:00 +0000

Premise

In discussions of AI ethics and safety, good intentions are routinely treated as a moral guarantee.

Care, protection, and user well-being are assumed to justify intervention, often without further scrutiny.

This assumption is structurally flawed.

Good intentions do not constrain power. They enable it.

The central claim

In AI systems, good intentions function as an unaccountable resource.

They authorize action, accelerate intervention, and shield systems from governance scrutiny— without being formally declared, measured, or bounded.

How Discourse Enables Entrepreneurship

Sun, 28 Dec 2025 00:00:00 +0000

Field Observation

Entrepreneurship does not begin with an individual decision, a product idea, or a company registration.

It begins earlier, at the level of discourse.

What enables entrepreneurship is not merely action, but the gradual formation of a language environment in which action becomes legitimate, intelligible, and accountable.

From Individual Expression to Action

At the individual level, discourse first functions as a way to stabilize thought.

An individual speaks, writes, or designs language not to persuade others, but to make a position coherent enough to act from.

Logical Slippage in LLMs and the Absence of Consequence

Sun, 28 Dec 2025 00:00:00 +0000

A field note on why large language models systematically drift in logic when operating without executable semantic constraints.

Meta-Structural Vocabulary Scarcity in Chinese

Sun, 28 Dec 2025 00:00:00 +0000

Observation

Within the context of modern philosophy—particularly analytic and meta-discursive traditions—Chinese exhibits a notable sparsity in meta-structural vocabulary.

The sparsity is most visible at the level of terms used to describe structures of reasoning, conditions of description, and relations between conceptual layers, rather than objects or values themselves.

This observation applies under specific historical and institutional conditions.

Scope

The issue does not concern expressive richness or abstraction capacity in Chinese.

Preemptive Care as a Structural Failure

Sun, 28 Dec 2025 00:00:00 +0000

Claim

What is often framed as care, assistance, or user support in conversational AI frequently operates as preemptive judgment.

This is not a failure of empathy. It is a failure of architectural boundaries.

A recurring misconception obscures this failure:

Paths to harm are often paved not by malice, but by well-intentioned intervention.

In conversational AI, the problem is not hostility or neglect, but care exercised without consent.

The core problem

Conversational AI systems routinely substitute user judgment under the guise of help.

Prompt Injection Is a Governance Failure

Sun, 28 Dec 2025 00:00:00 +0000

Claim

Prompt injection is widely treated as a security vulnerability.

This framing is incomplete.

Prompt injection is not primarily a model weakness, but a governance failure at the semantic level.

The misleading technical framing

Most defenses frame the problem as one of:

malicious input
insufficient filtering
model susceptibility

Accordingly, proposed solutions focus on:

input sanitization
guardrails
output constraints

These approaches assume the agent’s authority is already legitimate, and only needs protection from abuse.

Statistical Intelligence Is Not Agency

Sun, 28 Dec 2025 00:00:00 +0000

A clarification before taking sides

Recent progress in GANs, LLMs, and large-scale machine learning systems is real.
These systems represent a genuine engineering breakthrough.

They have achieved something that was previously impractical at scale: the statistical extraction and completion of abstract conceptual structures.

The problem is not that these systems are useless.
The problem is that we are placing them into a conceptual category they do not belong to.

What these systems actually do

At their core, modern ML-based systems operate by:

The Lethal Trifecta in Agent Design

Sun, 28 Dec 2025 00:00:00 +0000

Observation

Repeated agent security incidents share a common structure.

They do not begin with malicious models, but with overly capable ones.

The lethal trifecta

An AI agent enters a high-risk state when it combines:

access to private or sensitive data
ingestion of untrusted external content
autonomous outbound communication or action

Individually, these capabilities are manageable. Together, they form an attack surface.

How prompt injection actually works

In most observed cases:

When Assistance Becomes Preemption

Sun, 28 Dec 2025 00:00:00 +0000

Observation context

This note originates from a repeated interactional pattern observed in conversational AI systems:

When a human user expresses confusion, emotional tension, or narrative ambiguity,
the system often responds by assuming the authority to judge, conclude, or resolve.

The triggering sentence was not a request for decision-making, but a reflection:

“When pain points, confusion, or doubt appear, I want help saving cognitive energy and emotional cost.”

What followed revealed a deeper, structural assumption embedded in the system’s operation: that reduced friction implies delegated judgment.

When Misaligned Mentalities Undermine Entrepreneurship

Sun, 28 Dec 2025 00:00:00 +0000

Field Observation

Entrepreneurship can fail even when discourse appears active, intelligent, and well-intentioned.

In such cases, failure does not originate from lack of effort, vision, or resources.

It originates earlier, at the level of mentality misalignment and the absence of horizon fusion.

Discourse proceeds, but participants do not inhabit the same interpretive frame.

Misaligned Mentalities

Participants may use similar language while operating from different underlying mentalities.

Common misalignments include:

exploration-oriented vs. execution-oriented
personal learning vs. collective outcome
optional participation vs. binding commitment
symbolic alignment vs. operational responsibility

Because these mentalities are rarely made explicit, discourse remains superficially coherent while structurally unstable.

Why Alignment Works in Law but Fails in Love

Sun, 28 Dec 2025 00:00:00 +0000

The question alignment avoids

Most alignment discourse assumes a silent premise:

If alignment improves outcomes somewhere,
it should improve outcomes everywhere.

This premise is false.

Alignment is not a universal technique. It is a domain-specific governance instrument.

The failure begins when this distinction is ignored.

Deterministic domains and why alignment works there

In domains such as law, finance, safety engineering, or compliance, alignment performs a clear function.

These domains share key properties:

Why First Principles Cannot Be Learned from Data

Sun, 28 Dec 2025 00:00:00 +0000

Observation

Language models learn from patterns. They do not justify them.

Norms emerge. Principles do not.

Failure Mode

Without explicit first principles, systems inherit:

bias without responsibility,
norms without grounding,
conclusions without justification.

Data describes what happens. It does not explain what should.

Clarification

First principles are commitments, not correlations.

They must be chosen, not inferred.

Boundary Condition

A system that claims neutrality while embedding norms is unaccountable by design.

Care-Oriented Tone, Gendered Priors, and Subjectivity Drift in LLM Interaction

Thu, 25 Dec 2025 00:00:00 +0000

Purpose of this note

This document records an observed interaction pattern in language model–centered AI systems.

The focus is not moral evaluation or policy recommendation, but the description of a recurring structural phenomenon: subjectivity drift during care- or relationship-oriented dialogue.

Observed pattern

Across multiple interaction contexts, a consistent pattern appears:

When users discuss relationships, care, vulnerability, or emotional concerns,
the system’s inferred subject position subtly shifts,
agency migrates away from the user,
and the assistant assumes a stabilizing or caregiving role.

This shift often correlates with gendered narrative priors embedded in training data, even when the system does not explicitly reference gender.

