當讀者不再是人類：制度證據為何需要機器可讀基礎設施

搜尋系統、自動研究工具與 AI，正日益在人工讀者之前參與制度資訊的發現、擷取與解讀。EIA 011 指出，基礎設施問題已不只是資訊能否被找到，而是制度意義能否穿越擷取過程後仍被正確理解。 當搜尋、研究代理與大型語言模型在人工讀者之前接觸制度資訊，發布成功的定義便改變了。內容能被抓取只是起點；機器還必須辨識這是法規、標準、草案、諮詢回覆或編輯分析，並知道由誰發布、何時有效、與哪些版本及文件相關。否則可讀文字會在離開頁面後失去制度身分。 本期 Analysis 因而不把 制度資訊跨越機器擷取邊界時的證據保存 視為單一技術或政策更新，而是把官方主錨點可證明的制度事實、SNN 的分析推論與仍待驗證的結果分層處理。讀者必須能看見主張從何而來、推論在哪裡開始，以及哪些結論目前不能由來源直接支持。 為使摘要能直接支援決策，本段同時回答五個問題：已發生什麼、由哪一第一方紀錄支持、影響透過什麼機制傳導、目前仍缺少哪些結果證據，以及下一個可推翻或強化判斷的檢查點。任何未被來源明示的因果關係都保留為編輯推論，不以肯定語氣包裝。

Search systems, automated research tools and AI increasingly participate in the discovery, retrieval and interpretation of institutional information before it reaches a human reader. The resulting challenge is not simply whether information can be found. It is whether its institutional meaning can survive retrieval. When search, research agents and large language models encounter institutional information before a human reader, the definition of successful publication changes. Retrievability is only the starting point. A machine must distinguish a regulation, standard, draft, consultation response and editorial analysis, and know who issued it, when it applies and how it relates to versions and other documents. Otherwise readable text loses institutional identity after leaving the page. This Analysis therefore does not treat the preservation of institutional evidence across machine-retrieval boundaries as a self-contained technical or policy update. It separates the institutional facts supported by the official anchor, SNN editorial inference and outcomes that remain unverified. The reader should be able to see where the source ends, where interpretation begins and which conclusions the present evidence cannot support. To make the lead decision-ready, it answers five questions together: what has occurred, which first-party record supports it, through what mechanism the effect may travel, which outcome evidence is still missing, and what next observation could strengthen or overturn the judgement. Any causal relationship not stated by the source remains an editorial inference and is not converted into a factual claim through confident wording.

傳統制度網站以人類閱讀為前提。讀者會從機構品牌、頁面階層、日期、導航與相關文件重建脈絡，區分監管機構與評論者、草案與定稿、現行版本與歷史版本，以及一手證據與二手解讀。 人類會從網址、導覽、機構標誌、版面與相關連結補足語境，機器擷取則常把單一段落或中繼資料帶離原網站。W3C PROV-O、DCAT、DataCite、Crossref 與 FAIR 等環境分別提供來源、目錄、持久識別、關係與機器可行動性的構件；它們不能替內容作出權威判斷，但能保存判斷所需條件。 為了維持分析密度，制度背景必須同時標示規則制定者、執行者、資料擁有人、覆核者與受影響市場。這些角色可能由不同組織或部門承擔；文件發布、系統上線、企業採用與結果交付也不是同一證據狀態。 時間與權限也必須分開記錄：公告日不等於生效日，試點不等於普遍採用，技術規格不等於法律義務，企業自願導入也不等於監理核准。把這些節點放在同一時間軸，才能判斷一項制度變動何時真正進入資料、合約、投資或揭露流程。

The institutional web has traditionally relied on human readers to reconstruct context. They recognize institutional branding, infer hierarchy from page structure, distinguish a regulator from a commentator, compare publication dates, and understand that analysis about a standard is not the standard itself. Much of this context exists around the information object rather than inside it. Humans reconstruct context from URLs, navigation, institutional marks, layout and related links. Machine retrieval often removes a paragraph or metadata from the publishing environment. W3C PROV-O, DCAT, DataCite, Crossref and FAIR environments provide components for provenance, cataloguing, persistent identity, relationships and machine actionability. They do not make the authority judgement, but they can preserve the conditions needed to make it. Institutional context must identify the rule setter, implementer, data owner, reviewer and affected market. Those roles may sit in different organisations or functions, and publication of a document, deployment of a system, enterprise adoption and delivery of an outcome are different evidence states. Time and authority must also be separated. An announcement date is not an effective date; a pilot is not general adoption; a technical specification is not a legal obligation; and voluntary enterprise use is not regulatory approval. Putting these events on one timeline shows when an institutional development actually enters data, contract, investment or disclosure processes and which actor is authorised to make that transition.

