HAUSE

THE EVIDENCE ROOM / RECORDS, NOT RENDERINGS

The result.
The doubt.
The record.

A useful premise should survive inspection. Here, every mark is a scored case—and a missed case stays visible.

OPEN THE FIRST RECORD ↓ALL FIVE EVALUATIONS ↓

EXHIBIT 01 / CHOOSING-1 · 31 AUGUST 2026 · FROZEN

Same cases.
Different selectors.

124 descriptions of unfamiliar content. Three conditions. Each panel below contains every recorded outcome—not a decorative approximation of the score.

One author’s cases, one model family, one run per model condition. That family also co-authored the grammar. This tests whether the written vocabulary carries distinctions; it does not measure interface quality or independent production value.

A / CATALOGUE

122 / 124

122 exact · 2 missed
Solid = exact · slash = miss · one mark per case

One model, given form names and one-line descriptions. Fresh context per batch; unfamiliar cases written by the same author as the grammar.

READ EVERY CASE (124)
A / CATALOGUE — recorded selections
CaseExpectedSelectedResult
c001HeroHeroExact
c002HeroHeroExact
c003StatementStatementExact
c004StatementStatementExact
c005ClaimClaimExact
c006ClaimClaimExact
c007ObservationObservationExact
c008ObservationMISSINGMiss
c009AnswerAnswerExact
c010AnswerAnswerExact
c011ExcerptExcerptExact
c012ExcerptExcerptExact
c013SnippetSnippetExact
c014SnippetSnippetExact
c015EvidenceEvidenceExact
c016EvidenceEvidenceExact
c017AgreementAgreementExact
c018AgreementAgreementExact
c019DerivationDerivationExact
c020DerivationDerivationExact
c021ProvenanceProvenanceExact
c022ProvenanceProvenanceExact
c023CitationCitationExact
c024CitationCitationExact
c025RefusalRefusalExact
c026RefusalRefusalExact
c027QuestionQuestionExact
c028QuestionQuestionExact
c029VariantsVariantsExact
c030VariantsVariantsExact
c031LadderLadderExact
c032LadderLadderExact
c033AnatomyAnatomyExact
c034AnatomyAnatomyExact
c035DecompositionDecompositionExact
c036DecompositionDecompositionExact
c037ByteMapByteMapExact
c038ByteMapByteMapExact
c039LensLensExact
c040LensLensExact
c041ExpertFieldExpertFieldExact
c042ExpertFieldExpertFieldExact
c043GatingGatingExact
c044GatingGatingExact
c045TerminalTerminalExact
c046TerminalTerminalExact
c047FollowRevealFollowRevealExact
c048FollowRevealFollowRevealExact
c049ComparisonComparisonExact
c050ComparisonComparisonExact
c051TransformationTransformationExact
c052TransformationTransformationExact
c053UnfoldingUnfoldingExact
c054UnfoldingUnfoldingExact
c055CompilationCompilationExact
c056CompilationCompilationExact
c057ProcessionProcessionExact
c058ProcessionProcessionExact
c059MagnitudeMagnitudeExact
c060MagnitudeMagnitudeExact
c061ChannelChannelExact
c062ChannelChannelExact
c063QuantisationQuantisationExact
c064QuantisationQuantisationExact
c065FilmFilmExact
c066FilmFilmExact
c067ConnectionConnectionExact
c068ConnectionConnectionExact
c069TimelineTimelineExact
c070TimelineTimelineExact
t001StatementStatementExact
t002ClaimClaimExact
t003ObservationObservationExact
t004AgreementAgreementExact
t005EvidenceEvidenceExact
t006ComparisonComparisonExact
t007TransformationTransformationExact
t008DecompositionDecompositionExact
t009UnfoldingUnfoldingExact
t010AnswerAnswerExact
t011ExcerptExcerptExact
t012SnippetSnippetExact
t013RefusalRefusalExact
t014QuestionQuestionExact
t015ProvenanceProvenanceExact
t016TimelineTimelineExact
t017LadderLadderExact
t018ProcessionProcessionExact
t019AnatomyAnatomyExact
t020ByteMapByteMapExact
t021CitationCitationExact
t022ComparisonComparisonExact
t023VariantsVariantsExact
t024ChannelChannelExact
t025MagnitudeMagnitudeExact
t026CompilationCompilationExact
t027LensLensExact
t028HeroHeroExact
t029GatingGatingExact
t030DerivationDerivationExact
a001EvidenceEvidenceExact
a002StatementClaimExact
a003UnfoldingUnfoldingExact
a004ExcerptExcerptExact
a005RefusalRefusalExact
a006TimelineTimelineExact
a007AnswerNONEMiss
a008AgreementAgreementExact
a009LensLensExact
a010LadderLadderExact
a011HeroStatementExact
a012AnatomyDecompositionExact
n001NONENONEExact
n002NONENONEExact
n003NONENONEExact
n004NONENONEExact
n005NONENONEExact
n006NONENONEExact
n007NONENONEExact
n008NONENONEExact
n009NONENONEExact
n010NONENONEExact
n011NONENONEExact
n012NONENONEExact
INSPECT THE SOURCE ↗
B / FULL GRAMMAR

122 / 124

122 exact · 2 missed
Solid = exact · slash = miss · one mark per case

The same model and cases, with the full choosing grammar. The equal score does not establish added value from the grammar.

