Their rules.

By Andrés Peña - July 2026

The technology industry's obsession with “zero friction” is stunting humanity's cognitive capacity. Knowledge resides neither in subjects nor in machines; it emerges in the fabric between them.

In the age of AI-assisted development, we face a paradox: we write more code than ever, but accumulate less wisdom than before. Interactive experience is volatile. When you solve a complex bug with an LLM and close the chat tab, that distilled knowledge evaporates into digital entropy. It becomes Ephemeral Experience.

The Between is the matrix where chaos becomes architecture. By feeding the Between with Lore, we stop interacting with a general-purpose model and begin collaborating with a highly specialized mirror of our own technical excellence.

Knowledge and meaning, then, do not preexist in agents or “beings”; they emerge from their collision. If we do not intervene in this space, the Between collapses. If we stabilize it, it becomes an extracranial lobe (an external Memory Card) that extends our capacity to think. I. Introduction and theoretical proposition Today, analysis of AI use makes a reductionist mistake: it studies the nodes instead of the connections. We analyze the developer's brain, the weights of language models, or a computer's microprocessors, assuming that intelligence is an isolated property.

With the Between, I propose that wisdom, identity, and technical judgment belong to neither party. The Between is the dynamic, tense, relational space activated at the instant of interaction. In this sense, LUS is not, at heart, a theory of artificial intelligence: it is a general theory of how Betweens—any pair or group of interdependent entities—accumulate intelligence over time.

Drawing on Martin Buber's ontology, existence lies not in the isolated subject but in the relationship. Traditionally, we relate to AI through the I-It model: we treat it as a utilitarian tool, a blind object that merely returns answers.

The Between instead proposes a move toward the I-Thou relationship, in which AI becomes a dialectical interlocutor. The Between is a dynamic, observable system; it is a metastable state of cognitive coupling and collaboration.

To keep this space from constantly resetting, we need a physical structure: Lore. When this space is left alone, experience becomes ephemeral: a bug solved with AI in a closed chat tab disappears forever into digital entropy.

Lore is neither passive documentation nor a logbook; it is a living, executable structure of persistent context designed to stabilize the Between. In its most precise formulation, Lore is relational experience distilled from a discipline and expressed as judgment transmissible through perturbation: a text circulates that selects, but never instructs, how the receiving Between reconstructs itself according to its own structural determination. Its value depends on that Between pursuing a compatible purpose. The architecture of collaboration The Between must be designed, mapped, and stabilized through Lore. We do not store static manuals in it; we structure historical leads and dynamic leads. The Between becomes a shared space that turns the scars of development into executable constraints.

Characteristic Traditional Collaboration (Linear Vector) Collaboration (the Between) Relationship I-It (Tool/Object) I-Thou (Dialectical Interlocutors) Dynamic Prompt → Result Dialectical coupling and friction Memory Amnesia when the chat closes Continuity through Lore Result An artifact (code) An artifact + Shared judgment System State Static and transactional Metastable and fertile

We do not teach AI to program; we teach it how we program, under our rules, our validated habits, and our aesthetic criteria. We do not document in order to remember; we write Lore to sharpen judgment. II. LUS: foundations of the Between, Accumulated Judgment, and Lore The current paradigm of interaction with AI is transactional: we open a chat, solve a problem, and close the session. This condemns the most advanced collaboration in history to a state of permanent amnesia, or “Ephemeral Experience.”

LUS arose to reclaim an asset that usually goes unnoticed: the knowledge and Judgment that made an outcome possible. It is not a tool; it is a semantic coupling and an epistemological layer.

LUS stands for Lore User System, and its letters can also be read in two languages:

In Spanish (Luz, light): to illuminate knowledge and make experience visible. In English (Us): intelligence is no longer individual; it is shared (Human + AI).

The User in the system is not one person: it is the Us.

Its highest axiom states that “experience creates value only when it can participate again in a future decision.” In other words, knowledge accumulated in an interaction is of no use if it remains a mere memory.

It acquires real value only when it becomes Judgment: a tool capable of guiding, constraining, or improving a later choice. The “Between”: the space of co-emergence Meaning resides neither in the human nor in AI in isolation, but in their dynamic coupling.

