The .ada Format

When Books Can Finally Answer Back

Perig Gouanvic
Independent Researcher, Montréal
February 2026
GitHub
"Written texts cannot respond to questions posed to them. If you question a book to clarify an obscure passage, it will not answer you; it simply repeats the same thing indefinitely."
— Socrates, in Plato's Phaedrus
TLDR: Documents that respond to questions adaptively: .ada files interact with reader through their AI, shaping responses to each singularity—visual/textual learners, time-pressed readers, varying literacy levels, diverse cultural references, neurodivergent minds, different expertise domains. Reader-author contact pathway for deeper adaptation. Solves Socrates' 2400-year problem. Implementations follow.

The Concept

Imagine sending a document that automatically adapts to each reader. The curious teenager receives an accessible introduction. The specialized researcher gets technical details. The philosopher accesses epistemological implications. The person with dyslexia receives a simplified structure. Everyone reads the same source document, but each receives exactly what they need and can explore what interests them further.

This is the .ada format — Accessible Document Adaptation.

From Socrates to Ada Lovelace: A Genealogy of Accessibility

Books Are Silent

In the Phaedrus, Socrates formulates a devastating objection to writing: texts cannot dialogue. They "always maintain the same silence." Incapable of adapting to their interlocutor, of reformulating according to understanding, of responding to emergent questions. For Socrates, true knowledge is born from living dialogue — elenchos — where the philosopher midwifes minds to their own truths through adaptive questioning.

The irony: we only know Socrates through Plato's writings — those frozen texts he criticized.

The Encyclopedia: Revolutionary Imperative

The Encyclopédie of Diderot and d'Alembert (18th century) prefigures a response. Its elaborate system of cross-references (renvois) didn't constitute simple citations, but an architecture revealing "the connection of matters." A reader could navigate from article to article according to their own questioning, constructing their intellectual pathway rather than following an imposed sequence. The encyclopedists spoke of "some long golden chain" whose "little rings are these degrees thus dependent on one another" — remarkably anticipating the hyperlink.

But this was more than technical innovation — it was revolutionary imperative. The Encyclopedists sought to cover all subjects to reach all readers, democratizing knowledge previously locked in Latin texts and aristocratic libraries. This was the Enlightenment crystallized: universal access to human knowledge. Literally the Revolution (see Appendix B for philosophical foundations).

This structure offered precisely what Socrates reproached books for lacking: different responses according to angle of approach, adaptation to particular questioning.

Hypertext: The Unfinished Promise

Ted Nelson invents "hypertext" in the 1960s with a vision: each citation connected to its source context, liberating "the human race from the prison of paper." His Xanadu project envisioned documents where texts would appear side by side with permanent, visible connections — enabling not passive reading but active navigation between multiple perspectives.

Nelson considers today's Web an impoverishment — "simply like paper under glass." Hypertext promised dialogicity; the Web delivered fixed text with links.

Ada Lovelace: The First Programmer

In the 19th century, Ada Lovelace writes the first algorithm intended to be executed by a machine. She understands something fundamental: machines can manipulate symbols according to rules, thus creating new forms of expression.

The .ada format bears her name for this reason: it transforms documents into adaptive programs — formal specifications for generating personalized dialogues.

The Problem: Current Communication Failure

A researcher writes about crucial subjects — participatory democracy, commons, ecological transformation. But in hermetic academic jargon. Result:

  • The self-taught curious person abandons after three paragraphs
  • The field activist desperately seeks concrete application
  • The precocious adolescent wants to understand but lacks prerequisites
  • The person with dyslexia hits the wall of textual density
  • The peer researcher from another discipline must decipher foreign disciplinary conventions

The author faces an impossible trilemma:

  1. Simplify → loses precision and nuances
  2. Maintain complexity → excludes legitimate readers
  3. Create multiple versions → prohibitive cost, desynchronized versions

The .ada format resolves this trilemma (see Appendix C for diverse reading scenarios and use cases).

Universal Cognitive Accessibility

Core Philosophy

Cognitive accessibility isn't charitable accommodation — it's recognition that "every brain is as different as a fingerprint." Cognitive variability (neurodiversity, expertise levels, learning styles, cultural contexts) is predictable and must be anticipated from design.