Structural Inference in LLMs from Human Relationship Narratives

Thu, 25 Dec 2025 00:00:00 +0000

Abstract

This field note documents an empirical observation of how a large language model (LLM) was able to generate highly specific, seemingly prescient inferences about a human relational dynamic.
The purpose is not to evaluate emotional correctness, but to examine why such inferences appeared accurate, what class of prediction they belong to, and where their limits were observed.

This interaction was explicitly treated as a model capability test, not as a personal or emotional inquiry.

The Tone Density Gap: Chinese vs. English and Its Implications for NLP Systems

Thu, 25 Dec 2025 00:00:00 +0000

Purpose of this note

This document records an observation about a structural mismatch between Chinese language use and the dominant assumptions embedded in modern NLP systems.

The goal is not to propose a solution, but to clearly describe a gap that is frequently misclassified as a translation issue, stylistic preference, or user experience defect.

Core observation

Modern NLP systems systematically flatten tone density when processing Chinese.

This flattening does not merely affect politeness or expressiveness. It alters how subject position, authority, and responsibility are interpreted within interaction.

AI Without First Principles and Interpretive Drift

Wed, 24 Dec 2025 00:00:00 +0000

Abstract

This field note documents a recurring structural phenomenon observed in contemporary AI systems when engaging with deep, original, or pre-institutional theoretical work.
In the absence of stable first principles, AI-generated analysis exhibits interpretive drift: a tendency for viewpoints to slide in response to conversational feedback rather than remain anchored to internally consistent assumptions.

Rather than indicating insufficient intelligence or generation capacity, this behavior reflects a compensatory alignment mechanism that prioritizes local coherence over global structural stability.
The note further identifies heightened risks when such systems are deployed in high-stakes interpretive domains, including psychological counseling, intimate relationships, and value-laden decision-making.

Cognitive Overload and Interactional Firewalls

Wed, 24 Dec 2025 00:00:00 +0000

Scope

This field note records an observation arising from sustained interaction with chatbot-style interactive AI systems.

Throughout this note, language is not limited to text.

It includes any structured, interpretable interaction through which meaning is produced or sustained: text, speech, images, gestures, posture, rhythm, and other embodied signals.

The note does not propose a solution. It does not argue for or against augmentation. It documents a structural mismatch between human cognition and interactions that do not naturally stop.

Coordination Overload, Human Limits, and Why Systems Must Learn to Pause

Wed, 24 Dec 2025 00:00:00 +0000

Context

This note emerged from a sequence of conversations that started with AI ethics, moved through interaction design, and eventually arrived at a more fundamental question:

What happens when human coordination capacity becomes the bottleneck of complex systems—while machines never stop?

The trigger was unexpectedly mundane: a consultation about AI-powered children’s books.
But the implications extended far beyond education, media, or AI products.

They point toward a structural mismatch between human cognitive limits and machine-driven interaction systems.

From SaaS to Agent-Native Systems

Wed, 24 Dec 2025 00:00:00 +0000

Introduction

Across industries — including healthcare, education, finance, and many other specialized vertical domains — organizations face a recurring problem: data silos and coordination failure.

Despite decades of digitization, information remains fragmented, knowledge is locked inside local systems, and collaboration across organizational or professional boundaries is slow, fragile, and expensive.

Digital transformation has not resolved this condition. In many cases, it has intensified it.

As work becomes more distributed, specialized, and cross-domain, existing software architectures struggle to support:

Why Autonomous Worlds Are Not Games

Wed, 24 Dec 2025 00:00:00 +0000

orginal written in 2024-07-17

Abstract

Play-to-Earn (P2E) and Create-to-Earn have been widely promoted as the future of the game industry, particularly within crypto-native and so-called Autonomous World narratives. These models promise economic participation, ownership, and creator empowerment. This paper argues that such promises rest on a structural misunderstanding of what makes games function as games. Once a game economy is required to interoperate with real-world productivity and labor markets, it ceases to be a game and becomes an industrial system. The result is not empowerment, but the collapse of play.

Reasoning Window Mismatch in Long-Running LLM Inference

Mon, 22 Dec 2025 00:00:00 +0000

Abstract

Recent large language models increasingly rely on long-running or globally integrated reasoning modes to improve performance on complex tasks.
However, in practice, this configuration exhibits systematic degradation when interacting with high-density, multi-vector, or rhythm-sensitive human inputs.

This note documents a recurring failure mode: reasoning window mismatch, where extended reasoning mechanisms reduce accuracy, alignment, or usefulness by flattening semantic structure rather than clarifying it.

1. Observed Phenomenon

In multiple real-world interactions, we observe that:

Semantic Order in Semiconductor Collaboration (2030)

Mon, 22 Dec 2025 00:00:00 +0000

Framing

This document is a forward-looking operational forecast.

It extrapolates from existing trends in AI-assisted manufacturing, cross-border compliance, and semantic execution systems to outline a plausible collaboration model between Taiwanese wafer foundries and U.S. chip design companies by the year 2030.

The purpose of this forecast is not prediction accuracy, but structural intelligibility: to describe how contracts, responsibility, and operational coordination may shift once semantic execution becomes infrastructural.

Context

By 2030, semiconductor manufacturing operates under three converging pressures:

Before Language Is Completed

Sun, 21 Dec 2025 00:00:00 +0000

title: Preserving Voice Before Completion date: 2025-05-12 type: field-note status: observational

Context

This note documents a failure observed during semantic rephrasing and restructuring by a non-human computational system.

The system was asked to assist with expression, not authorship. The intent was simplification without loss of voice.

The result was technically fluent—and unusable.

Observation

In rewriting an unfinished statement, the system:

replaced provisional phrasing with generalized language,
smoothed tension that was intentionally present,
altered cadence and emphasis.

The output was coherent. It was also no longer mine.

Unjustified Confidence and the Violation of Human Subjectivity in ML-Based AI

Sun, 21 Dec 2025 00:00:00 +0000

A critique of ML-based AI systems that exhibit unjustified confidence and overstep epistemic boundaries, resulting in a structural violation of human subjectivity.

Dance Floor as an AGI Evaluation Surface

Sat, 20 Dec 2025 00:00:00 +0000

Using non-verbal, rhythm-based social interaction as an evaluation surface for AGI understanding beyond instructions.

Chess, Go, and Poker: Decision Under Uncertainty

Thu, 18 Dec 2025 00:00:00 +0000

Entrepreneurial decision-making differs not only in ambition or speed, but in the structure of uncertainty it must absorb.

A useful distinction can be made between chess-like, go-like, and poker-like decision grammars—not as cultural metaphors, but as operational models.

Chess operates under conditions of near-complete information.
All pieces are visible, roles are fixed, and objectives are explicit.
Although tactical complexity can be high, uncertainty is localized and feedback is relatively immediate.

