I3085 — An article explains how a doubt is settled in both languages: by an experiment, not a count
Publish site/content-src/a-doubt-is-settled-by-an-experiment.md, an English article on how a session settles a material engineering uncertainty (ADR 0098). It covers why a decision made silently cannot be audited and why advice pasted into a chat is lost, the assessment and the evidence its writer requires, the review budget a policy computes, advisors as optional capabilities whose absence is state, the structured local review, a disagreement settled by an experiment and never by a count, the conclusion whose review count is computed, how the records reach the next session, and what the protocol does not do. Its example is the doubt weighed on 2026-10-21 that docs/RELEASE.md records. Every statement is one that ADR 0098, rules project.advisors-are-optional and project.review-is-recorded-not-claimed, docs/REASONING.md, docs/RELEASE.md or the reasoning code on master states, and no count is typed by hand. site/content-src/pochybnost-rozhodne-experiment.md is its Czech translation, naming translated_from and its source's digest. site/data/nav.toml lists both. Both relate the plan-critique, continuity and mesh articles, each of which names them back in its own language.
BLOCKED wave 10 · p2 · implementation profile · parallel safe
Part of public-proof — A visitor sees the product prove itself, and can check every claim it makes.
Blocked. This issue cannot start until I3084, I3076 are done. The status is derived from that, not declared.
Objective
Publish site/content-src/a-doubt-is-settled-by-an-experiment.md, an English article on how a session settles a material engineering uncertainty (ADR 0098). It covers why a decision made silently cannot be audited and why advice pasted into a chat is lost, the assessment and the evidence its writer requires, the review budget a policy computes, advisors as optional capabilities whose absence is state, the structured local review, a disagreement settled by an experiment and never by a count, the conclusion whose review count is computed, how the records reach the next session, and what the protocol does not do. Its example is the doubt weighed on 2026-10-21 that docs/RELEASE.md records. Every statement is one that ADR 0098, rules project.advisors-are-optional and project.review-is-recorded-not-claimed, docs/REASONING.md, docs/RELEASE.md or the reasoning code on master states, and no count is typed by hand. site/content-src/pochybnost-rozhodne-experiment.md is its Czech translation, naming translated_from and its source's digest. site/data/nav.toml lists both. Both relate the plan-critique, continuity and mesh articles, each of which names them back in its own language.
Why
The owner asked on 2026-10-19 that the site's articles be developed and interlinked in every language. How a session decides under uncertainty, with evidence first and advisors optional, was explained only in docs/REASONING.md and ADR 0098, which are written for the sessions that follow them.
Current state
No article explains how a doubt is settled. The article groups relate each other, and none of them names it.
Desired state
/a-doubt-is-settled-by-an-experiment/ and /pochybnost-rozhodne-experiment/ are built, are counterparts of each other, are listed in the navigation, and are related both ways with three other groups. Check 21 passes, with the translation at its source's digest.
Scope
- .ai/repo/project/issues/I3085.yaml
- site/content-src/a-doubt-is-settled-by-an-experiment.md
- site/content-src/pochybnost-rozhodne-experiment.md
- site/content-src/a-plan-is-attacked-before-it-is-built.md
- site/content-src/plan-se-napadne-driv-nez-se-postavi.md
- site/content-src/a-session-ends-the-work-does-not.md
- site/content-src/session-skonci-prace-ne.md
- site/content-src/many-machines-one-journal.md
- site/content-src/mnoho-stroju-jeden-denik.md
- site/data/nav.toml
- scripts/ci/english-only-allow.txt
Out of scope
- Changing the reasoning protocol, its policy, its records or the ADR and rules the article explains
Dependencies
- I3084READY An article explains the mesh in both languages: many machines, one journal
- I3076BLOCKED The landing clock's first source is measured, a fresh clone's wait is a labelled bound, and case 3470 shows a decided version
What waits on this
Acceptance criteria
- Both articles are built, declare each other in [extra.translations], and the Czech one records translated_from = a-doubt-is-settled-by-an-experiment/ and the digest check 21 computes of its source
- Each statement is one a record, manual or the reasoning code on master states, and check 19 finds no count typed by hand
- scripts/site-check --articles passes the tree, counting the new related links
Validation
- majordomus plan validate
- scripts/site-check --articles site/content-src site/data/nav.toml site/config.toml
- env -u MAJORDOMUS_SHARE -u MAJORDOMUS_HTTP_HOST -u MJ_ROOT bash test/run.sh 3389_related_articles_answer_each_other_in_every_language
Evidence
None recorded. Every token above needs a command or an artifact behind it before this issue can be completed; narrative is refused.
Timeline
- started
- —
- verified
- —
- completed
- —
Those three fields, the evidence above and the state of the dependencies are all the status is made of. There is no status field to disagree with them.
Canonical record: .ai/repo/project/issues/I3085.yaml. Read it back with majordomus plan show I3085.