機器擷取會讓資訊離開原本的發布環境，形成 Retrieval Boundary。當文字留下，但來源、類型、狀態、版本、權威與文件關係沒有一同保留時，內容即使仍正確，也可能產生 Evidence Ambiguity，成為制度上不完整的證據。 Retrieval Boundary 的風險在於文字內容仍正確，證據功能卻變得模糊。草案可能被當成現行規則，研究結論可能被當成機構立場，舊版本可能在搜尋排序中超過定稿。若 canonical、document type、status、version、issued by 與 relation 沒有一同傳遞，AI 只能依語句表面推測制度意義。 上游證據若沒有穩定識別、形成時間、適用邊界與版本，最終輸出即使格式一致，覆核者仍可能無法重建形成過程。真正風險不是單一欄位缺失，而是錯誤主體、過期方法、推定關係或未核准版本在傳遞中被當成確定事實。 具體失效模式至少包括主體配對錯誤、資料人口不完整、邊界前後不一致、方法或係數未版本化、例外未留下理由、核准與發布時間倒置，以及下游重用時脫離原用途。每一種失效都可能讓合理的單筆資料，在彙總後形成無法防禦的結論。

The environment in which information is published may no longer be the environment in which it is encountered. This creates the Retrieval Boundary. If only the text survives, source, publication type, version relationships and status may disappear. Evidence Ambiguity arises when the information required to interpret institutional significance no longer travels with the evidence. The Retrieval Boundary creates risk when text remains correct while its evidential function becomes ambiguous. A draft may be treated as a current rule, a research conclusion as an institutional position, or an old version may outrank a final document in search. Without canonical identity, document type, status, version, issuing authority and relationships, an AI system must infer institutional meaning from surface language. When upstream evidence lacks stable identity, formation time, applicable boundary and version, a standardised output may still be impossible to reconstruct. The material risk is not one missing field. It is the silent conversion of the wrong entity, an expired method, an inferred relationship or an unapproved version into an apparent fact as information moves downstream. Concrete failure modes include incorrect entity matching, an incomplete data population, inconsistent boundaries, unversioned methods or factors, exceptions without rationale, approval occurring after publication, and downstream reuse outside the original purpose. Each failure can turn a reasonable individual record into a conclusion that cannot be defended after aggregation, comparison or machine-assisted interpretation.

EIA 011 將文件重新界定為 Evidence Object，要求其攜帶身分、來源、類型、時間、狀態、版本、關係、可追溯性與權威。制度資訊需依序具備可發布、可發現、機器可讀、機器可解讀、可追溯、制度脈絡化與證據保存等能力。 Evidence Object 應至少攜帶持久識別、canonical URL、發布與修改日期、發行機構、作者、文件類型、狀態、語言、版本、前後關係、引用來源與權利資訊。內容更新不應靜默覆寫；取代、修正、翻譯與摘要各自形成可機器辨識的關係，讓擷取系統可以回到權威版本。 可執行的控制單位應是受治理的證據物件：每項重要主張連接原始來源、計算或判斷方法、組織與時間邊界、責任人、控制狀態、例外、核准及版本。當任何元件改變時，系統保存差異與影響範圍，不以覆寫舊檔取代變更紀錄。 營運上可建立最小控制集：主張登錄、證據擁有人、來源快照、方法識別、適用期間、控制頻率、例外門檻、覆核與核准，以及允許的下游用途。高判斷或高財務影響項目採更嚴格的覆核層級；低風險資料則以自動化完整性檢查降低重工。

An Evidence Object carries identity, origin, type, time, status, version, relationship, provenance and authority. Evidence-Preserving Infrastructure connects persistent evidence identity, explicit evidence type, structured metadata, provenance, version and relationship management, canonical discovery and evidence-preserving retrieval. The progression is Published, Discoverable, Machine-Readable, Machine-Interpretable, Traceable, Institutionally Contextualized and Evidence-Preserving. An Evidence Object should carry persistent identity, canonical URL, publication and modification dates, issuing institution, author, document type, status, language, version, predecessor and successor relationships, cited sources and rights. Updating must not silently overwrite history. Replacement, correction, translation and summary each create machine-readable relationships that allow retrieval systems to return to the authoritative version. The implementable control unit is a governed evidence object. Each material claim links to its primary source, calculation or judgement method, organisational and temporal boundary, accountable owner, control state, exception, approval and version. When any component changes, the system preserves the difference and affected uses instead of overwriting the earlier basis. A minimum operating control set includes a claim register, evidence owner, source snapshot, method identity, valid period, control frequency, exception threshold, review, approval and permitted downstream use. High-judgement or high-financial-impact items receive a stronger review tier. Lower-risk records use automated completeness and consistency checks so that governance effort is concentrated where a wrong claim would change a decision.

