Learning Guides (Guías de Aprendizaje)
What this is. The document that accompanies the person who is learning throughout the learning: it tells them where they stand, what they are pursuing, what they are going to do, how they will know they achieved it and which decisions are theirs. Its reader is the Learning Subject (Sujeto de Aprendizaje) — the participant, the learner — and it is written for them, in the second person. It exists at three altitudes: that of the Learning Program, that of the Learning Experience and that of the Learning Unit (Programa, Experiencia, Unidad de Aprendizaje). They are three distinct documents, not three versions of the same one.
Why it exists. Smoother declares far more than one person can read: eight components of the Learning Object (Objeto de Aprendizaje), eight Unit Structures, six Task Types (Tipos de Tarea), a rubric anchored to the Learning Object, sixteen Personal Preferences, an Experience Layer (Capa de Experiencia), a General Learning Definition (Definición General del Aprendizaje). All of that is declared on the design side. The Learning Guide is the translation of that declared structure into the perspective of the person who is going to live it — and that is why it is a derivable artifact: almost everything it contains already exists somewhere else in the model.
What it is NOT
It is not the content — it does not explain the topic; if the guide teaches, something was written in the wrong place. It is not the topic list — it is not a list of topics with dates, but an orientation on how to travel through them. It is not a grade nor a record — it shows the rubric as a mirror, what will be seen when each component is achieved, never as an accumulated score. And it is not the guide of the Learning Facilitator: what the Learning Facilitator (Orientador de Aprendizaje — the role that accompanies the process) needs lives in the Guidance Guide (Guía de Orientación), the sibling document that uses the same skeleton.
The three altitudes, and inheritance
| Altitude | The question it answers | When it is read |
|---|---|---|
| Of the Learning Program | What am I getting into, why does it concern me, and how do I navigate it? | On entry; returned to in order to reorient. |
| Of the Learning Experience | What will I be able to do when this Experience ends, and how will I know I achieved it? | On entering each Experience. |
| Of the Learning Unit | What do I do before, during and after this Unit? | Alongside the Unit, while working. |
Nothing is repeated downward. The guide below inherits the orientation of the one above and only adds what belongs to its own altitude. Pointer, never copy: when the lower altitude needs something from the higher one, it links to it — a guide that reproduces the text of the one above falls out of sync at the first change. The only legitimate exception is local contradiction: if in this Experience or Unit something works differently from what the guide above says, it is said explicitly and flagged as an exception. And despite the inheritance, each guide reads on its own: the pointers are aids, not prerequisites for reading.
The origin of each section, which is the whole point
Each section declares its origin, and the origin is exactly one of three:
generada(generated) — it is derived mechanically from already declared structure, and the source is named: a component of the Learning Object, the Unit Structure, the Learning Tasks, the rubric, the Preferences, the Experience Layer, the General Learning Definition, platform fields.autorada(authored) — a person writes it, and who is stated: the design team, or the assigned Learning Facilitator.invariante(invariant) — fixed Smoother text, identical in every guide at that altitude, not rewritten per Learning Program.
That column is what allows the platform to generate the guide instead of asking for it to be written. The Program guide has fourteen sections — seven generated, two authored, five invariant; the Experience one, ten — seven, one and two; the Unit one, eleven — nine, one and one. The proportion is not accidental: it is the document’s whole argument.
Among the sections: what the Program pursues, stated from the eight components of the Learning Object; what it helps to arrive with; how it is organized into Learning Blocks and what Learning Paths exist; how you are going to know you progressed, with the rubric presented as what it is — rows = components of the Learning Object, four levels (Not achieved · In development · Achieved · Mastered) and the result as a profile by component, not as a grade; how much time and how to distribute it; the study techniques of one’s own; who can be counted on; and what comes before, during and after each Learning Unit.
Five guards, and a sixth
They hold at all three altitudes, and they hold equally for the hand-written guide and for the generated one.
- It never prescribes by preference. The Personal Preferences are offered as a menu of variety, never as an assignment. Every formulation that hints at matching — “your learning style is X, therefore…” — is prohibited.
- It never turns the rubric into a grade. It shows rows and descriptors, and returns a multi-component profile. It does not average, does not accumulate points, does not produce a single number or a ranking.
- It never restores the content. If a section of the guide can be studied instead of the material, it is badly written.
- It never uses the refused vocabulary. Not “estudiante” (student), not “alumno” (pupil), not “trainee”, not “curso” (course), not “módulo” (module), not “docente” or “profesor” (teacher, professor). It says participante / aprendiz / Sujeto de Aprendizaje, Unidad de Aprendizaje, Experiencia de Aprendizaje, Programa de Aprendizaje, Orientador de Aprendizaje. The institutional-register exception of the capacity rule does not reach this document: the Learning Guide is exactly what a participant reads, and there the prohibition is total.
- It never claims proven efficacy. It does not say “proven method”, nor “evidence-based”, nor “the research shows that this guide improves your results”. It may say which instrument it uses and where that instrument comes from.
And a sixth, minor but recurrent: it never invents what is not declared. If the Experience Layer does not declare a horizon or an offered choice, the corresponding section is omitted — no referent and no choice that the design did not offer is fabricated.
How it is generated
Each altitude has a precondition: without a populated Learning Object the Program guide is not generated, and this is reported as a block instead of issuing a mutilated guide; without a declared Structure there is no Unit guide. The generated sections are issued complete, the invariant ones are inserted from a template without ad-hoc translation, and the authored ones are issued as slots — an empty slot does not prevent publication, but filling it with generated text that simulates human authorship is prohibited. Regeneration rewrites the generated and the invariant and preserves the authored intact: human text is never lost to a regeneration. And every issued guide records version and date, so that the person reading it knows whether it corresponds to the current design.
Evidence
Level: Synthesized. This document contributes no findings of its own and no new instrument: it is an assembly of already existing canon, reordered from the perspective of the person who is learning and fixed as a generation contract. Each piece it carries keeps its own label: the Learning Object and the Structures, original; the six Task Types and the rubric, synthesized; the Pashler posture, backed in its critique; the Experience Layer, original / design position, authored by Carlos Miranda Levy; the participant’s techniques, mixed by technique.
What is its own — and therefore what is unbacked — are three design decisions: the section skeleton shared with the Guidance Guide, the origin taxonomy generada / autorada / invariante, and the pointer-inheritance rule between altitudes. They are document-architecture decisions, not claims about learning. No assembled instrument lends its label to the assembly: that the rubric is well founded does not make this guide well founded.
Permitted wording: “consolidates instruments already in the canon” · “derives from the declared structure” · “makes the guide generable” · “a Smoother design position” · “aligns with” · “is consistent with”. Prohibited wording: “proven method” · “evidence-based” · “the research shows that learning guides improve results” · “a guide personalized to your learning style” · any formulation that attributes to the guide a measured effect on learning.
Related
Guidance Guides · The Learning Object · Learning Units · Learning Tasks · Evaluation · Personal Preferences · Discovery · Experience Layer · Study and Practice Techniques · Learning Blocks · Learning Paths · Task Routing
Source
protocol/smoother-guias-de-aprendizaje.md — state of 2026-09-17 (open section of the CHANGELOG, after 0.40.0).