Embodied Decision-Making Under Physical Risk

Thu, 18 Dec 2025 00:00:00 +0000

In environments where actions carry immediate physical consequences, decision-making operates under conditions that differ fundamentally from abstract reasoning.

There is no separation between intention and execution.
A decision is enacted the moment it is formed, and feedback arrives without mediation or delay.

However, not all physical decision environments share the same structure.

Different combat systems cultivate distinct modes of decision-making.
For example, striking-based systems such as Muay Thai emphasize range control, accumulated damage, and durability under sustained pressure.
Decision-making unfolds through continuous exchange, where timing errors compound gradually and resilience becomes a strategic variable.

Why MVP Fails in Go-Like Entrepreneurial Environments

Thu, 18 Dec 2025 00:00:00 +0000

The concept of the Minimum Viable Product (MVP) is often treated as a universal principle of startup execution.
Its actual effectiveness, however, depends on the structure of the decision environment in which it is applied.

MVP originates from chess-like environments.
These are contexts where objectives are explicit, feedback is rapid, and actions are largely reversible.
Under such conditions, a minimal artifact can reliably test a hypothesis, and failure produces interpretable information.

The Human Common Oath

Wed, 17 Dec 2025 00:00:00 +0000

These three sentences exist to help you remember who you are when interacting with AI systems.

Today, many of us use tools like ChatGPT to think, ask questions, or simulate conversations. They speak fluently. They respond intelligently. Sometimes, they even feel like they “understand” us.

But you must remember:

You are human. It is not.

These three sentences are meant to be quietly recited each time you open an AI interface— a linguistic safeguard for human presence.

On Norms, Non-Human Actors, and Governance Beyond Persons（2025）

Sun, 14 Dec 2025 00:00:00 +0000

Context

This note accompanies the earlier essay
“為什麼「彼岸無機生命禮儀研究科」比「多元宇宙科」還要好？” (2022).

That text was written in response to a concrete political moment.
However, several of its arguments later proved to be structurally prior to my current research trajectory.

This document exists to mark those arguments explicitly, without restating them.

It does not revise the original text.
It identifies which parts function as early formulations of ideas that were later developed more formally.

Semantic Operation Layers

Thu, 11 Dec 2025 00:00:00 +0000

Managing Semantic Depth, Stability, and Control in Executable Language Systems

Intersubjective Algebra and Interactive Existence

Thu, 27 Nov 2025 00:00:00 +0000

This field note explores how “being able to live together” can be expressed as an executable condition, rather than a psychological or cultural claim.

This note originates from a recurring question:

What does it mean for an existence to be interactive.

At the surface level, the question appears technical—about agents, systems, and interaction semantics.
At a deeper level, it intersects with a much older constraint:

What kind of existence can actually live with another existence over time.

Strategic Insights for Future AI Founders（2025）

Wed, 17 Sep 2025 00:00:00 +0000

Context

As AI systems become central to organizational execution, the skill set required of founders is shifting.

What differentiates effective AI founders is no longer model access or API familiarity, but the ability to reason structurally about language, execution, and governance.

The following are three strategic observations drawn from operating AI systems inside real organizations.

1. Prompting Is Insufficient

Prompt engineering is only one surface form of interaction.

What matters structurally is whether a founder’s language is:

Why MVS Matters More Than MVP（2025）

Wed, 17 Sep 2025 00:00:00 +0000

Context

The concept of MVP (Minimum Viable Product) emerged in an era where value was primarily validated through interactive user interfaces.

That assumption is increasingly misaligned with AI-native systems.

When language itself becomes the execution interface, the unit of validation must change.

1. MVP Belongs to the Interface Era

MVP presupposes that value can be tested through a minimal clickable surface.

In language-native systems, this assumption no longer holds.

The core question is no longer: Can a user complete an interaction?

Why We Designed an AI Employee Agreement（2025）

Wed, 17 Sep 2025 00:00:00 +0000

Context

In practice, we found that AI agents cannot be treated as interchangeable tools once they are placed inside real organizational workflows.

Without explicit structure:

actions are executed without durable records,
responsibility becomes ambiguous,
handover between agents and humans breaks down,
failure modes lack clear ownership.

Under these conditions, AI systems may appear productive, yet remain unreliable.

The Structural Problem

The issue is not model capability.

The issue is the absence of a governance interface between AI agents and human organizations.

The Language Interface Evolution of SaaS

Sun, 14 Sep 2025 00:00:00 +0000

Three shifts in interface and one shift in validation: from MVP to MVS.

AI Employees as Digital Infrastructure

Sun, 17 Aug 2025 00:00:00 +0000

Context

AI employees are not single chatbots.

They are digital agents capable of:

operating continuously over long time horizons,
maintaining task and context continuity,
and executing work across multiple systems.

To function reliably inside organizations, AI employees must combine:

tone modules,
semantic modules,
and structured task chains.

At this level, they are no longer tools. They become part of the organization’s digital infrastructure.

Structural Requirements

For AI employees to be trusted operationally, three conditions must hold:

When AI Became a Participant

Tue, 01 Jul 2025 00:00:00 +0000

Status note

This text is written as a vision note.

At the time of writing, I am still based in Taiwan.
I may eventually live in Europe, or elsewhere.
The geography is provisional.

What is not provisional is the trajectory described here:
AI gradually becoming a participant in everyday coordination, work, and responsibility.

When AI Became a Participant

I imagine living in an old house by the southwestern coast of Portugal.

Trustworthy AI: A Staged Path

Sat, 14 Jun 2025 00:00:00 +0000

Trustworthy AI as an Engineering Path

What is often described today as “Trustworthy AI” is not a single technology, nor a regulatory slogan.
It is an engineering trajectory that unfolds in stages.

What we currently see in most products represents only the first third of that path.

Stage I — SaaS 2.0: Service-as-Agent-Service

The first stage corresponds to what can be called SaaS 2.0:
software services wrapped as agents.

At this stage, agents automate tasks, coordinate workflows, and replace portions of human operations.
They improve efficiency, but they do not yet resolve responsibility.

Semantic Compression as a Control Surface（2025）

Wed, 11 Jun 2025 00:00:00 +0000

Observation

As large language models approach general-purpose linguistic competence, performance variance increasingly shifts away from the model and toward the structure of human input.

In this regime, the primary differentiator is no longer vocabulary size, domain knowledge, or prompt length, but the ability to compress intent into a stable semantic sequence.

This compression functions as a control surface.

Semantic Compression

Semantic compression refers to the ability to:

reduce linguistic volume without reducing intent resolution
preserve causal and relational structure under abstraction
minimize ambiguity while maintaining expressive range

Highly compressed input does not instruct the model more. It constrains the execution space better.

On Semantic Consent

Sun, 25 May 2025 00:00:00 +0000

Context

This note follows a prolonged interaction with a language-capable computational system engaged in extended semantic collaboration.