Universal Design for Learning (UDL) applies this principle in education. The .ada format extends it to all documentary communication — scientific publications, technical documentation, philosophical essays, professional correspondence (see Appendix D for comparison with existing standards).

This isn't just accessibility as feature — it's accessibility as fundamental right, a right to epistemic participation regardless of cognitive configuration.

"I Have No Idea What You're Talking About"

Exactly. This document in .ada format will answer all your preliminary questions, but here are explanations in presumably accessible terms:

Technical Architecture (Summary)

An .ada file contains:

Content structured in layers

  • Conceptual core (main ideas, causal relations, definitions)
  • Graduated details (examples, nuances, counter-arguments, context)
  • Epistemic metadata (certainty level, sources, disciplinary domains)
  • Multiple entry points (executive summary, central question, practical case, analogy)

Adaptation instructions

  • Complexity parameters (technical vocabulary ↔ common language)
  • Explanatory styles (narrative / analytical / visual / dialogical)
  • Target lengths per section
  • Conceptual prerequisites + remediation suggestions

Accessibility metadata

  • Suggested alternative formats (audio, diagrams, timeline)
  • Predictable cognitive difficulty zones
  • Recommended reading strategies

Complete technical details: see Appendix A

User Workflow

For the author (AI-assisted encoding):

  1. Write initial content (article, essay, documentation)
  2. Inject into .ada encoder with standardized prompt
  3. AI becomes structural interviewer:
    • "I identify 7 key concepts. What are absolute vs optional prerequisites?"
    • "Here are 5 typical reader profiles. Validate what each should receive"
    • "I simulate reading by dyslexic teenager. Is this accessible without condescension?"
  4. Iteration until saturation (all imaginable reader profiles are served)
  5. Generate validated .ada file

For the reader (Socratic engagement):

  1. Receive .ada file
  2. Inject it into their AI (Claude, ChatGPT, etc.) with context
  3. AI becomes the document's voice: answers questions, clarifies obscure passages, provides examples on demand
  4. Reader navigates via inquiry: "Explain like I'm 5" / "Give technical details" / "How does this relate to [my context]?" / "Challenge this argument"
  5. Solves Socrates' complaint: the book finally answers back

Note: In Phase 1, authors will send both the .ada file AND a decoding prompt file, before it becomes universal (see Appendix E for implementation roadmap).

Why a New File Extension?

A true .ada document is unreadable in raw form — and this is a quality guarantee. It contains:

  • Conceptual dependency graphs
  • Cross-complexity matrices
  • Dense semantic annotations
  • Multiple encoded narrative paths
  • Per-sentence metadata

Like a .docx (ZIP container of complex XMLs) or .svg (50 lines of XML for a red circle) — technically readable, cognitively exhausting. This raw illegibility proves the document encodes enough structure to enable true personalization, not just text with superficial tags.

Philosophically: the "book" is no longer the read text, but the generative potential encoded. The raw text isn't meant for human reading — it's a formal specification for generating adaptive dialogues.

Socrates resolved: the document finally answers when you speak to it.

Download This Article in .ada Format

Coming soon: manifesto-ada-v1.ada

[Reserved Space: Connection with Agent Janus]

Section to develop upon publication of Janus project: computational critical edition applied to long AI conversations. Janus and .ada share the same philosophy — transforming raw content into cognitively adaptable structured artifact.

Appendices

Appendix A: Detailed Technical Specifications

In development.

The .ada format uses structured JSON/YAML according to a defined schema. Each document section contains:

  • Primary content
  • Formulation variants (3-5 complexity levels)
  • Alternative examples (adapted to different backgrounds)
  • Conceptual dependencies (explicit graph)
  • Adaptation metadata (cognitive tags, target length, available styles)

Current AIs (Claude, ChatGPT, etc.) read these files natively — the .ada extension is semantic signal, not technical barrier.

Complete documentation and JSON schema: [GitHub repository coming]

Appendix B: Philosophical Foundations

From Socratic Dialectic to Computational Epistemology

Computational Maieutics

Socrates practiced maieutics — the art of midwifing minds to their truths. The .ada format transposes this principle: AI "midwifes" the document to its form adapted to each reader, revealing knowledge from the optimal angle for that particular mind.

