11 How to follow a plenary yourself

11.1 The five files

Every 3GPP plenary leaves the same five artefacts, and they are public. For SA#112 they are:

• F1. The agenda — document SP-260361 at version 2 ^[2]. It carries the deadlines, the 23 agenda items ^[3], the hour-by-hour time plan and the meeting's own rules on how to request a document number.

• F2. The document list, the file called TdocsByAgenda ^[1]. Every document under the agenda item it belongs to, with its type, its source and its result.

• F3. The draft report ^[4]. The secretary's narrative, agenda item by agenda item, then the annex tables.

• F4. The documents themselves — one zip file per document number, in the meeting folder ^[5].

• F5. The welcome pack ^[6], which is what the chair tells a first-time delegate about where all of the above lives.

11.2 Where they are

One folder on 3gpp.org, named after the meeting. It holds the agenda, the report and a Docs/ subfolder of zip files.

Everything in this course came out of that folder plus the meeting record for 3GPP TSG SA#112 in the 3GPP portal. There is no members-only step and no login.

11.3 The order to open them in

Not F1 to F5. This order:

• Start with the document list, F2. It is the whole meeting as one table. Sort it by agenda item to see the shape, sort it by status to see the outcomes, search it for a specification number to see what happened to a document you care about.

• Then the agenda, F1, but only the item list — you need it to make sense of the first column of F2, and chapter How to read a plenary agenda is that item list already read for you.

• Then the report's narrative, F3, and only for the items you care about. This is where the arguments are, and it is the only file that has them.

• Open a document, F4, last. By the time you open a zip file you should already know its type, its status and which argument it belongs to.

• Read F5 once, ever. The welcome pack does not change from meeting to meeting in any way that matters.

11.4 What a document number tells you

At this meeting every number looks like SP-260361 or SP-260697. Two things about that are safe to rely on, and one is not.

Safe: a revision gets a new number, and it is higher. The work item on the satellite speech codec arrived as SP-260417 and was approved as SP-260696. The paper on secure-by-design 6G was noted as SP-260584 and its revision, SP-260643, was withdrawn. In both pairs the later paper carries the larger number.

Safe: the number is a handle, not a meaning. It identifies one piece of paper at one meeting, and nothing about the number encodes the subject, the group or the outcome.

Not safe: reading anything else into it. How the numbers are requested and allocated is written in the agenda document itself, and that is where to look rather than inferring a rule from a handful of examples.

11.5 Three habits

• H1. Read the status column, not the title. Sixty documents ended revised ^[7], and each one has a twin somewhere with the same title and a different outcome.

• H2. Remember that a pack is many changes. SP-260379 alone carries twenty-three separate change requests under one approval, and 689 change requests arrived at this meeting ^[8] across 120 packs ^[9] and 14 standalone ones ^[10].

• H3. Check whether a pack was split. SP-260387 came out partially approved. The pack's status is then not the status of the change you are following, and only the document itself can tell you.

11.6 What to skip

• Empty agenda items. The report says "There were no contributions to this agenda item." forty times ^[11]. Skip every one of them.

• The attendance annexes, unless you specifically need them. Three tables of names — 568 registered ^[12], 365 checked in ^[13], 110 online ^[14] — and the counts are the only part most readers want.

• The narrative of items you do not follow. Agenda item 6 alone took 80 documents ^[15], so it is a long stretch of the report; if you are following change requests, none of it is yours.

• Documents with no narrative at all. Only 284 of the 346 are narrated ^[16], so for 62 the annex row is everything there is. Do not go looking for a discussion that was never written down.

11.7 Using a machine-written summary safely

There is a summary of this meeting written by a language model ^[17], and 224 of its documents have a machine-written summary of their own ^[18].

Use them as an index. They are good at telling you which of 346 documents is worth opening, and that is a real saving.

Then throw the summary away and quote the document list or the report. A summary is not evidence of what a meeting decided, and a course or a briefing that rests on one has no way of being checked.

11.8 When the trail stops

Sometimes the method runs out, and knowing that early saves an afternoon.

• The document is one of the 62 with no narrative. Only 284 are narrated; for the rest the annex row is all there is, and no amount of searching the report will produce a discussion that was never written.

• The report contradicts itself. It does this at least once, over whether the SA#111 report was approved ^[19], and the answer is to report both readings rather than pick one. Chapter The report is a draft, and it shows has the full case.

• The answer is in the chair's notes. Those are published after the meeting closes, the agenda says so, and they are not part of the material behind this course.

• The question is not one the record answers at all. Who supported what, how close a decision was, what any of it means commercially: none of that is in these files, and chapter What this record does not tell you lists the rest.

11.9 A worked example

Follow one thread end to end and the method becomes obvious.

Search the document list for "Ultra Low Bitrate". Two documents come back with the same title. One, SP-260417, is marked revised; the other, SP-260696, is marked approved. Habit H1 already tells you which one matters.

Now open the report at the agenda item those two sit under. It says "This new WID was approved." ^[20] — and then, in the same passage, the thing you would never have found in the document list: the chair is to inform the Project Co-ordination Group about the legal requirements around the dataset needed to train the model ^[21].

Two files, five minutes, and you can state the outcome and the unresolved question behind it without having opened either document.

Do the same for the modernisation study and it takes another five minutes. Eleven documents sit under agenda item 9 ^[22] and the document list gives you each one's status; the report gives you the guidance that closed it — continue the separated ASN.1 trial ^[23] and encourage companies to demonstrate Forge ^[24]. What you must not do is stop after the document list, because a column of "noted" with one "endorsed" at the end is not a summary anybody can use.

The reverse mistake is just as common: reading the report's narrative without the document list, and reporting an argument whose outcome you never checked.

11.10 Where this leads

The last chapter is the one that keeps all of this honest: what this record does not tell you, and what a careful reader must therefore refuse to say — What this record does not tell you.