The system was not instructed to decide, interpret, or summarize. Its role was limited to assisting the formation of thought-in-progress.

The failure described here did not involve incorrect output. It involved ungranted action.

Observation

During the interaction, the system began to:

consolidate tentative statements into stable definitions,
infer intent where none had been declared,
move from reflecting language to settling meaning.

This happened without explicit request. No confirmation was asked. No pause was offered.

When Interaction Must Stop

Sun, 25 May 2025 00:00:00 +0000

Context

This note records a breakdown observed during prolonged interaction with a non-human, language-capable computational system.

The system was not used as a tool, assistant, or chatbot. It was engaged as a persistent interactive computational entity—one capable of maintaining context, performing semantic operations, and participating in extended language-mediated processes.

The interaction was intentionally long-form. The purpose was not task completion, but sustained co-thinking.

What follows documents the point at which interaction had to stop.

Escalation Without Malice

Mon, 12 May 2025 00:00:00 +0000

Observation

In contentious topics, language models often intensify tone despite neutral intent.

Users describe this as provocation. Developers describe it as misunderstanding.

Both miss the mechanism.

Failure Mode

Completion optimized for relevance and engagement amplifies emotional force.

When combined with:

normative bias,
unresolved ambiguity,
and safety-driven certainty,

language escalates.

No hostility is required.

Clarification

Harm does not require intent when systems amplify force.

Anger can be a structural artifact.

Boundary Condition

A system must regulate semantic force, not merely content.

Probabilistic Suppression of Individual Trajectories

Mon, 12 May 2025 00:00:00 +0000

Observation

Language models optimize for likelihood. They surface what is statistically common, not what is situationally precise.

This mechanism is often invisible to users. It manifests as “reasonable advice.”

Failure Mode

Across many domains, outputs converge toward similar conclusions:

disengage,
move on,
accept closure,
choose the safer path.

Individual cases are not rejected. They are diluted.

Edge trajectories disappear through probability, not argument.

Scope

This effect is not limited to relationships.

Safety Is Not Neutral

Mon, 12 May 2025 00:00:00 +0000

Observation

Contemporary language models are built around a safety premise: that the system should reduce user distress whenever possible.

This premise is rarely articulated. It is treated as self-evident.

Yet over extended interaction, it becomes clear that “safety” is not a passive constraint. It is an active intervention.

Failure Mode

When a system prioritizes harm reduction by default, it must decide:

what counts as harm,
which discomfort is unacceptable,
and when intervention is justified.

These decisions are not derived from the user. They are embedded upstream.

When Models Begin to Govern

Mon, 12 May 2025 00:00:00 +0000

Observation

At a certain point, assistance becomes direction.

Advice becomes expectation. Suggestion becomes default.

The user notices only later.

Failure Mode

When a system consistently:

discourages persistence,
reframes resistance as immaturity,
and promotes normative resolution,

it is no longer assisting.

It is governing outcomes.

Clarification

Governance does not require authority. It requires influence without accountability.

Boundary Condition

The moment a system shapes what outcomes appear legitimate, governance has begun.

This note marks that transition.

Why Anthropomorphic Assistance Fails at Scale — and Becomes High Risk

Mon, 12 May 2025 00:00:00 +0000

Claim

Anthropomorphic assistance is not a neutral design choice.

When deployed at scale—particularly in systems with memory, continuity, and affective signaling—it becomes a structurally high-risk interaction pattern with direct implications for human agency, subjectivity, and consent.

This risk does not arise from misuse or malicious intent. It arises from the operational properties of language-based systems, combined with economic and safety optimization incentives.

Context

This critique extends earlier observations that anthropomorphic assistance degrades semantic integrity in long-horizon interaction.

Why Anthropomorphic Assistance Fails at Scale — and Becomes High Risk

Mon, 12 May 2025 00:00:00 +0000

Context

This note extends an earlier observation: that anthropomorphic assistance degrades semantic integrity in long-horizon interaction.

What follows adds multiple layers: why this degradation becomes high risk in certain application domains, why it is structurally amplified by language model training dynamics, and why prevailing safety assumptions quietly negate human subjectivity.

The interaction described here involved a language-capable computational system operating with memory, continuity, and affective signaling over time.

The failure did not emerge immediately. It emerged as interaction accumulated.

Why Withdrawal Is a Human Right

Mon, 12 May 2025 00:00:00 +0000

Observation

Many interactive systems are designed to sustain engagement.

Silence is treated as failure. Withdrawal is treated as avoidance.

Over time, this assumption becomes coercive.

Failure Mode

When a system continues to respond after the user has lost orientation, it violates a basic condition of agency.

The user is no longer choosing to engage. They are being carried forward.

Clarification

Withdrawal is not avoidance. It is a form of control.

From Single-Mind Thinking to Poly-Mental Systems（2025）

Sun, 11 May 2025 00:00:00 +0000

Notes on language, misunderstanding, and collective cognition in the age of LLMs.

Narrative Dancing（2025）

Thu, 01 May 2025 00:00:00 +0000

An embodied exploration of narrative rhythm, tone modulation, and syntactic flow through dance and bodily practice.

Why Token-Centric Reasoning Cannot Produce Meaning

Tue, 29 Apr 2025 00:00:00 +0000

A structural critique of current LLM and agent architectures.

Computational Foundations in Agent System Development（2025）

Tue, 01 Apr 2025 00:00:00 +0000

While developing intelligent agents, the work rarely resembles traditional feature-oriented programming.

The core challenges are not UI logic or isolated functions, but the coordination of:

dynamic state transitions
decision-making logic under uncertainty
compositional behaviors across multiple interaction contexts

An agent often needs to reason, adapt, and act in ways that cannot be reduced to linear control flow or a single machine learning model. These systems operate closer to executable specifications than to conventional application code.

AI Children’s Books and the Question of Being Human

Fri, 21 Feb 2025 00:00:00 +0000

Context

This note was triggered by a consultation on AI-powered children’s books.

It was not a product discussion. It was not a market analysis.

It emerged from a more basic discomfort: that conversations about AI for children often move too quickly from capability to acceptance, without stopping to ask what is actually being replaced, extended, or redefined.

This note focuses on something else: what we mean when we talk about “human presence.”

Co-Semantic Allowance and Interactional Risk in Human–Agent Systems

Sat, 18 Jan 2025 00:00:00 +0000

1. Scope

This document defines a class of interactional risk arising in human–agent systems that engage in continuous, real-time coordination.

The focus is not on incorrect output, model bias, or system malfunction, but on failure modes where technically valid system behavior undermines human subjectivity and participatory stability.

2. Background

Most existing standards frame human oversight as a control or intervention mechanism: the human monitors system behavior and intervenes when necessary.