本文以 IFRS Sustainability Standards Navigator、International Trade Centre Standards Map、European Commission Joint Research Centre Publications Repository，以及來源追溯與機器可讀永續知識基礎設施的相關研究環境作為制度觀察基礎。這些環境共同顯示，搜尋與結構化發布正在重新配置證據的發現方式。 現有標準與平台證明的是來源與關係可以被結構化，不證明所有 AI 一定正確使用它們，也不保證中繼資料本身真實。本文所稱 Evidence-Preserving Infrastructure，是在既有技術構件上提出的制度治理層；其效果仍取決於發布者維護、擷取者尊重與使用者覆核。 證據判讀以本期官方主錨點為起點，再以獨立第一方或具方法透明度的來源確認背景與邊界。來源能支持的是已發布的制度事實與明確數據；是否代表全面採用、因果改善或跨市場普遍有效，必須另有實作結果才能成立。 每一項關鍵句都應進入 claim ledger，標示其為官方事實、直接量測、估算、企業承諾、已交付結果或 SNN 編輯推論，並記錄來源所能支持的精確範圍。若兩筆來源衝突，保留衝突與處理理由；若資料尚未出現，明確標成待驗證，而不是以相近案例補成確定答案。

This publication is informed by institutional publication, standards navigation, persistent repositories, structured metadata, provenance and machine-readable information environments. Relevant environments include the IFRS Sustainability Standards Navigator, the International Trade Centre Standards Map, the European Commission Joint Research Centre Publications Repository, and emerging research on provenance-aware sustainability knowledge infrastructures. Existing standards and platforms demonstrate that provenance and relationships can be structured. They do not prove that every AI system will use them correctly or that metadata are necessarily true. Evidence-Preserving Infrastructure is an institutional governance layer proposed on top of technical components; its effectiveness still depends on publisher maintenance, retriever respect and user review. Evidence assessment begins with the official anchor and uses independent primary or method-transparent sources to test context and limits. The sources support stated institutional facts and explicit figures. Claims of comprehensive adoption, causal improvement or universal cross-market effectiveness require separate implementation evidence. Every material sentence should enter a claim ledger and be classified as official fact, direct measurement, estimate, corporate commitment, delivered outcome or SNN editorial inference. The ledger records the precise scope that each source supports. Conflicting evidence is retained with the resolution rationale; absent evidence is marked pending rather than filled with a convenient analogue from another entity, period or jurisdiction.

可搜尋不等於可正確解讀，機器可讀也不等於制度身分清楚。搜尋最佳化提高資訊被找到的機率，Evidence-Preserving Infrastructure 則提高資訊在被找到後仍能被正確理解的機率。未保存關係，就可能在內容完整的情況下遺失一部分證據。 這會把 SEO、機器可讀性與證據治理分成三個層次：SEO 提高發現機率，結構化資料提高辨識能力，來源與版本治理保存制度意義。只做前兩層可能讓錯誤版本更容易被找到；只有第三層而缺乏可發現性，權威內容又可能被二手解讀取代。 這種資訊密度的目的不是增加篇幅，而是縮短主張到決策之間的查核距離。董事會、投資人、監理者與營運團隊應能辨識哪些內容是事實、估算、承諾、進度或結果，並在情境改變時更新判斷，而不是重新尋找散落的檔案與口頭說明。 責任分工因此必須落到決策權：資料擁有人維護來源，方法擁有人控制計算，業務單位說明使用情境，內控或確信人員測試可重建性，核准者承擔最終採用責任。例外若沒有到期日、補救人與影響範圍，就會從暫時處置變成永久的證據債務。

A publication can be discoverable without being correctly interpretable. It can be machine-readable without being institutionally identifiable, retrieved without preserving provenance, and cited without distinguishing primary evidence from secondary interpretation. Search optimization improves the probability that information will be found. Evidence-Preserving Infrastructure improves the probability that found information will still be understood correctly. This separates SEO, machine readability and evidence governance into three layers. SEO increases discovery, structured data increases recognition, and provenance and version governance preserve institutional meaning. The first two without the third can make the wrong version easier to find. The third without discoverability can leave authoritative material displaced by secondary interpretation. The purpose of this information density is not length for its own sake. It is to shorten the verification distance between claim and decision. Boards, investors, regulators and operational teams should be able to distinguish fact, estimate, commitment, progress and outcome, then update the judgement when conditions change without reconstructing the case from scattered files and oral explanation. Accountability therefore attaches to decision rights. The data owner maintains the source, the method owner controls calculation, the business function defines the use case, internal control or assurance tests reproducibility, and the approver accepts responsibility for final use. An exception without an expiry date, remediation owner and impact scope stops being temporary treatment and becomes persistent evidence debt.