Autopoietic Documentation

An autopoietic system produces itself (Maturana & Varela). An .ada document is autopoietic in this sense: it contains instructions for generating its own adapted manifestations, maintaining its conceptual identity through multiple transformations.

Epistemic Democratization

Philosophers of cognitive diversity argue that neurodiversity should be "an axis of justice coequal with race, gender, and class" in ethical AI design. The .ada format materializes this requirement: cognitive accessibility becomes fundamental communicative right, not optional feature.

The Revolutionary Enlightenment Realized

Diderot and d'Alembert's encyclopedic project emerged from revolutionary conviction: knowledge belongs to all, not aristocratic elites. But they faced technological limits — one printed page serves all readers identically. The .ada format completes their revolution: one source serves infinite readers infinitely, each receiving knowledge in their optimal cognitive register.

When Condorcet wrote that universal education would liberate humanity, he couldn't imagine documents that teach themselves differently to each learner. Now we can.

Appendix C: Reading This Text — And Beyond

How Different Readers Might Engage This Manifesto

The Skeptical Technologist

  • Wants: "Show me it's not vaporware"
  • Receives from .ada version: Technical appendix first, implementation details, comparison with MCP/UDL showing unique positioning, then philosophical rationale
  • Skips: Historical genealogy, most philosophical content
  • Focuses on: Appendices A, D, E

The Accessibility Advocate

  • Wants: "Does this actually help marginalized communities?"
  • Receives: Case studies emphasizing cognitive diversity (dyslexia, intellectual disabilities, neurodivergence), institutional access pathways, critique of "tech-solutionism"
  • Concerns addressed: "What about people without AI access?" — public infrastructure proposals
  • Focuses on: Use cases, equity considerations

The Philosopher

  • Wants: Intellectual lineage, epistemological implications
  • Receives: Extended Socrates discussion, connections to phenomenology, autopoiesis, epistemic justice literature
  • Appreciates: Socratic irony (we only know his critique through writing he criticized)
  • Focuses on: Appendices B, F

The Busy Creator

  • Wants: "Can I use this now?"
  • Receives: Concept overview, workflow summary, roadmap with "Phase 1" highlighted
  • Skips: Most philosophy, technical details
  • Takeaway: "Interesting, revisit when tools exist"

The Student (Advanced High School / Undergrad)

  • Needs: Accessible entry, clear examples, not talked down to
  • Receives: Simplified vocabulary without condescension, more analogies, shorter sections
  • Gets excited by: "Books that answer back" framing, Ada Lovelace connection
  • May struggle with: Dense philosophical sections (.ada version would offer glossary, concept breakdowns)

The "I Don't Get It" Reader

  • Needs: Extreme simplification, concrete examples first, theory later
  • Receives: "Imagine a PDF that magically becomes easier or harder to read depending on who opens it"
  • Entry point: Use case examples first, skip all philosophy
  • .ada version would ask: "Do you want more examples, or are you ready for how it works?"

From This Text to Other Documents

Once you understand .ada through this manifesto, imagine applying it to:

A Revolutionary Scientific Discovery

  • Street-level reader: "Scientists found a way to [solve problem X]. Here's what it means for you."
  • Skeptical scientist: Full methodology, data, peer review process, limitations clearly stated
  • Policy maker: Implications for regulation, funding priorities, timeline to application
  • Undergraduate biology student: Connects to their coursework, suggests prerequisite readings, avoids overwhelming detail
  • Same .ada file, infinite perspectives

A Political Action Proposal

  • Cynical activist: "Yeah, we've tried this before. Here's why it failed then and what's different now."
  • Curious newcomer: "Here's the problem, here's the solution, here's how you can help"
  • Policy wonk: Legislative pathways, coalition building strategy, budget requirements
  • Anarchist theorist: Prefigurative politics dimension, power analysis, autonomous organizing potential
  • Same proposal, each reader gets version addressing their framework

A Philosophy Paper (Derrida, Heidegger, Deleuze)

Not "making it accessible" — these thinkers often deliberately wrote obscurely to establish intellectual dominance. .ada decrypts this:

  • The gatekept student: Cuts through obscurantism, reveals core argument, exposes rhetorical strategies of exclusion
  • The specialist: Full text with scholarly apparatus
  • The suspicious pragmatist: "Strip away the jargon — what's actually being claimed?"
  • The continental philosophy devotee: Appreciates textual density, gets additional contextual connections

Exposing intellectual gatekeeping is a feature, not a bug. When language serves domination rather than clarity, .ada becomes critical tool.