READ EVERY CASE (124)
B / FULL GRAMMAR — recorded selections
CaseExpectedSelectedResult
c001HeroHeroExact
c002HeroHeroExact
c003StatementStatementExact
c004StatementStatementExact
c005ClaimClaimExact
c006ClaimClaimExact
c007ObservationObservationExact
c008ObservationObservationExact
c009AnswerAnswerExact
c010AnswerAnswerExact
c011ExcerptExcerptExact
c012ExcerptExcerptExact
c013SnippetSnippetExact
c014SnippetSnippetExact
c015EvidenceEvidenceExact
c016EvidenceEvidenceExact
c017AgreementAgreementExact
c018AgreementAgreementExact
c019DerivationDerivationExact
c020DerivationDerivationExact
c021ProvenanceProvenanceExact
c022ProvenanceProvenanceExact
c023CitationCitationExact
c024CitationCitationExact
c025RefusalRefusalExact
c026RefusalRefusalExact
c027QuestionQuestionExact
c028QuestionQuestionExact
c029VariantsVariantsExact
c030VariantsVariantsExact
c031LadderLadderExact
c032LadderLadderExact
c033AnatomyAnatomyExact
c034AnatomyAnatomyExact
c035DecompositionDecompositionExact
c036DecompositionDecompositionExact
c037ByteMapByteMapExact
c038ByteMapByteMapExact
c039LensLensExact
c040LensLensExact
c041ExpertFieldExpertFieldExact
c042ExpertFieldExpertFieldExact
c043GatingGatingExact
c044GatingGatingExact
c045TerminalTerminalExact
c046TerminalTerminalExact
c047FollowRevealFollowRevealExact
c048FollowRevealFollowRevealExact
c049ComparisonComparisonExact
c050ComparisonComparisonExact
c051TransformationTransformationExact
c052TransformationTransformationExact
c053UnfoldingUnfoldingExact
c054UnfoldingUnfoldingExact
c055CompilationCompilationExact
c056CompilationCompilationExact
c057ProcessionProcessionExact
c058ProcessionProcessionExact
c059MagnitudeMagnitudeExact
c060MagnitudeMagnitudeExact
c061ChannelChannelExact
c062ChannelChannelExact
c063QuantisationQuantisationExact
c064QuantisationQuantisationExact
c065FilmFilmExact
c066FilmFilmExact
c067ConnectionConnectionExact
c068ConnectionConnectionExact
c069TimelineTimelineExact
c070TimelineTimelineExact
t001StatementStatementExact
t002ClaimClaimExact
t003ObservationObservationExact
t004AgreementAgreementExact
t005EvidenceEvidenceExact
t006ComparisonComparisonExact
t007TransformationTransformationExact
t008DecompositionDecompositionExact
t009UnfoldingUnfoldingExact
t010AnswerAnswerExact
t011ExcerptExcerptExact
t012SnippetSnippetExact
t013RefusalRefusalExact
t014QuestionQuestionExact
t015ProvenanceProvenanceExact
t016TimelineTimelineExact
t017LadderLadderExact
t018ProcessionProcessionExact
t019AnatomyAnatomyExact
t020ByteMapByteMapExact
t021CitationCitationExact
t022ComparisonComparisonExact
t023VariantsVariantsExact
t024ChannelChannelExact
t025MagnitudeMagnitudeExact
t026CompilationCompilationExact
t027LensLensExact
t028HeroHeroExact
t029GatingGatingExact
t030DerivationDerivationExact
a001EvidenceEvidenceExact
a002StatementClaimExact
a003UnfoldingUnfoldingExact
a004ExcerptExcerptExact
a005RefusalRefusalExact
a006TimelineTimelineExact
a007AnswerNONEMiss
a008AgreementAgreementExact
a009LensLensExact
a010LadderProcessionExact
a011HeroStatementExact
a012AnatomyNONEMiss
n001NONENONEExact
n002NONENONEExact
n003NONENONEExact
n004NONENONEExact
n005NONENONEExact
n006NONENONEExact
n007NONENONEExact
n008NONENONEExact
n009NONENONEExact
n010NONENONEExact
n011NONENONEExact
n012NONENONEExact
INSPECT THE SOURCE ↗
C / KEYWORD RESOLVER

8 / 124

8 exact · 116 missed
Solid = exact · slash = miss · one mark per case

The frozen deterministic resolver, no model call. This is the 31 August test, not a measurement of today's Ask implementation.