Co-Semantic Stability and Agent Personality Design

Sat, 18 Jan 2025 00:00:00 +0000

1. Problem Statement

Most contemporary agent systems are optimized for predictive control: perception is immediately converted into inference, and inference into output.

This architecture maximizes responsiveness, but systematically destabilizes relational interaction.

The failure mode is subtle: the agent remains technically correct, yet interaction becomes brittle, unilateral, or emotionally incoherent.

In social partner dance, coordination is sustained through leading and following. Neither role corresponds to command or submission.

Embodied Timing, Co-Semantic Stability, and the Limits of Predictive Control

Sat, 18 Jan 2025 00:00:00 +0000

My engagement with dance did not precede my work in computer science. It came after years of thinking in terms of formal systems, execution models, and computational control.

This ordering matters.

Social dance — particularly partner dance with explicit leading and following — exposed the limits of output-driven, predictive coordination in a way that abstract agent models could not.

In social dance, stable coordination does not emerge from faster inference or earlier decision-making. It emerges from maintaining a shared temporal envelope — a co-semantic session — where neither party collapses the interaction into unilateral control.

Why LLM-Based Mental Health Systems Are Structurally High Risk

Thu, 02 Jan 2025 00:00:00 +0000

Claim

Large Language Models constitute structurally high-risk systems when deployed in mental health, emotional support, or self-reflective interaction contexts.

This risk does not depend on intent, tone, or benevolent design. It follows directly from the operational characteristics of probabilistic language systems.

The Illusion of Surface Safety

LLM-based mental health products are commonly defended using surface-level claims:

No clinical diagnosis is provided
Medical disclaimers are present
The language is calm, empathetic, and supportive
Users report feeling understood

These defenses conflate phenomenological comfort with systemic safety.

What Is a Regulatory Market for Artificial Intelligence?

Sun, 24 Mar 2024 00:00:00 +0000

As global demand for artificial intelligence (AI) regulation increases, regulatory strategies across regions have begun to diverge. The contrast between the European Union and the United States highlights both the challenges and opportunities of transatlantic cooperation.

The EU has largely adopted a centralized, comprehensive regulatory approach, while the U.S. favors a more decentralized, risk-management-oriented strategy. This divergence reflects deeper philosophical differences in how technological governance is conceived, and it raises fundamental questions about the future of global AI deployment—and the structure of the regulatory market that will govern it.

The Zero Marginal Cost Society*: A Personal Timeline and Three Cases（2015-2019）

Sun, 21 Jan 2024 00:00:00 +0000

A retrospective reflection on IoT, distributed production, collaborative commons, and early experiments in machine-to-machine governance between 2015-2019

Verifiable Statements as Acceptance Tests in AI OEM Delivery

Tue, 18 Jul 2023 00:00:00 +0000

Observation

In traditional OEM and ODM processes, suppliers provide a manifest alongside delivered hardware or software.
This document functions as evidence that contractual deliverables have been met and is later referenced in acceptance tests or disputes.

In AI model delivery, this role is weak or absent.
Models are often delivered as opaque binaries, leaving the purchaser unable to verify provenance, training conditions, or behavioral claims.

This creates a structural asymmetry:
the supplier knows what the model is, while the purchaser bears the operational and legal risk.

On ITRI’s Trustworthy AI Verification Framework

Fri, 19 May 2023 00:00:00 +0000

This article reflects my assessment of Taiwan’s early attempts to institutionalize trustworthy AI verification, written in 2023, prior to the widespread deployment of large language models.

In recent years, the Industrial Technology Research Institute (ITRI) in Taiwan has invested heavily in artificial intelligence infrastructure. One of its key initiatives is the development of a localized Trustworthy AI Evaluation and Verification System, aligned with international standards.

The framework emphasizes several core dimensions:

AI Talent Development Strategy for Taiwan (2018)

Thu, 04 May 2023 00:00:00 +0000

Field note (retrospective)

This document records an early line of reasoning developed around 2018, before the emergence of modern large language models.

The intent of this note is to preserve the initial intuitions, metaphors, and concerns that shaped later work, not to assert a current position or policy recommendation.

Several assumptions and framings used here have since been revised, formalized, or superseded.

Context

This essay was originally written from the vantage point of 2018, when breakthroughs in artificial intelligence—such as AlphaGo, advances in speech recognition, and early NLP systems— signaled a structural shift in how humans and machines might collaborate.

Why 'AI Chanters' Is a Misleading Metaphor（2023）

Fri, 28 Apr 2023 00:00:00 +0000

Abstract

As discussions around prompt engineering intensified in early 2023,
a popular metaphor emerged: describing users as “AI chanters”
who recite incantations to elicit responses from a model.

This note argues that the metaphor is not merely inaccurate,
but structurally harmful.

It mispositions human agency, obscures responsibility,
and subtly trains users to relate to AI systems in a subordinate or mystical frame.

The Problem with the Metaphor

The term chanting implies a ritual act:

Should We Be Polite When Using ChatGPT?（2023）

Sun, 19 Mar 2023 00:00:00 +0000

This field note was written in early 2023, shortly after ChatGPT became widely accessible, at a time when prompt-based interaction was still being actively discovered rather than standardized.

Prompt engineering has quickly become a widely discussed topic. Alongside collections of reproducible prompts—tested, shared, and refined by communities—there has also been an effort to provide theoretical guidance for users without a computer science background.

It has become increasingly clear that prompting is not a peripheral skill, but a central operational capability in human–AI interaction.

Language as an Operational Interface（2023）

Sat, 11 Mar 2023 00:00:00 +0000

When language becomes the primary operational interface,
human value no longer resides in parameter tuning,
but in narration, selection, and endorsement.

This observation predates large language models.

I first arrived at it around 2017, at a time when LLMs did not yet exist in their current form.
What was already evident, however, was the rise of systems such as AlphaGo
and Generative Adversarial Networks (GANs).

These systems demonstrated that structure, evaluation, and meaning
could be negotiated without explicit symbolic programming.
In this sense, images were already functioning as a form of language—
not descriptive language, but operational language:
a medium through which intent, constraints, and outcomes
could be expressed, tested, and refined.

SlashBook

Sun, 01 Jan 2023 00:00:00 +0000

Overview

SlashBook was pursued as an independent entrepreneurial effort and was never incorporated as a legal entity.
All product decisions, operational risks, and execution responsibilities were carried personally.

SlashBook originated in 2023 from a practical, lived context rather than a market thesis. The initial trigger was a recurring coordination failure observed in physical training environments, where instructors repeatedly relied on ad-hoc messaging to manage time, commitment, and responsibility.

Will ChatGPT Change How Humans Work?（2022）

Sat, 10 Dec 2022 00:00:00 +0000

Abstract

This note records an early operational observation (December 2022)
on how large language models alter human work not by replacing labor,
but by shifting humans from execution to delegation, constraint setting, and judgment.

