12 What this record does not tell you

12.1 Why a chapter of limits

Everything before this chapter was what the record says. This chapter is what it does not, and it is the more useful half.

A reader who knows the limits below can quote SA#112 in a briefing, a specification comment or an argument and not be caught out. A reader who does not will eventually write a sentence the paperwork cannot carry.

Each limit comes with the sentence it forbids.

12.2 The record in hand is a draft

The file is SP-112_Draft_Report_v005, version 0.0.5, and its own cover page gives its status as "Draft - for comment" ^[1].

Forbidden: "the official record of SA#112 states". Correct: "the draft report of the meeting records that…".

And part of the record is somewhere else. The chair's final notes are published after the meeting closes — the agenda says so ^[2] — and they are not in this material. Nothing here can say what they contain, or whether they settle something this draft leaves open.

12.3 The agenda is complete and the report is not

The two files disagree about how much of the meeting they cover, and only one of them admits it.

The document list is complete by construction: every document of the meeting is in it, which is why the agenda-item counts add to 346 ^[3] and the report's own annex agrees at 346 ^[4].

The narrative is not. It follows what was said, so an item where nothing was said gets one sentence, forty times over ^[5], and a document nobody discussed gets none.

Forbidden: treating the report's silence as evidence that nothing happened. Correct: checking the document list, where the status word will be there whether or not anybody spoke.

12.4 Sixty-two documents have no story

Only 284 of the meeting's 346 documents get a paragraph of their own ^[6]. The other 62 appear only as a row in an annex table.

Forbidden: "the meeting discussed X" for one of those 62. Correct: "the document list records X as approved", and nothing further.

12.5 It contradicts itself, and it has not been proof-read

Under agenda item 1.5 the narrative says "The report of meeting SA#111 was approved." ^[7] while the status line under the same document says "Not concluded." — and the meeting's database records that document, SP-260362, as approved.

Forbidden: picking whichever of the three reads best. Correct: report them all, or leave the point alone.

The chair's surname appears as both "Jaine" and "Jain" in the same report ^[8]. A spelling in this draft is not authoritative, and neither is any other single word in it that a second pass would have caught.

12.6 What it never contained

Three absences, and none of them can be read around.

• No vote counts, no attendance per company, no reasons. The report records who spoke and what was decided. Where a decision was contested it names one objection ^[9] and says nothing about anybody else's position. Forbidden: "unanimous", "widely supported", "controversial".

• Approved means approved by TSG SA, not published. The meeting approved 26 specifications and reports ^[10], and 30 new work and study items ^[11]. Each still had to be produced by the support team and put on the server afterwards, and nothing here says when. Forbidden: "available since June 2026".

• Nothing about what any of this means for a network operator. No deployment dates, no product impact, no market view. Forbidden: every sentence beginning "this means operators will".

12.7 One thing, two true numbers

689 comes from splitting the specification column of every change request and pack ^[12]. 673 comes from adding up the four release counts, which look only inside packs — 487 for Rel-20 ^[13] and its three neighbours.

Both are right. Forbidden: "689 change requests" with no statement of the question. Correct: say which count you mean, every time.

12.8 What is not evidence

There is a summary of this meeting written by a language model ^[14], and 224 of its documents have a machine-written summary of their own ^[15].

They are useful for finding your way around 346 documents and they are worth citing as summaries. They may never be the source of a claim about what the meeting decided.

Forbidden: any statement about an outcome whose only backing is a generated summary. Correct: use it to find the document, then cite the document.

12.9 What this course cannot give you

This course stands on a register of sources, not on a specification. There is no clause text behind it, so it has no map of a document, no glossary, no version history and no list of related specifications.

It may name a specification and link to it — TS 23.501, TS 23.502, TS 23.482 — and that is the whole of what it can do with one. What any of those documents actually says is a different course.

The same limit applies to the working groups. This record shows how much paper each group sent — SA WG3 sent 57 documents ^[16] and SA WG4 sent 28 ^[17] — and which specifications its change requests touched. It does not describe what any group is for, and nothing here should be read as such a description.

12.10 What the register can and cannot be

This course's evidence is a register of 150 entries, generated from files and databases rather than typed by hand. That buys three things and no more.

• Every figure carries its question. A count entry records the query that produced it and the day it was asked, so a reader can restate it. That is what makes the 689 and its awkward twin possible to explain instead of embarrassing.

• Every link points somewhere checkable. The addresses go to 3gpp.org or to the machine that holds the mirror, and nowhere else, so there is no citation in these chapters that nobody can open.

• Every quotation is character for character. "It was decided to use '6GC' or '6G CN'." ^[18] and the caution about frozen releases ^[19] are the meeting's own words, cut where the register says they were cut.

What it does not buy is completeness. A register holds what somebody thought to extract. The 62 unnarrated documents are in it as rows and not as stories, twenty kinds of document are counted but only ten broken out ^[20], and no entry anywhere records what a decision was for.

12.11 What one meeting is, and is not

A last limit, and it is the widest one. Everything in this course was read from one plenary 3GPP TSG SA#112.

The shape it shows — the agenda, the status words, the packs, the letters — is that meeting's shape. Whether the next plenary looks the same is a question this material cannot answer, and the way to answer it is to open the next one's folder and count.

12.12 Where to go next

Three places, in this order.

• The meeting folder ^[21] — the agenda, the report and one zip file per document number, exactly as 3GPP publishes them. Chapter How to follow a plenary yourself says what to open first.

• This course's sources page — all 150 entries of the register, each with the file or the query it was read from and the day it was read. Every citation in these twelve chapters resolves there.

• The draft report itself ^[1], for any argument you want in the secretary's own words rather than in this course's.

And then the next plenary, which will have its own folder, its own twenty-three-odd agenda items, and its own draft report with its own typographical errors.