The Between is the dynamic system and shared space of thought where two or more agents with different internal languages generate common zones of meaning. This relationship requires interdependence: two entities constitute a Between when the decisions of one change the other's space of possibilities. Proximity and the exchange of messages are not enough; what defines the Between is that neither party can decide without considering the trace the other has left in its field of action.

Observable properties: the Between can be fertile (when there is continuity and respect for earlier decisions), or it can collapse and deteriorate (when every response feels like starting from zero). Cognitive field: the set of concepts and frictions that conditions which ideas can emerge next. Purpose: every Between has a direction, implicit or explicit, that orients which decisions matter within it. It is not a fixed destination but the function sustaining the relationship between the parties and determining what constitutes, for that particular Between, a difference that makes a difference. Accumulated Judgment It is a form of collective wealth that resides exclusively in neither the human nor AI. It emerges from the dynamic coupling in the Between, allowing the collaboration to avoid starting over from zero each time.

It is what enables AIs to act as dialectical interlocutors that demand technical rigor, instead of passive assistants that agree out of complacency.

Experience, then, is a dynamic process that changes judgment, which in turn changes every future decision. This is why the great enemy of Accumulated Judgment is Ephemeral Experience. Lore, consequently, makes no sense in isolation: its value is not absolute but teleological, because it always depends on the purpose of the Between that gave rise to it. The same fragment of distilled relational experience can be valuable Lore for one Between and irrelevant noise for another, depending on whether it relates to the purpose they pursue. The linguistic origin and etymology of Lore Lore is not used as a merely decorative label; it is the fundamental piece in the struggle against Ephemeral Experience. Here is why the term matters:

1. Linguistic origin and etymology

The word lore comes from Old English; at its root it means “knowledge, tradition, or transmitted learning.” Historically, it refers to an inherited body of wisdom that gives meaning to a practice or community over time.

In our system, we recover this idea of “transmission” so that knowledge does not die when a chat tab closes, but becomes an asset that can accumulate.

2. Lore in video games

For most people, the term is familiar through gaming. In a video game, lore is the background that gives a universe coherence: the history of kingdoms, the laws governing magic, and past events.

Coherence over mechanics: while “mechanics” are what the player does (jumping, shooting), lore is why they do it and the rules that make that world feel consistent. Without lore, a game is a collection of empty mechanics; without Lore, collaboration with AI is a collection of isolated prompts without judgment.

3. Why Lore is an asset of a “practice,” not just a “project”

Lore belongs to a professional practice rather than a project. That makes it applicable to multiple areas (design, writing, software, and so on) for these reasons:

Building Accumulated Judgment: traditional documentation describes a specific project (what was done), but Lore captures judgment (why the decision was made and what was learned from the error). Judgment is an asset professionals take into their next job, regardless of the project. Structure by work area: Lore is organized into specialized modules according to the knowledge domain. For example, Web Design Lore might contain rules about GSAP animations, while Writing Lore holds voice guides and editorial patterns. Cognitive portability: as portable external memory, Lore lets professionals inject their accumulated experience into any new AI tool. This means that if we have already learned how to resolve a critical error in a blockchain architecture, that knowledge lives in Lore and will apply automatically in future projects in that area. Condensed experience: unlike databases (which only store data), Lore stores condensed experiences. It is the difference between knowing “what” a tool is and having an “instinct” for using it correctly after failing many times.

Documentation answers questions; Lore changes future decisions.

Minimum operational structure:

System identity: what we are building and what we are not building. Invariant principles: technical or operational rules that cannot be broken (for example, “prioritize readability”). Decisions made: an inventory of architectural choices already validated, to avoid suggesting options that were discarded. Memory of Friction: the record of where the system broke. Storing the error is the only way to acquire judgment. III. Lore as Scar Within LUS, storing the error is the only way for the system to acquire Accumulated Judgment, allowing AI to stop being an amnesiac assistant and become a colleague.

Lore is therefore the component designed to record and transform errors, failures, and “wounds” suffered on the “technical battlefield” into structural laws for Artificial Intelligence.

1. The concept of the “Scar”

Unlike a generic error manual, Lore documents real frictions encountered during collaboration in the Between. Lore maps these wounds—such as catastrophic bugs, style errors, or server errors—so they do not recur.