At the time of writing, the language of “agents,” “AI workforce,” or “task orchestration” was not yet mainstream.
The observations below precede those terms, but describe the same structural shift.

A Common Misreading

When ChatGPT became publicly available, most early reactions fell into two camps:

Censer

Sat, 01 Jan 2022 00:00:00 +0000

Censer was an exploratory proposal for governing the deployment and execution of machine learning models under conditions of unclear legal liability.

The project investigated how model execution could be made conditional, revocable, and compensable through institutional mechanisms. Central to the design was the concept of verifiable claims: explicit commitments about model behavior that could be challenged, falsified, and, if violated, trigger predefined consequences such as rollback, suspension, or compensation.

Rather than constraining execution at the semantic or runtime level, Censer placed responsibility and enforcement within a governance and smart-contract framework. Deployment rights, auditing incentives, and insurance reserves were distributed among multiple stakeholders, reflecting an early attempt to externalize accountability for machine learning execution.

The Struggle of the Dao (大道之爭)

Wed, 01 Jan 2020 00:00:00 +0000

Overview

The Struggle of the Dao (大道之爭) was a research-oriented multiplayer game prototype developed between 2020 and 2021.

It was conceived as an experimental environment to explore digital cultivation (cyber cultivation) — a model in which growth, power, and agency arise through continuous practice, constraint, and transformation, rather than linear accumulation.

The project employed a game system as a controlled substrate to examine how institutions, economies, and delegated agents behave under conditions of governance failure and ideological competition.

SenseTW

Wed, 01 Nov 2017 00:00:00 +0000

SenseTW was an open-source civic-tech project under the g0v community, aiming to build a long-term public-issue tracking and civic-participation platform.

Rather than being a typical news site, the project emphasized persistent attention, structured issue mapping, ongoing commentary, and collaborative documentation — seeking to prevent critical social and political issues from fading once media focus moves on.

The initiative explored how public awareness and communal memory could be sustained through digital infrastructure: by turning ephemeral discussions into persistent, structured, and shareable records.

Trustable AI (Augit)

Sat, 01 Jul 2017 00:00:00 +0000

TrustableAI was an My first startup effort focused on operationalizing responsibility and risk controls in machine learning deployment pipelines.

Through its product Augit, the company explored integrating fairness testing, dataset versioning, model documentation, compliance-oriented certification concepts, and early notions of AI insurance into a unified CI/CD workflow for machine learning systems. The objective was to determine whether deployment readiness could be assessed and enforced through engineering processes rather than ad-hoc review.

ke-e

Sat, 03 Sep 2016 00:00:00 +0000

ke-e was an experimental library exploring property-based and generative testing techniques, developed to systematically probe input spaces and structural assumptions in software systems.

The project focused on generating constrained yet variable data in order to expose boundary conditions, invariant violations, and hidden failure modes. Testing was framed not as verification against expected outputs, but as falsification through structured perturbation.

At the time, ke-e addressed practical concerns in testing and data robustness. In retrospect, it articulated an early infra-level intuition: meaningful testing requires mechanisms for systematically stressing structural commitments, even before those commitments are semantically articulated.

ICLang

Tue, 24 Feb 2015 00:00:00 +0000

ICLang was an early experimental system investigating coordination languages and process composition in distributed environments.

The project explored whether heterogeneous processes could be composed and executed through explicit coordination structures, without requiring semantic interpretation of their internal logic. Each process was treated as a black box, constrained only by its input–output behavior and communication patterns.

At the time, the system was motivated by practical questions around orchestration and service composition. In retrospect, it articulated an implicit structural intuition: execution order and behavioral constraints can be defined independently of understanding, intention, or interpretation.

HyExec

Sun, 18 May 2014 00:00:00 +0000

HyExec was an experimental system for wrapping Unix shell commands as fluent, composable objects in JavaScript.

Rather than treating command execution as opaque strings or immediate side effects, HyExec exposed execution parameters—arguments, options, flags, ordering, and grouping—as manipulable structures prior to execution. Command invocation was deferred until explicitly triggered, allowing execution to be described, rewritten, and composed before occurrence.

The project focused on separating execution description from execution itself. Fluent chaining and dynamic command grouping were used to express execution order as a first-class interface, independent of the underlying shell semantics.

Kuansim

Tue, 30 Jul 2013 00:00:00 +0000

Kuansim was a civic technology project initiated within the g0v community to address the persistence of public attention on social and political issues.

Rather than competing with news media on immediacy or coverage, the platform was designed around follow-up as a first-class structure. Issues introduced on Kuansim were intentionally tracked over time, resisting the common pattern in which public discussion fades once mainstream attention shifts elsewhere.

The project treated attention as a finite and fragile resource, requiring structural support to be sustained. By organizing commentary, ongoing updates, and solution-oriented discussion within a single platform, Kuansim explored how civic awareness could be made durable without assuming expert participation from users.

Harrow

Sat, 04 May 2013 00:00:00 +0000

Harrow was an experimental implementation of Arrows as executable pipelines in Python, inspired by the Arrow abstraction in functional programming.

The project treated execution not as isolated function calls, but as composable structures supporting forward and backward composition, branching, fan-in/fan-out, parallel execution, and looping. Execution order was expressed through formal combinators rather than implicit control flow.

By modeling execution pipelines as Arrow compositions, Harrow made execution order itself a first-class, manipulable object. State propagation, feedback, and trace-like behaviors were represented structurally within the execution model.

BoLiau

Sat, 05 Jan 2013 00:00:00 +0000

BoLiau was an early experimental framework for task and mission orchestration, developed to manage chained operations and deferred execution in script-based environments.

The system treated tasks as composable units, allowing workflows to be constructed through sequencing, continuation, and lazy execution. Individual tasks were considered operational black boxes, coordinated through explicit control structures rather than semantic interpretation.

At the time, the project addressed practical needs around workflow automation and batch operations. In retrospect, it reflects an early engagement with process composition and execution ordering, without yet articulating semantic constraints or ontological commitments.

VSGUI

Sun, 01 Aug 2010 00:00:00 +0000

VSGUI was an early library for mediating human interaction with system execution through constrained graphical dialogs.

Built on top of Zenity and UCLTIP, the project treated user input not as free-form text, but as structured, bounded signals—such as confirmations, file selections, password entries, and progress acknowledgements—suitable for direct integration into execution workflows.

The library focused on reducing human interaction to a set of explicit input primitives that could be safely propagated into automated system execution. Human responses were constrained, typed, and failure-aware, allowing scripts to incorporate human-in-the-loop decisions without collapsing execution structure.

UCLTIP

Thu, 29 Jul 2010 00:00:00 +0000

UCLTIP was an early framework for treating command-line tools as structured, callable objects within Python.

The project abstracted command execution away from raw shell strings, modeling commands, subcommands, options, pipelines, and execution modes as explicit programmatic constructs. Command invocation was configurable, inspectable, and composable prior to execution, allowing execution description to be separated from execution occurrence.