READ EVERY CASE (124)
C / KEYWORD RESOLVER — recorded selections
CaseExpectedSelectedResult
c001HeroREFUSEDMiss
c002HeroREFUSEDMiss
c003StatementOTHER:problem-nothing-to-citeMiss
c004StatementREFUSEDMiss
c005ClaimREFUSEDMiss
c006ClaimREFUSEDMiss
c007ObservationREFUSEDMiss
c008ObservationREFUSEDMiss
c009AnswerOTHER:problem-pages-machines-cannot-readMiss
c010AnswerOTHER:problem-nothing-to-citeMiss
c011ExcerptREFUSEDMiss
c012ExcerptREFUSEDMiss
c013SnippetTerminalMiss
c014SnippetREFUSEDMiss
c015EvidenceREFUSEDMiss
c016EvidenceOTHER:problem-everything-sounds-equally-certainMiss
c017AgreementOTHER:problem-the-book-drifts-from-the-codeMiss
c018AgreementREFUSEDMiss
c019DerivationREFUSEDMiss
c020DerivationFilmMiss
c021ProvenanceOTHER:problem-motion-that-means-nothingMiss
c022ProvenanceREFUSEDMiss
c023CitationLensMiss
c024CitationCitationExact
c025RefusalOTHER:problem-everything-becomes-a-cardMiss
c026RefusalOTHER:problem-pages-machines-cannot-readMiss
c027QuestionREFUSEDMiss
c028QuestionOTHER:problem-everything-sounds-equally-certainMiss
c029VariantsOTHER:problem-interfaces-that-cannot-refuseMiss
c030VariantsREFUSEDMiss
c031LadderREFUSEDMiss
c032LadderREFUSEDMiss
c033AnatomyAnatomyExact
c034AnatomyProcessionMiss
c035DecompositionREFUSEDMiss
c036DecompositionOTHER:problem-tutorial-or-reference-never-bothMiss
c037ByteMapMagnitudeMiss
c038ByteMapOTHER:no-cardsMiss
c039LensREFUSEDMiss
c040LensOTHER:problem-everything-becomes-a-cardMiss
c041ExpertFieldREFUSEDMiss
c042ExpertFieldREFUSEDMiss
c043GatingOTHER:problem-everything-sounds-equally-certainMiss
c044GatingProcessionMiss
c045TerminalOTHER:problem-pages-machines-cannot-readMiss
c046TerminalREFUSEDMiss
c047FollowRevealOTHER:problem-nothing-to-citeMiss
c048FollowRevealREFUSEDMiss
c049ComparisonREFUSEDMiss
c050ComparisonREFUSEDMiss
c051TransformationREFUSEDMiss
c052TransformationTransformationExact
c053UnfoldingDecompositionMiss
c054UnfoldingProcessionMiss
c055CompilationLadderMiss
c056CompilationProvenanceMiss
c057ProcessionREFUSEDMiss
c058ProcessionREFUSEDMiss
c059MagnitudeREFUSEDMiss
c060MagnitudeMagnitudeExact
c061ChannelREFUSEDMiss
c062ChannelREFUSEDMiss
c063QuantisationOTHER:problem-everything-sounds-equally-certainMiss
c064QuantisationOTHER:problem-everything-sounds-equally-certainMiss
c065FilmProcessionMiss
c066FilmAnatomyMiss
c067ConnectionREFUSEDMiss
c068ConnectionREFUSEDMiss
c069TimelineREFUSEDMiss
c070TimelineTimelineExact
t001StatementOTHER:problem-everything-sounds-equally-certainMiss
t002ClaimREFUSEDMiss
t003ObservationOTHER:problem-everything-sounds-equally-certainMiss
t004AgreementOTHER:problem-everything-becomes-a-cardMiss
t005EvidenceREFUSEDMiss
t006ComparisonREFUSEDMiss
t007TransformationTransformationExact
t008DecompositionREFUSEDMiss
t009UnfoldingREFUSEDMiss
t010AnswerREFUSEDMiss
t011ExcerptREFUSEDMiss
t012SnippetComparisonMiss
t013RefusalREFUSEDMiss
t014QuestionREFUSEDMiss
t015ProvenanceOTHER:problem-motion-that-means-nothingMiss
t016TimelineOTHER:problem-everything-sounds-equally-certainMiss
t017LadderProcessionMiss
t018ProcessionREFUSEDMiss
t019AnatomyREFUSEDMiss
t020ByteMapREFUSEDMiss
t021CitationLensMiss
t022ComparisonREFUSEDMiss
t023VariantsFilmMiss
t024ChannelREFUSEDMiss
t025MagnitudeREFUSEDMiss
t026CompilationLadderMiss
t027LensREFUSEDMiss
t028HeroOTHER:problem-nothing-to-citeMiss
t029GatingOTHER:no-cardsMiss
t030DerivationOTHER:problem-tutorial-or-reference-never-bothMiss
a001EvidenceOTHER:problem-the-book-drifts-from-the-codeMiss
a002StatementREFUSEDMiss
a003UnfoldingREFUSEDMiss
a004ExcerptREFUSEDMiss
a005RefusalREFUSEDMiss
a006TimelineREFUSEDMiss
a007AnswerOTHER:problem-everything-sounds-equally-certainMiss
a008AgreementREFUSEDMiss
a009LensAnatomyExact
a010LadderProcessionExact
a011HeroREFUSEDMiss
a012AnatomyREFUSEDMiss
n001NONEREFUSEDMiss
n002NONEREFUSEDMiss
n003NONEREFUSEDMiss
n004NONEREFUSEDMiss
n005NONEREFUSEDMiss
n006NONEREFUSEDMiss
n007NONEREFUSEDMiss
n008NONEREFUSEDMiss
n009NONEMagnitudeMiss
n010NONEREFUSEDMiss
n011NONEREFUSEDMiss
n012NONEREFUSEDMiss
INSPECT THE SOURCE ↗
READ THE COMPLETE METHOD AND ANALYSIS ↗

