Summary

  • The WTPF-26 Chair's report says five Working Groups revised and endorsed the draft Opinions and that the Forum adopted all five as presented in Part II on 4 September 2026.
  • The last public temporary documents for Opinions 1–3 are dated 3 September; those for Opinions 4–5 are dated 4 September. The five standalone final Opinions are dated 4 September.
  • Every draft/final DOCX pair has different bytes and a different file hash. Plain-text comparison finds only a new document number, a date change where applicable and removal of the DRAFT label.
  • Once those administrative header fields are removed symmetrically, all five pairs have identical text hashes. The finding does not reach earlier Working Group versions, oral debate or participants' private preferences.
  • ITU should publish an adoption-delta receipt that binds each final Opinion to its last public predecessor and records authority change and substantive text change as separate facts.

Finality arrived as a new document number

The WTPF-26 final-document register now lists five standalone Opinions. They are numbered WTPF-26/5-E through WTPF-26/9-E and dated 4 September. The Chair's report records the institutional act behind those labels: five Working Groups considered the drafts, their chairs presented the results to Plenary, and the Forum adopted Opinions 1 through 5 as presented in Part II.

That act matters. Before it, a temporary document recorded wording proposed for the Forum's decision. After it, a final document recorded wording adopted by the Forum. The final file does not need a fresh sentence to acquire that status. Authority can move even when language stays still.

The public interface nevertheless makes the reader reconstruct the transition. The temporary documents live in one register; the final Opinions live in another. Their filenames, package sizes and hashes change. Nothing on the individual final record says which temporary document was its immediate predecessor or whether the policy language changed at the handoff.

Five pairs, one result

The comparison used the English Word files exposed by the two official registers. Each DOCX was converted to plain text. The document-number line, document-date line and DRAFT status token were then removed from both sides under the same rule. The Opinion number, title and every character after the header were preserved.

Opinion Last public temporary document Final document Administrative change Substantive text delta
1 · Bridging digital divides WTPF-26/DT/1-E · 3 Sep WTPF-26/5-E · 4 Sep Number, date, draft label 0 added / 0 removed
2 · Green digital transformation WTPF-26/DT/2-E · 3 Sep WTPF-26/6-E · 4 Sep Number, date, draft label 0 added / 0 removed
3 · Resilience of telecommunications/ICTs WTPF-26/DT/3-E · 3 Sep WTPF-26/7-E · 4 Sep Number, date, draft label 0 added / 0 removed
4 · Space connectivity WTPF-26/DT/4-E · 4 Sep WTPF-26/8-E · 4 Sep Number, draft label 0 added / 0 removed
5 · ICT-centric innovation ecosystems WTPF-26/DT/5-E · 4 Sep WTPF-26/9-E · 4 Sep Number, draft label 0 added / 0 removed

The normalized hashes are different across the five Opinions, as they should be: these are five different texts. Within each predecessor/final pair, however, the normalized hash is identical. That is a stronger and narrower statement than saying that the files “look the same.” It defines the fields excluded, preserves the operative text and makes the result reproducible.

A changed file is not necessarily changed policy

All five original DOCX pairs have different SHA-256 hashes. Opinion 1's temporary package contains 122,914 bytes and its final package 121,553. The other four pairs also change size. A system that watches only file hashes would correctly announce five new digital objects. It would be wrong to convert that observation automatically into five substantive amendments.

Office files contain more than the sentences a reader sees. Styles, internal XML, package relationships, timestamps and other metadata can change. Here the extracted text also changes, but only in fields that describe the document: WTPF-26/DT/1-E becomes WTPF-26/5-E, 3 September becomes 4 September for the first three pairs, and DRAFT OPINION becomes OPINION in all five.

Those are not disposable differences. They establish identity, chronology and status. They tell a reader that the Forum's act occurred. They simply do not alter what the Opinions say about digital divides, environmental sustainability, resilience, space connectivity or innovation ecosystems.

The distinction works in both directions. Calling the handoff “no change” would erase adoption. Calling every new hash a “revised policy” would invent wording that is not in the record. The accurate account has two columns: authority delta, yes; substantive text delta from the last public draft, no.