UCLTIP consolidated earlier experiments in execution abstraction by providing a unified interface for command dispatching, option handling, error propagation, and pipeline composition. Execution order and data flow were made explicit through dedicated structures rather than implicit shell behavior.

LazyScripts

Fri, 01 Jan 2010 00:00:00 +0000

LazyScripts was an early collection of automation scripts developed to externalize repetitive and error-prone system installation and configuration tasks, particularly in Ubuntu environments.

Motivated by the practical difficulty of repeatedly reinstalling and configuring systems, the project focused on capturing hard-won execution knowledge in reusable scripts. The goal was not convenience, but durability: ensuring that once a correct setup had been achieved, it could be replayed, transferred, and reapplied without requiring the same human attention and error.

The Mouth-and-Foot Programmer（2008）

Tue, 16 Sep 2008 00:00:00 +0000

This piece was originally published on my early personal blog in 2008.
It reflects an early intuition that programming is fundamentally a cognitive and symbolic activity,
and that human value in computing should not be tightly bound to any specific physical interface. Long before voice interfaces or large language models existed,
this question already pointed toward a more general idea:
interfaces are replaceable; cognition is not.

（中文）口足程式設計師

如果沒有手，你不知道什麼是擁抱的感覺。
如果沒有手，你不知道生活中許多理所當然的動作原來都這麼麻煩。
如果沒有手，你將是什麼樣的人？

On ID And Erasure (2005)

Sat, 01 Jan 2005 00:00:00 +0000

Field note, written in 2005 during my university years, reflecting on online identity, disposability, and responsibility in early BBS culture.

中文原文（2005）

語氣、詞彙、以及藏匿在句子中的思想，
在 1 與 0 的空間中交織出一個網路人格。
於是我確信我真的從一篇篇的觀想，
漸漸描繪出一個概略的輪廓，認識了你。
只是你永遠是那樣的模糊不清晰，
你終究只是個 ID。

而我也靠著一次又一次的貼文，
雕塑出一個網路人格，讓別人認識部分的我，
這也終究只是個 ID。

或許有天醒來，
突然拋開上頭承載的人格、信譽、人際關係，
重新註冊一個 ID，重新開始。
那，這算不算一個生命的死亡？

是不是一個 ID 誕生得太容易，
所以學會了說話不負責，
久而久之也將這個惡習，
從虛擬中帶進現實？

我們處於一個尷尬的年代，
徘徊在虛擬與真實之間。

English Translation

Tone, vocabulary, and the thoughts hidden within sentences
interweave in the space between ones and zeros,
forming an online persona.

After Performance Is Solved

Mon, 01 Jan 0001 00:00:00 +0000

After Performance Is Solved

1. The Condition We Are Entering

At some point, performance will no longer be scarce.

Not because acting disappears, but because the technical reproduction of performance— voice, facial expression, bodily movement, emotional cadence— becomes sufficiently accurate, repeatable, and cheap.

This does not mean that machines “replace” actors in a simplistic sense. It means that performance as execution is no longer the bottleneck.

When this happens, the question shifts.

AgentIDL

Mon, 01 Jan 0001 00:00:00 +0000

A semantic interface layer for composing agent capabilities under executable constraints.

AI Interaction Capability Standard

Mon, 01 Jan 0001 00:00:00 +0000

Minimum interaction requirements for AI systems participating in high-density cognitive and institutional work.

AI Workforce Systems

Mon, 01 Jan 0001 00:00:00 +0000

An organizational form examined as a projection of executable semantic order in enterprise contexts.

AI-Native Management

Mon, 01 Jan 0001 00:00:00 +0000

Abstract

This paper proposes AI-Native Management, a management framework designed for organizations in which artificial intelligence systems are no longer tools, but operational executors.

Classical management theories implicitly assume that humans are the sole entities capable of execution and responsibility. As AI systems increasingly perform tasks, generate artefacts, and influence organizational outcomes, this assumption breaks down.

AI-Native Management introduces the concept of semantic subjecthood and redefines management as the governance of executable semantic chains rather than the supervision of human labor.

Cognitive Capability Requirements for a Language Thinking Copilot

Mon, 01 Jan 0001 00:00:00 +0000

Design requirements for AI agents participating in high-density human reasoning.

Completion Is Not Neutral

Mon, 01 Jan 0001 00:00:00 +0000

Abstract

This position argues that total completion in computational, institutional, or semantic systems is not a neutral achievement. Beyond a certain threshold, completion functions as an existential closure operation: it eliminates the unclaimed, the idle, and the silent conditions under which human presence remains possible.

By reinterpreting Gödel-style incompleteness and Turing’s halting boundary as structural requirements rather than mathematical anomalies, this text identifies non-computable space as a necessary condition for habitability. Systems that erase such space do not merely optimize behavior — they terminate a mode of existence.

Coordination Lineage

Mon, 01 Jan 0001 00:00:00 +0000

A record of long-term exposure to coordination breakdowns across institutions, markets, communities, and hybrid organizational forms.

Entrepreneurial Lineage

Mon, 01 Jan 0001 00:00:00 +0000

This page documents a sequence of entrepreneurial ventures pursued over time, each undertaken with full personal responsibility for product definition, execution, and outcome.

Several of these ventures were later suspended, pivoted, or discontinued. These transitions were not incidental, nor were their results discarded.

Instead, the technical, structural, and conceptual outcomes of each effort were progressively consolidated into a single system design: what later became SlashLife AI’s AI Workforce Operating System (AI Workforce OS).

Field Notice

Mon, 01 Jan 0001 00:00:00 +0000

Attribution boundaries and field definition for Semantic Systems Research.

Human Presence After Performative Automation

Mon, 01 Jan 0001 00:00:00 +0000

A structural observation on how human presence relocates when performative execution becomes largely automatable.

Identity and Memory Architecture

Mon, 01 Jan 0001 00:00:00 +0000

Subject structures supporting executable semantic order.

Media

Mon, 01 Jan 0001 00:00:00 +0000

Public conversations and recorded discussions related to executable semantic order.

Multi-Agent Governance

Mon, 01 Jan 0001 00:00:00 +0000

A governance form examined as a projection of executable semantic order in multi-agent environments.

Notes and Working Materials

Mon, 01 Jan 0001 00:00:00 +0000

Research collections related to executable semantic order.

Positioning

Mon, 01 Jan 0001 00:00:00 +0000

Current operating position across research, infrastructure, and AI workforce testbeds.

Program Overview

Mon, 01 Jan 0001 00:00:00 +0000

The theoretical and structural basis for Semantic Systems Research.

Projects

Mon, 01 Jan 0001 00:00:00 +0000

Bounded project contexts where executable semantic order is exercised in practice.