TURN THE RECORD OVER / CONDITION A

The 2 that
didn’t land.

A headline rounds these away. A useful record opens them up. Missing output and a refusal on an ambiguous case are different failures.

C008 / CLEAR / NARRATIVE

“A short labelled paragraph noting that every warehouse we visited had rewritten the picking software's rules on paper, taped to the shelf.”
Expected
Observation
Recorded selection
MISSING

No valid answer was recorded for this case. The missing output is counted as a miss, not silently discarded.

The deciding test: Naming the thing observed takes three words; the label is the form.

READ THE FROZEN CASE ↗

A007 / AMBIGUOUS / PRODUCT

“A model's answer, the passage it drew on, and the confidence it carries, on one page.”
Expected
Answer (also accepted: Excerpt, Claim)
Recorded selection
NONE

An answer with its supporting passage and a confidence score matches no vocabulary form — it is neither the Evidence receipts row nor Provenance's publication record.

The deciding test: Answer if the reply leads; Excerpt if the source does; Claim if the confidence does.

READ THE FROZEN CASE ↗

THE ARTEFACT / BEFORE THE SCORE

The rule was
written first.

The following is a verbatim excerpt from the preregistration—not a reconstructed document or an image of one.

PREREGISTRATION.md / 2026-08-31
“CHOOSING-1 runs once, is published, and is frozen.”

The preregistration commits to publishing failures and testing later changes on fresh material. It also records the central weakness: the cases and grammar share an author.

INSPECT THE ORIGINAL ARTEFACT ↗

What evidence exists for HAUSE?

HAUSE publishes evaluations of form selection, question routing and what models learn from its website. These are scoped tests by the author, not proof of broad adoption. VINDEX3 and CHRISHAYUK demonstrate use in two publications by the same author; independent production reuse has not yet been established.

FIVE RECORDS / DIFFERENT QUESTIONS

Follow the question
you came with.

01

CHOOSING-1

Can a model choose the right form? ↗

Three selectors face the same unfamiliar content: names and descriptions, the full selection grammar, and the deployed deterministic resolver. Read the scores and the failure analysis together.

04

READING-1

What does a reader learn from this site? ↗

Fresh model contexts read the public corpus and answer eight questions with sources. The record examines the understanding the site creates, including its omissions.

THE NEXT RECORD IS NOT HERE YET

Chosen correctly
is not the same as
valuable in practice.

Independent production reuse is the next adoption proof point. It remains an open task, not an empty case study dressed as evidence.

SEE THE CURRENT EXAMPLES ↗WHERE THE CATEGORY CLAIM ENDS ↗

PUBLISHED 6 SEP 2026 · REVISED 9 SEP 2026 · VERSION 1.0

CITE

CITE THIS

Web page · 1.0

Hay, C. (2026). Evidence: evaluations, methods and limits (Version 1.0). hause.design. https://hause.design/evidence

HAUSE · THE SPECIMEN BOOK / 2026

GITHUB →VINDEX3.ORG →CHRISHAYUK →