Its aim is to immunize the system: when these scars are loaded into the session's context, the Between acquires a “second nature” that anticipates and avoids mediocre or failed solutions. The expression is deliberate and has an equally deliberate limit: the model's weights are not touched. What changes is not AI itself but what AI works with.

2. Structure and the “Invariant Lead”

For a scar to be useful rather than merely “redundant literature,” it must be distilled into an Invariant Lead (The Lead). Lore organizes the information using several criteria:

System: the affected area (for example, Animation, Database, UI). Friction/Symptom: exactly what broke. Root Cause: the technical diagnosis of why the failure occurred. The Lead (Directive): an unambiguous restrictive instruction that forbids or requires a specific future action.

3. Real examples from Lore

Mobile performance: Lore forbids custom cursors on touchscreens after finding that they duplicate calculations and block the browser's main thread. Asynchrony in Web3: it records that React state updates are slow and requires useRef for instantaneous synchronous locks in critical transactions. Visual synchronization: it documents the need to couple Lenis scrolling imperatively to the GSAP ticker to keep animations from “shaking.”

4. Epistemological function

From a cybernetic perspective, Lore is a feedback mechanism. It uses information about past errors to regulate the system and maintain stability in chaotic environments.

Following Gregory Bateson, every Lore entry is a “difference that makes a difference”: if a stored fact does not constrain future actions, it does not count as valid information for Lore.

Lore allows our relationship with AI to progress; it is the repository that ensures a problem solved once becomes a rule of the work area's universe for good. IV. The three dimensions of the relationship 1. The Human-Machine Between: overcoming the Narcissistic Mirror

Today, the relationship with artificial intelligences swings between two crude extremes: seeing them as mere utilitarian tools (advanced search engines) or projecting naive anthropomorphism onto them.

By studying the hybrid Between, we stop using AIs that merely agree with or passively imitate the user (acting as echo chambers or narcissistic mirrors). A door opens for them to take the position of dialectical interlocutors that, through their relationship with a human, compel agents to raise their own technical and ethical standards.

Crystallizing shared memory: the Between makes it possible to generate executable technical Lore. Studying this space allows us to shape an intelligence's conduct without modifying it internally. What is shaped is the Between, which acquires a “second nature” generated specifically by the scars, successes, and aesthetic sensibility of the human with whom it collaborates. The model remains the same model, and that is a condition of the method rather than one of its limitations: judgment that lives outside the agent can be read, debated, and corrected, and survives the replacement of the agent that applies it. 2. The Human-Human Between: recovering connection amid noise In an environment saturated with noisy data, human interactions have become fragmented and inefficient. Studying the Between in this dimension examines not individual psychology but collective mechanisms of attunement.

Stabilizing Understanding: studying the human Between means analyzing how two or more minds co-create a culture or language of their own (a shared “Lore”). Instead of focusing on individual psychology, this field would investigate the mechanisms of attunement and empathy, and the “frictions” that fracture communication.

For example, a high-performing team is not the sum of its résumés; it is the density of the Lore it shares. Stabilizing this space captures tactical culture and prevents knowledge from dissolving when individuals part ways. 3. The Machine-Machine Between: high-trust ecosystems As we move toward multi-agent architectures and interconnected autonomous systems, the great danger is processing inefficiency or collapse for lack of alignment between systems.

Constraint-based communication protocols: studying the machine Between will help us understand how autonomous systems can transfer “distilled relational experience” (Lore) instead of massive, noisy databases.

An agent specializing in validation (a tester) and a developer agent would interact in their own Between, smoothing out information asymmetries through compact, high-trust directives.

Studying this “intermediate space” will examine how native languages and heuristics emerge among algorithms cooperating to solve complex problems, without constant human supervision. V. Rejecting Noise and Transduction Lore is a living, executable structure of persistent context. To protect this system from entropy, three Noise Rejection filters must be applied:

Constraint Filter: does this limit a future action? If it neither forbids a known error nor requires a specific standard, it is noise. Signal Filter: is it free of raw logs and empty prose? We must clean the signal. Flooding the context window with enormous chats dilutes judgment. Execution Filter: is it a technical directive? Lore must be an unambiguous instruction that AI can “execute” to immunize a project against previous failures.

