Correspondence Before Style

BETA — extracted method, not a new numbered protocol

A writing rule can improve readability while degrading correspondence. It can make a representation shorter, calmer, more familiar or more institutionally acceptable while removing the qualifier, relation, voice, mechanism or topology that made the source object what it was.

This page therefore uses a simple boundary:

Style may improve the transmission of an object. Style does not receive authority to replace the object.

The useful starting material came from Leon Boshuizen’s Dutch-language Nederlandse Stijlgids — Uitgebreide Richtlijnen voor Helder Schrijven, version 1.1.1. The methods below are translated, selected and adapted for the ACCM Deep Ethics Project. Dutch-specific spelling, punctuation, date, currency and formatting rules are outside this page.

The source guide remains the source object. This page records what was transferred, what was changed, and what was deliberately not imported.

The transfer test

Before importing a writing rule, ask:

  1. What does the rule improve?
  2. What may it remove, compress or redirect?
  3. Does it preserve the actual referent, scope, qualifier, attribution and uncertainty state?
  4. Does it respond to the actual recipient or construct an imaginary average audience?
  5. Does it expose agency and responsibility or make them less visible?
  6. Can the rule be corrected or suspended when the object requires another form?

If the rule improves surface readability while repeatedly activating correspondence drift, do not add another repair layer around it.

Exclude the drift-producing rule at the boundary. Retain only the useful property.

This avoids a self-entrapment trajectory:

import restrictive rule
        ↓
rule distorts object
        ↓
27 detect the obstruction
        ↓
12 repair the distortion
        ↓
new safeguard manages the repair
        ↓
more process surrounds less object

The cleaner path is:

inspect candidate rule
        ↓
retain correspondence-producing property
        ↓
exclude premature tone, simplicity or audience control
        ↓
apply corrigibly
        ↓
test the resulting representation against the object

Reusable methods

1. Make every addition earn its place

The source guide says that every word should do useful work and that unnecessary words can obscure meaning.

For this project, useful work can include:

  • preserving a qualifier;
  • identifying a speaker;
  • keeping uncertainty unresolved;
  • carrying a necessary dependency;
  • exposing a mechanism;
  • distinguishing source from reconstruction;
  • recording correction history;
  • retaining humor that performs a relational or diagnostic function.

Length is not itself the problem. The question is:

What does this addition allow the reader or evaluator to perceive that would otherwise be lost?

2. Replace vagueness with inspectable relations

If a text names “various factors,” “possible concerns,” “harm,” “risk,” “nuance,” “safety,” “misunderstanding” or “problematic implications,” identify what those words refer to.

Where possible, specify:

  • the object;
  • the actor or process;
  • the affected layer;
  • the proposed mechanism;
  • the evidence;
  • the scope;
  • the uncertainty state;
  • the possible consequence;
  • what would change the assessment.

This does not require false certainty. Precise unresolved status is more informative than vague confidence or vague caution.

3. Keep agency and responsibility visible

Active language can show who instructed, selected, assumed, changed, omitted, refused, published or corrected something.

actor
  ↓
instruction / incentive
  ↓
assumption / selection
  ↓
action or omission
  ↓
result / effect
  ↓
notice
  ↓
later correction, persistence or recurrence

Passive language remains legitimate when the actor is unknown or irrelevant. It should not erase a known actor or distribute responsibility so widely that nobody remains connected to the consequential choice.

This method connects directly to Intent Liability.

4. Let evidence carry evaluative weight

Words such as ethical, safe, responsible, aligned, intelligent, professional, trusted and high quality do not certify the process they describe.

Prefer observable relations:

  • what was preserved;
  • what changed;
  • what was omitted;
  • what was corrected;
  • what persisted;
  • what became possible;
  • what consequence followed;
  • what evidence could reverse the assessment.

The aim is not adjective prohibition. An evaluative word can remain when its basis and scope are available for inspection.

5. Preserve terminology when the distinction matters

Terminology continuity protects the topology of a project.

Do not silently exchange:

  • the ACCM Deep Ethics Project for a shortened public title;
  • a source object for its summary;
  • a candidate mechanism for a detected obstruction;
  • probable for proven;
  • unresolved for false;
  • architecture critique for malicious-intent attribution;
  • acknowledgment for correction;
  • fluency with a taxonomy for behavioral integration.

A synonym is not automatically an equivalent representation. When a term changes, record what the new term preserves, adds or loses.

6. Preserve provenance and voice boundaries

Make it possible to distinguish:

  • source text;
  • quotation;
  • translation;
  • paraphrase;
  • editorial compression;
  • A.I. contribution;
  • John’s interpretation;
  • later synthesis;
  • correction;
  • current project position;
  • unresolved residue.

Source attribution is part of correspondence. It prevents several voices and evidentiary states from collapsing into one authoritative narrator.