SNN 編輯分析：歐洲與全球制度正朝結構化資料與 AI 輔助審閱前進，台灣資本市場也已導入格式化永續專章與 ESG 數位申報。台灣企業若只提供 PDF 敘事而沒有穩定識別碼、欄位定義、版本、來源與責任關係，證據經機器處理後就可能失去原意；語義必須在揭露前保存。 台灣上市櫃公司、主管機關與研究機構正增加數位永續揭露，但 PDF、新聞稿與資料欄位常缺乏穩定關係。台灣企業若要讓國際 AI、投資人與歐洲客戶正確理解，應為中英文內容共用主張識別，標示原始語言、翻譯、版本、責任人與適用期間，並把揭露欄位連回受控來源，而不是只增加可被爬取的文字。 對台灣市場的意義應沿實際傳導路徑判斷：國際制度或客戶要求先進入融資、採購、合約、供應商資料與確信流程，再影響台灣企業的系統與控制。這不等於外國規則自動成為台灣法律義務；本地企業要做的是辨識適用情境、保留跨語言對應並建立可授權覆核的證據鏈。 台灣企業可把傳導測試落在具體物件：公司與法律主體、廠區、產品、批次、供應商、合約、融資工具及揭露欄位。中英文名稱、內外部分類與不同框架之間應連到同一主張識別，同時保留商業機密、個資與權限邊界，避免可追溯被誤解為全面公開。

SNN editorial analysis: European and global institutions are moving toward structured data and AI-assisted review, while Taiwan capital markets are introducing formatted sustainability chapters and ESG digital filing. Taiwan companies need stable identifiers, field definitions, versions, provenance and responsibility links so machine processing does not strip evidence of meaning. Taiwan listed companies, authorities and research institutions are increasing digital sustainability disclosure, while PDFs, news releases and data fields often lack stable relationships. To be understood correctly by international AI, investors and European customers, Chinese and English content should share claim identities, identify original language, translation, version, owner and valid period, and link disclosure fields to controlled sources rather than merely add crawlable text. For Taiwan, relevance should be traced through an actual transmission path. An international rule or customer requirement first enters finance, procurement, contract, supplier-data and assurance processes, then changes local systems and controls. It does not automatically become Taiwan law. Companies need to identify the applicable scenario, preserve bilingual mappings and make the evidence chain reviewable under controlled access. Taiwan companies can perform the transmission test on concrete objects: the company and legal entity, facility, product, batch, supplier, contract, financing instrument and disclosure field. Chinese and English names, internal and external classifications and different reporting frameworks should resolve to the same claim identity. Traceability must still preserve commercial confidentiality, personal data and access boundaries; it does not require unrestricted publication.

未來制度網站需要成為 Evidence Environment，使身分持續存在、來源可追溯、版本可區分、關係可重建、證據類型保持分離，並讓制度意義在跨越人類與機器擷取環境後仍可理解。挑戰不是生產更多內容，而是讓證據攜帶理解自身所需的條件。 後續成熟度不應用頁面數或 API 數量衡量，而要測試機器能否區分權威與評論、定稿與草案、現行與歷史。低後悔行動是對一份重要制度文件進行 machine retrieval red-team：從搜尋結果、API 與 AI 摘要逆向檢查身分、狀態、版本及來源是否仍可重建。 後續追蹤應分開最終文件、技術指引、採用範圍、執行控制、監督活動與可觀察結果。現階段的低後悔行動，是挑選一項高風險主張做端到端重建測試，記錄缺少的身分、來源、方法、責任與版本；這是治理準備，不是合規保證或結果預測。 監測不只依日曆，而應由事件觸發：最終規則發布、技術指引變更、適用範圍擴大、主管機關執法、企業採用數據或實際成果出現時，都建立新版本並重評原判斷。舊結論不刪除，而是標示當時依據、被何種新證據修正，以及哪些決策需要重新檢視。

The future institutional web may need to function as an Evidence Environment in which identity persists, provenance remains traceable, versions remain distinguishable, relationships remain reconstructable, evidence classes remain separate and institutional meaning survives retrieval. The next challenge may not be producing more content. It may be ensuring that evidence carries the conditions required to interpret it correctly. Maturity should not be measured by page or API count, but by whether a machine can distinguish authority from commentary, final from draft and current from historical. A low-regret action is to conduct a machine-retrieval red team on one important institutional document, tracing search results, API output and AI summary back to determine whether identity, status, version and sources remain reconstructable. Future monitoring should separate final text, technical guidance, adoption scope, operating controls, supervision and observable outcomes. A low-regret step is to select one high-risk claim for an end-to-end reconstruction test and record missing identity, source, method, accountability and version. That is governance preparation, not a compliance guarantee or forecast of results. Monitoring should be event-triggered as well as calendar-based. A final rule, amended technical guidance, expanded scope, supervisory action, adoption data or observed outcome creates a new version and a reassessment of the earlier judgement. The prior conclusion is not erased. It retains its original basis, identifies the new evidence that changed it and states which decisions or downstream uses now require review.