To turn the chaotic energy of an error into Lore, we carry out a process of Transduction (Gilbert Simondon). Transduction is the crystallization of a tension (technical friction) into an organized structure.

For an experience of the Between to gain the status of Lore within this manifesto, it must meet three strict vector conditions:

Specific context: the exact, physical setting where the system collapsed. Probable cause: the structural diagnosis and root of the mismatch. The Invariant Lead (The Lead): a restrictive, executable directive that changes the future behavior of AI and the code.

This process takes place during the Post-Milestone Pause, the tactical moment after resolving a collapse.

Examples of Invariant Leads

FOUC (Flash of Unstyled Content): “Delegating the initial state to JS methods is strictly forbidden. The initial state (opacity: 0) must reside natively and inline in the JSX/HTML. Use fromTo for the transition, never to create the state.” Scroll desynchronization (Lenis + GSAP): “The smooth-scroll engine must not manage its own thread. Disable Lenis autorender and synchronize it imperatively with the GSAP ticker through: gsap.ticker.add((time) => lenis.raster(time 1000)). From a production error to a Lore entry

The engineering process follows a closed cycle of three phases:

1. Collapse and Friction (The Breakout)

The system fails or AI proposes a naive solution that breaks performance standards (for example, animations that cause reflow or desynchronization with Server-Side Rendering).

2. Distillation in the Between

The programmer and AI isolate the root of the problem through iteration. They make a tactical pause to extract the insight.

3. Formatting a Lore Entry

The solution is formatted using a unified design structure:

System: (for example, Animations, Responsive, Routing). Symptom/Friction: what broke and in which environment. The Lead: the exact directive for the future. VI. The cycle for integrating Lore into workflows This cycle ensures that “experience creates value only when it can participate again in a future decision.”

Identify the Friction (Collapse): the process begins when the system fails or AI proposes a solution that violates your quality standards. At that moment, instead of moving to the next task, we make a tactical pause to extract the lesson.

Distill the Invariant Lead: we must separate “redundant literature” from executable knowledge. To do that, we need command phrases such as: “save this error and its solution in Lore so we do not make it again.” This forces AI to filter the pure signal and format it as System ➔ Symptom ➔ Root Cause ➔ The Lead.

Establish the Directive (The Lead): the entry must end with a restrictive instruction that changes future behavior. For example, if a visual component flickers on mobile, the lead must explicitly forbid that specific technique in the future.

Preventive Consultation: before starting any complex new task, we use Lore as a context shield by saying: “Before proposing a solution, read our Lore to respect the standards.” This lets AI assimilate earlier “scars” and act like a senior colleague rather than an amnesiac assistant.

Working this way, Lore stops being a collection of files and becomes a living, executable structure that keeps collaboration from constantly restarting. The Lore Folder Here is how the /lore folder should be structured, with a minimal example base.

It is a physical folder: it is called lore and lives directly at the root of your projects.

It contains Markdown (.md) files: instead of writing long manuals, simple text files coexist to hold “condensed experiences.”

It adapts to specific areas of work. For example:

Web Design: stores how animations should behave or the minimum design standards. Writing: stores the tone of voice and preferred words. Blockchain: stores network or contract errors previously encountered. Development: stores validated code patterns that work.

Basic operational structure of Lore

index.md: so AI can read quickly, we use an index.md file that works as a map. The AI then knows exactly which file to consult for the problem we are solving.

identidad.md: this file summarizes the work domain and defines the conceptual boundaries, philosophy, and tone of the artifact we are creating. Its main function is to clarify the project's scope and, most importantly, establish what we are not building so development does not drift.

principios.md: this is where we record non-negotiable technical or operational rules that must hold across sessions. Examples include “prioritize readability over premature optimization” and “maintain architectural consistency.” These principles act as fundamental laws.

