8 The map: fourteen groups, 309 meetings and 374 names

8.1 Why an area of work needs a map and a document does not

A course about one specification has one owner, and the question of who does the work is answered in a line. An area has no owner.

IMS has been argued in the requirements group, the architecture group, three protocol groups, the smart card group, the media group, the security group, the management group, two radio groups and three plenaries. Each was answering a different question.

So the first thing an area needs is a map: who appears at all, how much each of them filed, and over how many meetings. This chapter lays the whole map out at once and reads its shape. The two chapters after it take the groups one at a time.

Fourteen groups have IMS documents on the record, and 309 separate meetings have had one on the table ^[2]. The whole record is 4939 documents ^[3].

8.2 Everybody who filed

Group	Documents	Meetings	First and last
CT1	2169	66	C1-90, C1-162
CT4	746	48	C4-68, C4-135
CT3	719	47	C3-81, C3-144
SA2	429	30	S2-107, S2-175
CT	221	44	CP-67, CP-112
SA4	186	27	S4-84, S4-136
RAN5	147	14	R5-74, R5-109
SA3	96	10	S3-100-e, S3-128
SA1	88	7	S1-80, S1-105
SA5	72	7	S5-100, S5-123
SA	34	17	SP-68, SP-112
RAN	25	13	RP-75, RP-110
CT6	6	2	C6-78
RAN2	1	1	R2-92

The first letter tells you which family a group belongs to, and it holds across the whole record: C for the CT groups, S for the SA groups, R for the RAN groups.

The second character says which kind of body it is. A digit is a working group, a P is the plenary above them: C1, S2 and R5 against CP, SP and RP.

8.3 What the map is built from

Two counting rules made that table, and knowing them changes what you can say about a row.

A document is IMS work when its work item column names one of the 177 acronyms the boundary rule takes in ^[4]. On the day it was read that gave 4939 documents, and 159 different spellings of that column matched — because a document sometimes carries two work item names at once ^[3].

A document belongs to the group whose meeting it was filed at. A row of the table above is documents whose meeting code begins C1-, or S2-, or R5- CT1. That is the meeting, not the author: a paper written by one company and filed at a CT1 meeting counts as CT1's row.

So the table answers "where was this argued" and not "who wrote it". The second question is answered further down this chapter, by a different count over a different column, and the two must never be mixed in one sentence.

8.3.1 How the work is handed out, in a work item's own words

The map exists because a work item hands its parts to named groups, and the descriptions say so on the page. A Rel-9 item on the border control function is as plain as it gets:

"For CT3:" is the map being drawn. That work item covers three rows across CT3 and CT4 and has no document in this record against it IMS_IBCF, which is a reminder that the handing out is recorded even where the paper is not.

One more number belongs here because it explains a gap you will meet: 237 IMS documents carry no upload time at all ^[5]. That is why a count over one year and a count over the whole record answer different questions, and why the per-year figures never add up to the total.

8.4 What the shape of that table says

The three largest counts all belong to CT working groups, and that single fact tells you what kind of subject IMS is.

CT groups write stage 3: the bytes on the wire, the message fields, the interfaces between core network functions. A subject whose weight sits there is a core-network subject. A radio subject would put RAN1 and RAN2 at the top of the same table, and here RAN2 has one document RAN2.

CT1 alone carries 2169 documents over 66 meetings CT1 — more than CT4, CT3 and SA2 have each. It owns the SIP profile, and a protocol specification that everybody implements attracts change requests the way nothing else does.

8.4.1 Two columns, two different meanings

Documents counts paper. It rises when many companies file into the same document at the same meeting, which is what a subject at full stretch looks like.

Meetings counts persistence. It rises when a subject stays on a group's agenda for years, whether or not much was filed at each one.

A group can be high on one and low on the other. The CT plenary sits at 221 documents over 44 meetings CT — a few papers at a great many meetings, which is what approving and reporting on work looks like.

SA1 sits at 88 documents over 7 meetings SA1, its first stored document in 2017 and its last in 2024: a short, concentrated involvement in a long subject.