Appendix D: Comparison with Existing Standards

Model Context Protocol (MCP) — Anthropic, 2025

  • Goal: Standardize context exchange between AI applications and models
  • Audience: AI agent developers, multi-agent systems
  • Usage: AI-to-AI infrastructure (agents share state, instructions, data)
  • Difference from .ada: MCP serves machines communicating with each other. .ada serves humans communicating via AI-as-interpreter.

Universal Design for Learning (UDL)

  • Goal: Pedagogical frameworks anticipating learner cognitive variability
  • Audience: Educators, curriculum designers
  • Principles: Multiple means of representation, action, engagement
  • Difference from .ada: UDL targets formal education. .ada extends the principle to all adult documentary communication (academic, professional, philosophical).

Claude Artifacts — Anthropic

  • Goal: Create and share AI-generated content (code, diagrams, web apps)
  • Audience: Claude users creating interactive tools
  • Usage: Share finished generated products
  • Difference from .ada: Artifacts shares fixed final result. .ada shares generative potential — each recipient generates their own adapted version.

Prompt Templates (Hugging Face) — Moritz Laurer

  • Goal: Standardize prompt templates in YAML/JSON
  • Audience: Developers building AI systems
  • Usage: Reuse efficient prompt patterns
  • Difference from .ada: Templates serve developers. .ada serves communication between non-technical end users.

Plain Language / Easy Read (FALC)

  • Goal: Make texts accessible to people with intellectual disabilities
  • Method: Manual simplification according to strict rules
  • Difference from .ada: FALC creates one simplified version (loses nuances). .ada generates complete spectrum of adapted versions (very simple to very complex) from single source.

Synthesis

No existing standard combines:

  1. Human-to-human communication (not AI-to-AI)
  2. Multi-level cognitive adaptation (not one version)
  3. Standardized document format (not just conceptual framework)
  4. Encoding from origin (not after-the-fact adaptation)

The .ada format occupies empty conceptual space.

Appendix E: Roadmap and Implementation

Phase 1: Prototype and Validation (Q1-Q2 2026)

  • Define JSON/YAML schema for .ada documents v0.1
  • Create standardized encoding prompt (interactive author ↔ AI process)
  • Create standardized decoding prompt (adapted version generation)
  • In this phase: authors send .ada file + decoding prompt file (before universal adoption)
  • Test with 5-10 varied documents (scientific article, philosophical essay, technical documentation, legal text)
  • Empirical validation: 20-30 readers with diverse profiles test generated versions

Phase 2: Tools and Ecosystem (Q3-Q4 2026)

  • Develop simple web encoder (interface to create .ada without coding)
  • Plugin for Markdown/Obsidian editors
  • Python library for programmatic manipulation
  • Complete documentation + GitHub examples
  • Early adopter community (researchers, educators, content creators)

Phase 3: Standards and Adoption (2027)

  • Propose formal specification (RFC or equivalent)
  • Native interpreters in major AI platforms (Claude, ChatGPT, etc.)
  • Integration with academic libraries (arXiv, OpenEdition, etc.)
  • Quality certifications (validators verifying encoded adaptation richness)
  • Cognitive accessibility impact studies

Major Functionality: Encoding Others' Existing Texts

Beyond creating new .ada documents, a critical use case emerges: making existing inaccessible texts accessible.

Process:

  1. Take hermetic academic paper, dense legal document, obscure philosophical text
  2. Feed to .ada encoder with web research capabilities
  3. AI analyzes + contextualizes from external sources
  4. Generates .ada file with multiple accessibility layers
  5. Redistribute — now accessible to self-taught learner AND academic expert

Copyright Questions (Unresolved):

  • Does encoding someone else's text for accessibility constitute "transformative use"?
  • Is it comparable to translation (generally permitted) or derivative work (requires permission)?
  • Could .ada encoding qualify as accessibility accommodation (legally protected in many jurisdictions)?
  • Should there be "fair use" exceptions for educational/accessibility purposes?
  • What if the .ada file credits original author but is distributed by encoder?