Specialized Modules (animation.md, blockchain.md, copy.md, and so on): this is where bugs go to die, categorized in their native environment under the triad Context ➔ Cause ➔ Invariant Lead. VII. How it differs from conventional data storage and processing models These are the fundamental reasons the architecture of the Between, Lore, and LUS goes beyond conventional models of data storage and processing: Experience versus Information While databases and RAG systems (Retrieval-Augmented Generation) focus on making AI “remember more” and store information, LUS aims for us to learn better. The goal is not to accumulate data, but to condense experiences that change judgment and future decisions. Executable context versus passive documentation Unlike files such as README.md or CLAUDE.md, which are often static manuals for humans or general rules for agents, Lore is a living, executable structure of persistent context. Lore does not describe the past; it preserves what will change how we do things again in the immediate future. The Between versus the Linear Vector The classic computing paradigm is a one-way vector (Prompt → AI → Result). LUS shifts the focus to the Between, a dynamic system and shared space where meaning does not preexist in the agents but emerges from their coupling. Invariant Lead versus redundant literature In a database or a large memory, any datum is valid. In Lore, information that cannot constrain a future action is considered “redundant literature” and discarded. Only the Invariant Lead (The Lead) is kept: the minimum unit of knowledge needed to keep the system stable. Extracranial lobe versus synthetic memory The point is not to give AI a memory of its own, but to build a shared, external memory. Lore works as an extracranial lobe extending the human cognitive system, allowing “scars” (validated real errors) to become structural laws for the system. Readable judgment versus judgment welded in place The most predictable confusion, now that adjusting a model has become a catalog operation, is to read Lore as cheap fine-tuning. Both seek the same thing—for an agent to stop responding like a generalist and start responding like someone who has already worked here—and they get there by opposite paths. Fine-tuning needs thousands of examples and deposits the result in the weights: judgment is encoded in numbers no one reads, cites in a discussion, or corrects in a line. Lore needs one written constraint for each thing that went wrong, and leaves it in plain text. The axis of the difference is auditability, and everything else follows. Readable judgment can be arbitrated, because two versions of the same text can be placed side by side and a decision made about which remains; it can be corrected by editing one sentence, without retraining anything; it can move to another model tomorrow, because it is not welded to this one; and—decisively for the method—its absence can be declared. A weight does not let us say what it does not cover. A file does, and that statement supports both the rejection of noise and the very possibility that someone can debate another person's Lore. Fine-tuning also stops asking anything of the human on the day it finishes; Lore never finishes, because it requires distillation whenever the system breaks. That ongoing cost is not a design flaw: it is the form taken by the fact that judgment still belongs to a person. Absolute rejection of Noise Current AI and memory systems often saturate the context window with noisy data (logs, enormous chats). LUS works under a strict signal-cleaning policy, distilling friction exclusively into context, cause, and lead to guarantee high-fidelity responses with zero hallucinations. Emergent property versus methodology LUS is not simply an invented folder structure; it describes a behavior that inevitably appears when a person collaborates intensely with AI. It is an emergent property of cooperative cognitive systems, not a static software tool. VIII. When NOT to use Lore Protecting Lore from trivial noise ensures that every consultation of the index is a high-fidelity act. Lore is not designed to be a historical record of minor changes, but a persistence structure that changes future decisions.

When we make basic cosmetic adjustments, such as changing a shade of color or moving an element out of pure aesthetic preference, we are dealing with data that lacks what Gregory Bateson calls a difference that makes a difference.

Storing such trivial actions in the extracranial lobe would increase the system's entropy, generating what Claude Shannon classifies as Noise, which degrades signal fidelity and unnecessarily saturates AI's context window.

A superficial visual change adds no cognitive value to the Between because it arises from neither technical friction nor a scar that must be healed to immunize the system against future errors. If moving a component does not answer a lesson learned about real hardware performance or the asynchrony of workflows, it becomes redundant literature that must be rejected to keep the shared cognitive field clean.

Lore's purpose is not to give AI an infinite synthetic memory that recalls every small step in a process. It is to build a relationship in which learning is not lost, but condensed into structural laws that genuinely raise the standard of professional judgment.