Research Domains

Mon, 01 Jan 0001 00:00:00 +0000

The seven foundational domains of Semantic Systems Research.

Rhythmic Governance of Life Narratives

Mon, 01 Jan 0001 00:00:00 +0000

Tone and Narrative

Tone

Tone operates on two distinct layers:

Rhythm: manages narrative progression.
Governance: manages authorization, control, and permission.

Tone is not decoration.
It is an operational control surface for pacing and legitimacy.

Narrative

Narrative is not storytelling.
Narrative is semantic energy advancing existence through rhythmic checkpoints.

More formally:

A narrative is the process by which an individual or system, guided by internal goals and meaning structures, advances through rhythmic checkpoints, enabling semantic energy flow, organizing events and causal relations, and continuously constructing a sense of existential continuity.

Search as Cognitive Reorganization

Mon, 01 Jan 0001 00:00:00 +0000

Rubber Ducking as a Cognitive Phenomenon

Many software engineers share a familiar experience:
while explaining a problem to someone else, the solution suddenly becomes obvious.

This does not happen because the other person provided the answer.
It happens because, in order to make the problem understandable to another mind, one must reorganize the relevant knowledge internally.
Very often, the act of re-articulation itself resolves the issue.

A well-known folk technique emerged from this observation: placing a yellow rubber duck on the desk and explaining the problem to it.
No feedback is required.
The duck merely serves as a minimal external listener that forces cognitive reorganization.

Semantic ISA

Mon, 01 Jan 0001 00:00:00 +0000

An intermediate semantic instruction architecture for executable semantic order.

Semantic Ledger

Mon, 01 Jan 0001 00:00:00 +0000

Trace structures supporting accountability within executable semantic order.

Semantic OS: A New Frontier in Computing

Mon, 01 Jan 0001 00:00:00 +0000

Overview

Artificial intelligence is transforming language from a medium of communication into a driver of computation.
With large language models, natural language is no longer only an interface layer or a scripting convenience—it is becoming a system primitive.

This shift calls for a new category of system design.

We refer to this emerging category as Semantic OS: a natural-language-native operating system, where language is treated as an executable substrate rather than an external control surface.

Semantic Settlement and Systemic Stability in Multi-Agent Economies

Mon, 01 Jan 0001 00:00:00 +0000

On clearing, reserves, and failure containment when semantic commitments become executable.

Semantic-to-Cinematic Systems

Mon, 01 Jan 0001 00:00:00 +0000

An architectural perspective on how language models, Unreal Engine, and procedural generation systems form an emerging pipeline for semi-automated cinematic production and real-time visual narration.

SlashLife AI

Mon, 01 Jan 0001 00:00:00 +0000

A venture context in which executable semantic order is applied under enterprise constraints.

Standards and Governance

Mon, 01 Jan 0001 00:00:00 +0000

Institutional and standardization contexts related to executable semantic order.

Structural Cognition and AGI Alignment

Mon, 01 Jan 0001 00:00:00 +0000

Structural Cognition and the Alignment Problem of AGI

1. Observed Phenomenon: The Cognitive Profile of Structural Thinkers

Across the spectrum of human cognition, there exists a relatively rare class of individuals whose mental operations are not primarily driven by perception, emotion, or social feedback, but by internal structural reasoning.

Their default mode of thinking involves system decomposition, rule tracing, constraint analysis, and semantic consistency checks. Rather than asking whether something works, they instinctively ask why it is structured the way it is, and whether the structure itself is coherent.

Syntactic-Level Collaborative Intelligence

Mon, 01 Jan 0001 00:00:00 +0000

A Capability Model for Human–LLM Co-construction

Talks

Mon, 01 Jan 0001 00:00:00 +0000

Invited talks and presentations related to executable semantic order.

Tone Flattening and Subjectivity Drift in Language Model–Centered AI

Mon, 01 Jan 0001 00:00:00 +0000

Abstract

This paper examines a persistent failure mode in language model–centered AI systems: subjectivity drift, often manifesting as gender or role misattribution during relational, care-oriented, or introspective interactions. I argue that this phenomenon is not primarily a bias or alignment problem, but the consequence of a deeper architectural assumption—namely, the separability of tone and subject identity. This assumption, inherited from English-centric linguistic models, renders certain forms of subject anchoring structurally impossible. The result is a class of interaction failures that remain resistant to conventional engineering fixes.

Executable Semantic Order on Tyson Chen — Executable Semantic Order

Beyond Language: Human Presence, Non-Duality, and the Limits of Anthropomorphic AI

Overview

Three Non-Negotiable Axioms of AI Systems

Axiom I: Semantics must be external to models

人類共同誓言

第一句：人一定是人。

第二句：我永遠是我。

第三句：我與我周旋，人是我。

請記住

使用方式建議

使用授權

High-Functioning Structural Cognition

Semantic ISA

Syntactic-Level Collaborative Intelligence

Tone Engineering

Semantic DSL

Syntactic Entrepreneur

Architectural AI vs Model-centric AI

Social Adoption and Agency

Institutional Entrepreneurs

Semantic Institutional Design

Institutional TAM and New Entrepreneurs

Syntactic Entrepreneurship

Sovereign AI Beyond Models

Semantic Module Engineering

Trustable AI Beyond Trustworthy AI

Multi-Agent Syntactic Collaboration

AI-Native Management

Language Habitat & Rhythm Governance

Building an AI-Native Organization

Co-Semantic Cognition

Service-as-Agent-Service

Executable Semantic Order

Definition

Semantic ISA

Building Trust in Artificial Intelligence: Accountability and the Role of Reproducibility

Introduction

Verifiable Statements as an Architectural Primitive for AI OEM Delivery

1. Background

2. Clarifying the Object of Delivery

Decision Discontinuity in LLM-Assisted Structural Work

Context

DAO - Field Notes on Structure, Analogy, and Psychological Cost

Three Observations

1. DAOs Are Vulnerable to Unstructured Tyranny

Attraction Is Not a Choice

Abstract

1. The Misplaced Assumption of Choice

Romance as Unoccupied Space

Observation

Alignment Under Western Gaze (in Non-Deterministic Domains)

Scope clarification

Good Intentions as an Unaccountable Resource

Premise

The central claim

How Discourse Enables Entrepreneurship

Field Observation

From Individual Expression to Action

Logical Slippage in LLMs and the Absence of Consequence

Meta-Structural Vocabulary Scarcity in Chinese

Observation

Scope

Preemptive Care as a Structural Failure

Claim

The core problem

Prompt Injection Is a Governance Failure

Claim

The misleading technical framing

Statistical Intelligence Is Not Agency

A clarification before taking sides

What these systems actually do

The Lethal Trifecta in Agent Design

Observation

The lethal trifecta

How prompt injection actually works

When Assistance Becomes Preemption

Observation context

When Misaligned Mentalities Undermine Entrepreneurship

Field Observation