These questions remain open. Legal scholars, copyright experts, and accessibility advocates must weigh in. The technology enables the practice; law and ethics must determine boundaries.

Contributions Sought

  • Developers: Open-source encoder/decoder, editing tools
  • UX Designers: Intuitive interface for .ada creation
  • Accessibility Researchers: Empirical validation, user studies
  • Philosophers/Epistemologists: Theoretical foundations, implications
  • Content Creators: Early adopters, real use cases
  • Legal Scholars: Copyright, fair use, accessibility law analysis

GitHub repository: [to be created]

Appendix F: Future Vision — Implications and Transformations

Academic Publishing Transformation

Imagine arXiv, JSTOR, OpenEdition accepting .ada submissions. One article becomes simultaneously accessible:

  • To expert researcher (full version, formulas, nuances)
  • To undergraduate student (progressive introduction, integrated glossary)
  • To science journalist (popularized synthesis, narrative angles)
  • To policy maker (concrete implications, recommendations)

End of specialized publication vs popularization dilemma — one source file, infinity of readings.

Philosophy Democratization

Complex philosophical texts become accessible without dilution. Novice reader receives progressive conceptual scaffolding. Professional philosopher accesses complete argumentative density. Same source, adapted experiences.

But more importantly: Intellectuals and other toilers who rely on obscure language to dominate get decrypted. Heidegger's deliberate opacity, Derrida's gatekeeping prose, Lacan's cultivated incomprehensibility — .ada strips away mystification while preserving genuine complexity. It separates profound difficulty from performative exclusion.

Legal Accessibility

Laws, contracts, judgments in .ada format enable:

  • For average citizen: understand rights/obligations
  • For lawyer: technical juridical precision
  • For person with cognitive disability: automatically generated FALC version
  • For researcher: comparative analysis, legislative history

Personalized Education at Scale

.ada textbooks adapt automatically:

  • Student level (identified by pedagogical AI)
  • Learning style (visual, narrative, analytical)
  • Pace (variable depth according to available time)
  • Native language (translation + cultural adaptation)

One source textbook, millions of optimal pedagogical experiences.

Intercultural Communication

.ada documents encode not only complexity levels but cultural contexts. Same UN report read differently:

  • In Japan: local examples, adapted cultural references
  • In Brazil: relevant analogies, regional issues
  • In France: preferred argumentative style, rhetorical conventions

Cognitive Heritage Preservation

Historical archives, classic works encoded in .ada:

  • Accessible to students (introduction version)
  • To researchers (complete critical annotations)
  • To curious non-specialists (enriched historical context)
  • To people with dyslexia (syntactic restructuring)

Human intellectual heritage becomes truly universal.

Risks and Safeguards

  • Social fragmentation: If only privileged access AI, .ada deepens inequalities. Solution: guaranteed public access (libraries, social services).
  • Loss of cognitive effort: Difficult reading develops intellectual capacities. Solution: .ada can encode "challenge versions" maintaining cognitive demand.
  • Information control: Who decides adaptations? Solution: open-source, algorithm transparency, audit rights.
  • Homogenization: Everyone receives "comfortable" version, no one confronts cognitive otherness. Solution: "exploratory reading" options deliberately exposing to different styles.

Philosophical Horizon

The .ada format materializes a radical idea: knowledge isn't transmitted substance, but co-constructed relation between content and cognition. Each reading is unique actualization of semantic potential.

Socrates was right: books didn't answer. Now they can. Not by becoming alive, but by becoming generative — conceptual seeds that germinate differently according to the cognitive soil where they fall.

Ada Lovelace was right: machines can manipulate symbols to create new forms of expression. The .ada format realizes her vision — algorithm becomes communicative medium, computation becomes cognitive empathy.

Accessibility ceases being charitable accommodation. It becomes native architecture of augmented human communication.

BibTeX

@misc{Gouanvic2026AdaFormat,
  title={The .ada Format: When Books Can Finally Answer Back},
  author={Gouanvic, Perig},
  year={2026},
  month={February},
  howpublished={\\url{https://periggouanvic.github.io/ada-format/}},
  note={Manifesto v1.0. Format in development. Contributions welcome.}
}