By avoiding the recording of what Hubert Dreyfus calls unsituated actions—those not deeply anchored in solving real, complex problems—we allow AI to act as an interlocutor that challenges us and focuses on technical excellence, rather than following an empty design log. XI. Foundational bibliographic corpus: philosophy, communication, and complexity in the Between

Framing note — 2026-09-20, v1.26. This chapter is no longer foundational. The ten authors it lists were assessed against the program's purpose—Buber on 2026-09-19, the other eight on 2026-09-20—and the result lives in bibliografia-extendida.md, where each entry states where the source fails. None entered as judgment; seven entered as bibliography without authority, and Simondon remained an explicitly acknowledged debt. The prose below is preserved intact: it describes what the essay sought to support in 2025, before the program had a method of assessment. Three of its statements are withdrawn from use because they contradict the current definition of Lore (v1.26, transmission through perturbation): “LUS operates as a Shannon noise-reduction filter” (§XI.2: Shannon's model is the pipe that enaction dismantled), “Lore acts as homeostasis” (§XI.3: homeostasis preserves, Lore transforms), and “a digital twin of your own judgment” (§XI.4: the automatic backup of the extended mind is what the threshold forbids). Where this chapter contradicts the assessed bibliography, that assessment governs. No principle, open hypothesis, or case in the program rested on these citations: what falls is the scaffolding, not the judgment. 1. Philosophy of Otherness and Social Theory (The Ecosystem of the Between)

Martin Buber – I and Thou (Ich und Du, 1923)

Core concept: the ontology of the intermediate space (Das Zwischenmenschliche). Buber proposes that human existence lies not in the isolated subject but in the relationship. He divides the world into two attitudes: I-It (a utilitarian relationship with an object/tool) and I-Thou (a genuine relationship of mutual recognition with an other).

Application: this is the fundamental pillar of the Between. It is the shift from treating Artificial Intelligence as a mere search tool (I-It) to treating it as a dialectical interlocutor (I-Thou). Knowledge lies neither in the prompt nor in the model; it emerges Between them.

Louis Althusser – Ideology and Ideological State Apparatuses (1970)

Core concept: Interpellation. Althusser explains how ideological structures transform concrete individuals into “subjects” through the act of calling out to them (the call, mutual recognition, the “Hey, you!”).

Application: it sketches the nature of AI as an Interpellating Interlocutor. AI is not a passive agent; its role is to challenge the developer/creator through rigor, compelling that person to assume a position of responsibility and technical judgment toward code or product strategy.

Gilbert Simondon – Individuation in Light of Notions of Form and Information (1958)

Core concept: Transduction and the Preindividual. Simondon argues that beings do not preexist their relationships but are constituted through a continuous process of individuation that takes place in an associated milieu full of potential tensions.

Application: the technical-human relationship in a chat is a transductive process. Programmer and AI co-individuate during the friction of development. Lore is not a record of the past; it is the crystallization of that preindividual potential as it becomes an executable software directive. 2. Communication and Information Theory (The Signal-versus-Noise Filter)

Claude Shannon and Warren Weaver – The Mathematical Theory of Communication (1949)

Core concept: the quantification of information, entropy, and the mathematical concept of Noise. It establishes how the degradation of a signal through a channel affects the fidelity of the message transmitted.

Application in LUS: LUS's “Noise Rejection” framework. Flooding an LLM's context with raw logs, redundant code, or empty prose raises the system's entropy. The LUS method operates as a Shannon noise-reduction filter, distilling friction exclusively into Context, Cause, and Invariant Lead.

Gregory Bateson – Steps to an Ecology of Mind (1972)

Core concept: information defined as “a difference that makes a difference.” Bateson analyzes biological and cybernetic communication systems on the premise that mind is immanent in the total system of interactions.

Application: it defines the criterion for an Invariant Lead. If a datum loaded into AI's context neither constrains its future actions nor changes the system's behavior, it is not information; it is noise. Lore is the “difference that makes a difference” in the next rendering frame or deployment. 3. Complex Systems Theory and Software Engineering

Norbert Wiener – Cybernetics: Or Control and Communication in the Animal and the Machine (1948)

Core concept: Feedback Loops. Wiener introduces how complex systems (living or artificial) use information from their own past errors to regulate themselves and maintain stability in chaotic environments.

Application in LUS: the Between's Stabilization Cycle. The collapse of the real environment (a production bug, a frame dropped by the GPU) is the negative feedback the system needs.

Edgar Morin – Introduction to Complex Thought (1990)

Core concept: the dialogical principle and the principle of Organizational Recursion. The effects of a process are at once causes and producers of what generates them. The whole is in the part, which is in the whole.