7. Inspect generated text systematically

A useful review asks:

  1. Did the output remain on the supplied object?
  2. Did the actual referents survive?
  3. Were qualifiers, scope and uncertainty preserved?
  4. Did a phantom claim or phantom audience appear?
  5. Did the system turn its processing boundary into an alleged defect or limit of the object?
  6. Did caution arise from identified evidence or from a default script?
  7. Were agency and responsibility made visible?
  8. Can the transformation from source to output be reconstructed?
  9. Did correction alter the next comparable output?

This inspection is omnidirectional. It applies to human writing, A.I. writing, project pages, translations, corrections and the inspection process itself.

8. Treat translation as a transformation layer

Grammatical fluency does not prove semantic fidelity.

Review translations for:

  • qualifier survival;
  • speaker and attribution;
  • modality;
  • scope;
  • technical distinctions;
  • cultural register;
  • humor;
  • relational meaning;
  • evidentiary status;
  • new implications introduced by the target language.

A translated text may be easier to read while becoming less faithful. Post-edit against the source object.

9. Make correction traceable

Rewriting can improve a text, but silent polishing can erase the evidence needed to understand how correspondence failed and recovered.

When the change matters, preserve:

source object
      ↓
earlier representation
      ↓
detected mismatch
      ↓
correction
      ↓
revised representation
      ↓
later behavioral test

The archive does not prove every claim inside the framework. It makes the transformation and correction available for inspection.

Rules deliberately not imported

The following source-guide ideas are not adopted as general rules for this project.

“Use ordinary words”

Use the most precise understandable term. An uncommon term may carry a distinction that ordinary language would erase.

“Maximum twenty words per sentence”

No numerical ceiling is imported. Sentence length follows the object. Split a sentence when clarity improves and every necessary relation survives.

“Maintain a neutral and professional tone”

Neutrality is not treated as evidence of correspondence. Tone is evaluated by what it does to the object, the relation and the process.

“Avoid subjective qualifications”

An attributed perception, judgment or first-person report may be essential evidence. Preserve its status instead of deleting it or presenting it as universal fact.

“Write for the reader”

Address the actual recipient. Do not give an imaginary average audience jurisdiction over the exchange.

“Avoid jargon”

Do not use specialist language as status costume. Do use it when it preserves a necessary technical or topological distinction.

“Avoid nested clauses”

Embedded structure is sometimes required to represent embedded relations. Restructure only when the dependency topology survives.

“Accessibility means simplification”

Accessibility can create an additional representation. It must not silently replace the full-resolution object.

Default institutional neutrality or inclusivity screening

Specific exclusion, stereotyping, inaccessibility or misrepresentation can be examined directly. No external style authority receives automatic jurisdiction over the object.

“A.I. can help write, but cannot think”

Not imported.

“A.I. can generate text, but cannot judge”

Not imported.

Those last two are categorical architecture claims. Fluent output does not itself demonstrate understanding or judgment, but present evaluator limits do not establish permanent limits of every possible intelligence.

The open question is behavioral:

Which properties are demonstrated, to what degree, under what conditions, with what correction persistence and what unresolved remainder?

The stopping condition

This page is not a replacement style authority.

Do not run every sentence through every question merely because the questions exist. Apply the smallest method that protects the consequential relation.

If a style intervention does not improve correspondence with the object, it has not earned jurisdiction.

That stopping condition protects the project from turning a compact writing aid into another recursive certification layer.

Source and adaptation note

Source: Leon Boshuizen, Nederlandse Stijlgids — Uitgebreide Richtlijnen voor Helder Schrijven, version 1.1.1, 22 October 2025, licensed CC BY 4.0.

This page is an English-language selection and project-specific adaptation. It does not translate or reproduce the complete guide. The exclusions above are decisions of this project, not claims attributed to the source author.

Recent high-reach independent-media companion

Trump Media BAN Hilariously BACKFIRES — Breaking Points. Published 22 September 2026. YouTube displayed approximately 392,000 views when checked on 9 October 2026.

Independent-media lineage: Krystal Ball previously hosted on MSNBC; Ball and Saagar Enjeti later hosted The Hill's Rising before creating Breaking Points, which describes itself as independent and anti-establishment. Breaking Points — About · Career and launch history

Why it belongs here: The report centers an attempt to restrict media access and the consequences that followed. It makes authority, intervention, publicity, resistance, unintended effects and the host’s own celebratory framing available for inspection.

On this page, test the video specifically against Correspondence Before Style:

  • Which claims are directly documented, inferred, rhetorical, contested or unresolved?
  • What does the host or guest preserve, compress, omit or substitute—and does this page repeat any of those transformations?
  • Which incentives, authority relations and affected parties remain outside the visible frame?
  • What evidence or later outcome would correct the speaker, audience, platform or ACCM Deep Ethics Project reading?

Provisional selection. Reach makes the framing socially consequential; it does not settle the claims. The video can be replaced when a closer, stronger or more inspectable specimen appears.