The Chair's report preserves the earlier boundary

The Chair's report says the Working Groups worked over two and a half days and that the drafts presented to Plenary had been “revised and endorsed” by those groups. The zero-delta result therefore cannot be projected backwards to the documents that entered the Forum, the Secretary-General's report or earlier IEG drafts.

The last public temporary documents may already incorporate Working Group changes. Opinions 1–3 carry 3 September dates; Opinions 4–5 carry 4 September dates. The record supports a comparison from those five snapshots to the final standalone files. It does not support the larger claim that deliberation produced no textual movement.

Nor does matching text reveal how much disagreement preceded the final state. A sentence can survive because everyone favoured it, because objections were answered, because compromise occurred earlier, because no amendment commanded consensus or because the remaining differences were accepted. A final-draft comparison cannot distinguish those histories. The Chairman's report supplies disposition; it is not a roll-call map of belief.

That evidence boundary matters more than a dramatic headline. A small, fully reproducible finding is more useful than an accusation assembled from absence.

Build an adoption-delta receipt

The document system already contains nearly all the needed facts. ITU could join them in a small public receipt for each Opinion.

The first block would identify the last public predecessor: WTPF-26/DT number, source language, publication date and original file hash. The second would identify the final Opinion: C-document number, adoption date, public posting time if available and original file hash. A direct link in both directions would prevent the two registers from becoming separate histories.

The third block would classify the delta. Administrative could list document number, date, status label, layout and package metadata. Substantive could list added, removed or modified provisions, or state none detected under published normalization rule. The rule itself must be published, versioned and narrow enough that no operative sentence can disappear as “formatting.” Both original and normalized hashes should remain available.

The fourth block would name the authority event: WTPF-26 Plenary adoption, its date, the relevant Chair's report and any recorded reservation or separate statement that the procedure treats as part of the disposition. It should say what the receipt does not establish—earlier amendment history, individual assent or downstream implementation.

This is not a demand that ITU make a blockchain out of conference documents. A stable JSON or CSV record beside the human page would be enough. The value comes from the join, not from technological ceremony.

Why zero delta is useful evidence

An unchanged final text can indicate that the last working draft had already absorbed the changes needed for adoption. It can let ministries, companies and researchers cite the final instrument without hunting for a late substantive amendment. It can also focus historical work on the earlier revisions where negotiation actually moved the language.

The result is useful for automated monitoring too. A crawler may detect ten different binaries and infer five revisions. A semantic-delta field would let it distinguish status finalization from operative change. An archive could preserve both artifacts without displaying a false alarm. A later correction could point to the precise final text instead of silently replacing a file.

The same discipline protects institutions. If a final document truly does add a sentence after the last public draft, the receipt would expose the change and identify its authority path. If it does not, the record would stop outsiders from narrating metadata churn as hidden policy-making.

Lu Heng's Policy Mirror offers the relevant discipline: a record reveals what an institution thinks it may control, but the record should not become a source of invented reality. Here the final label truthfully records an act. It should not be asked to imply a textual amendment that the paired evidence disproves.

What this comparison does not prove

It does not prove that every delegate agreed with every line. It does not measure participation quality, bargaining power or whose proposal prevailed. It does not say that the five Opinions are regulations, Recommendations or operating commands. The WTPF objective page describes them as non-prescriptive Opinions adopted by consensus; earlier BTW coverage has already examined that authority boundary.

It also does not authenticate invisible server history. The comparison uses the public files obtained on 5 September. If ITU later replaces one at the same URL, the original hashes recorded with this analysis will reveal that a new binary exists, but only a renewed extraction and comparison can classify the change.

Finally, identical English source text does not prove synchronized publication or equivalence across every language presentation. The temporary register exposed the checked working documents in English. This article therefore makes an English-document finding and no broader one.

The precise conclusion is enough: WTPF-26 adopted five Opinions, and each published final English text matches the last public temporary English draft after administrative identity fields are separated. Finality changed who could attribute the words to the Forum. It did not change those words.

Sources