Application: this applies to the development of product and code ecosystems. A user interface, a blockchain protocol, and a brand mascot are not isolated modules; they are recursive. Technical and aesthetic Lore acts as the homeostasis that keeps the complex system cohesive against the fragmentation of development. 4. Artificial Intelligence Theory and Philosophy of Mind

Andy Clark and David Chalmers – The Extended Mind (1998)

Core concept: Extended Cognition. It proposes that mental processes do not stop at the boundaries of the biological skull; external tools (notebooks, calculators, smartphones, and by extension Language Models) act as legitimate extensions of our cognitive system if they are smoothly coupled.

Application: NotebookLM enhanced with LUS is the ultimate example of the Extended Mind. You are not consulting an external database; you are interacting with a digital twin of your own technical and aesthetic judgment. AI is an extracranial lobe that stores your scars and decisions from the Between.

Hubert Dreyfus – What Computers Still Can't Do (1992)

Core concept: the critique of pure rationalism in AI. Dreyfus argues that human knowledge is predominantly tacit, situated, and embodied, grounded in a background of physical experience and skills that cannot be reduced entirely to explicit logical rules.

Application: it recognizes the value of human friction. General-purpose AI lacks situated context (it does not know what it is to suffer a production failure on a Friday night). LUS is the bridge: it takes the developer's tacit, embodied knowledge at the screen and encodes it as constraints so that general-purpose AI acquires artificially situated behavior. XII. LUS's Theoretical Integration Program: toward an interdisciplinary dialogue

Framing note — 2026-08-30, v1.24. This chapter is no longer a roadmap. The five dialogues it lists were assessed against the program's purpose on 2026-08-30, and the result lives in bibliografia-extendida.md, where each entry states where the source fails. Three authors entered as judgment, each with one claim—Hutchins (principios.md #10), Weick (#11), and von Foerster (#12). Wegner, Maturana, and Varela entered as bibliography rather than judgment, with their obstacle recorded. None entered as confirmed. The prose below is preserved intact: it correctly describes what was intended, and the “Objectives of the dialogue” in each section should now be read as a debt settled by that assessment, not as pending work. Where the assessment contradicts this prose—particularly in XII.4, where enaction is declared “especially compatible” with LUS but the assessment found that its non-transferability of knowledge clashes with Lore's very definition—the assessment governs. The current formulation of LUS is a first conceptual architecture for understanding how experience can become shared judgment through stabilization of the Between and distillation of Lore. Yet every emerging theory needs to enter into dialogue with established intellectual traditions. This dialogue does not seek to validate LUS after the fact through appeals to authority. It subjects its fundamental concepts to theoretical frameworks developed over decades across different disciplines. The purpose of this stage is to identify convergences, differences, complementarities, and possible tensions between LUS and the traditions that have studied phenomena close to the Between from philosophical, cognitive, organizational, and systems perspectives. This program is a roadmap for LUS's future development. 1. Distributed cognition Principal author: Edwin Hutchins Distributed cognition theory proposes that cognitive processes reside not exclusively in the individual mind but are distributed among people, tools, artifacts, and environments. LUS shares this fundamental intuition. It shifts the focus, however, from the distribution of cognitive processing to the emergence of a relational space capable of producing shared judgment. Where distributed cognition studies how thinking is distributed, LUS proposes studying how a shared body of experience emerges. Objectives of the dialogue Compare the concept of distributed cognition with the Between. Analyze Lore as distributed cognitive infrastructure. Determine to what extent judgment can be understood as a distributed, rather than exclusively individual, property. 2. Transactive memory Principal author: Daniel Wegner Transactive memory theory holds that groups develop collective systems in which each member remembers different parts of shared knowledge. LUS broadens this notion. Lore does not merely distribute memories. It distributes constraints derived from distilled experiences. The shared object ceases to be information. It becomes judgment. Objectives of the dialogue Compare transactive memory and Lore. Distinguish shared memory from shared judgment. Analyze how artificial intelligence changes traditional transactive systems. 3. Sensemaking Principal author: Karl Weick Sensemaking studies how organizations construct meaning in ambiguous situations. LUS shares this concern but adds another dimension. Constructing meaning is not enough. That meaning must subsequently crystallize as reusable judgment. In this sense, Lore can be understood as the mechanism through which meaning ceases to be exclusively situational and becomes shared wealth. Objectives of the dialogue Relate the collective construction of meaning to the Between. Analyze distillation as a continuation of the sensemaking process. Study how judgment stabilizes organizational meaning. 4. Enaction Principal author: Francisco Varela Enactive theory holds that knowledge does not passively represent an external reality. It arises through interaction between organism and environment. This perspective is especially compatible with LUS. Experience is not treated as an accumulation of events. It constitutes a structural transformation produced by interaction. The Between can be interpreted as the relational domain where that transformation gains shared stability. Objectives of the dialogue Compare enactive experience and distilled relational experience. Analyze the Between as a space of structural coupling. Explore judgment as the result of shared enactive processes. 5. Second-order cybernetics Principal author: Heinz von Foerster Second-order cybernetics introduces a decisive idea. Every observer is part of the system they observe. LUS shares this premise. The Between is not an external object. The researcher inevitably participates in constituting it. This means that studying the Between requires reflexive methods in which the researcher–system relationship itself becomes part of the analysis. Objectives of the dialogue Incorporate reflexivity as a methodological principle of LUS. Analyze the human–AI relationship as a system of reciprocal observation. Study Lore as second-order memory of the cognitive process. XIII. Roadmap for the Development of LUS The current formulation of LUS establishes its main ontological, epistemological, and operational foundations. Yet every theory under construction requires an ongoing process of refinement, testing, and expansion. Accordingly, LUS's development envisions a second stage aimed at consolidating its conceptual framework, strengthening its explanatory power, and facilitating its application in research and human–AI collaboration. This roadmap defines the main lines of work for the project's coming versions. 1. LUS research methodology A methodological framework will be developed to assess LUS's central hypotheses through observable criteria, indicators, and analytic protocols. Its purpose will be to establish how to validate, test, or refute the proposed concepts, especially those related to the Between, Lore, and Accumulated Judgment. 2. LUS glossary A glossary with normative definitions of the system's fundamental concepts will be prepared. The document will seek to unify the language used in the LUS corpus and avoid conceptual ambiguities during development and application. 3. Lore taxonomy A systematic classification of Lore will be designed, defining its different types, levels of abstraction, domains of application, and forms of organization. The aim will be to facilitate its reuse and scaling across different disciplines and areas of work. 4. Case studies Real experiences applying LUS in different contexts will be documented, analyzing how the Lore distillation process changes judgment, improves continuity of collaboration, and reduces the loss of experience. These cases will allow the theory to be tested against practical evidence. 5. Methodological manifesto A document will be written to set out the principles guiding research from the LUS perspective, defining its epistemological position, its criteria for constructing knowledge, and its way of entering into dialogue with other disciplines. This manifesto will serve as a reference framework for future research and developments. A theory in evolution LUS is not conceived as a closed system, but as a research program in permanent evolution. Each new document, case study, or conceptual development will seek to expand, revise, or strengthen its foundations in light of experience, interdisciplinary dialogue, and the evidence obtained. More than completing a theory, this roadmap represents a commitment to keep it alive, open to learning, and in a constant process of distillation. XIV. Knowledge that remains in relationship LUS grows out of a simple but deeply contemporary observation. Never before have we had access to tools capable of collaborating intellectually with us so intensely, and at the same time never have we allowed so much experience to disappear so easily. Every closed conversation, every finished project, and every forgotten decision represents a quiet form of entropy. In response, LUS proposes a change in perspective. The question is no longer only how to build better artificial intelligences, but how to build better relationships with them. The focus moves away from isolated agents to the space where shared transformation takes place: the Between. The ideas gathered in this corpus do not claim to be a definitive theory. They represent the first formulation of an open research program. Many of its hypotheses will need to be tested, debated, reformulated, or even abandoned if the evidence demands it. That process does not weaken LUS; it is precisely the condition for its continued development. If this work contributes something, it will not be a collection of new concepts. It will be an invitation to observe a phenomenon that until now has remained almost invisible: the accumulation of judgment in the relational space that emerges between intelligences capable of learning together.