8.4.2 What the plenary rows are for

Working groups argue and write; plenaries decide. Every work item and every study in this subject was opened by a document filed at one of the three plenaries — CT with 221 documents CT, SA with 34 SA and RAN with 25 RAN.

Those counts are small, and their being small is right. A plenary paper opens work, revises it or reports on it; the text itself is written downstairs.

8.4.3 The two nearly empty rows

CT6, the smart card group, has six documents CT6. RAN2 has one RAN2.

RAN2's single document is tagged with the work item IMS-CODEC at meeting R2-92 RAN2. The register does not record what kind of document it is, so it may not be called a change request, a liaison or work of RAN2's own.

It is one row, and one row is all the record supports.

8.5 Who writes the paper

374 different names appear as the writer of an IMS document ^[6]. The largest by a clear margin is Ericsson, with 602 ^[7].

Name as the documents spell it	Documents
Ericsson [7]	602
Nokia, Nokia Shanghai Bell [8]	258
Huawei [9]	173
China Mobile [10]	167
Orange / Marianne [11]	160
Huawei, HiSilicon [12]	158
Ericsson / Nevenka [13]	130
BlackBerry UK Ltd. [14]	123

Read the second column of that table before the first. Four of the eight rows are two companies wearing four names.

8.5.1 The free-text trap

The column that says who wrote a document is free text. Whatever the document spelled is what got stored, and nothing in this register resolves a spelling to a company.

So Ericsson has a second row of 130 documents under "Ericsson / Nevenka", and Huawei has a second row of 158 under "Huawei, HiSilicon" ^{[13] [12]}. The second name in each is a person or a business unit that ended up in the field.

Never add two of those rows together. A sum of two free-text spellings is a number nobody counted, and this register holds no entry for it — which is exactly why it may not be written here.

8.5.2 A name in that column is not always a company

Open a few individual documents and the column turns out to hold three different kinds of thing.

• A company. C4-244485 was filed by China Telecom Corporation Ltd. C4-244485, C1-246921 by Nokia C1-246921 and C3-250657 by Ericsson C3-250657.

• A working group. S2-2413007 was filed by SA WG2 itself S2-2413007. A group answering another group's letter files under its own name, so a group is one of the 374 names too.

• A company with something extra stuck to it, which is where the 130 documents under "Ericsson / Nevenka" ^[13] and the 158 under "Huawei, HiSilicon" ^[12] come from.

The counting is honest about this and does no tidying at all: it groups by the string each document spelled, and the register says so in the method line of every one of those counts.

8.6 What the map does not tell you

Three things a reader reasonably expects from a table like the one above, and none of them is in it.

• Who cared. The map shows who filed. A group can have every meeting spent on a subject and file nothing, and a group can file once and never think about it again.

• Whether anything was agreed. A document's outcome word is about that document alone. No count here says whether a group got its way.

• Anything about the first ten years. IMS work opened in Rel-5 and the earliest labelled document in this store went up on 2015-01-23 ^[15]. Every row of the table above is a floor, and for the founding releases it is a floor of nearly nothing.

• How many people were in the room. The attendance store reaches back only a little way. 31 of the meetings this register describes carry no registration count at all: C1-162 in Prague has one, at 149 registered C1-162, and C1-112 has none C1-112.

8.7 How to use the map

Three questions this table answers well, and the way to ask each one.

• Where was this argued? Find the groups with a real count and read their work item lists. For IMS that is CT1, CT4, CT3 and SA2 before anybody else CT1 SA2.

• For how long? Read the meetings column and the first and last meeting beside it. CT1's 66 meetings from C1-90 to C1-162 is a subject that has never left the agenda CT1.

• Who should I be reading? Take the group, then take the document it owns — which is what the two chapters after this one are for.

And one question to stop asking of it: which group mattered most. Nothing in this record measures that, and a document count is the wrong instrument for it in every direction.

8.8 Where this meets the rest of the course

The four CT groups carry most of the record and are taken one at a time in CT1, CT3, CT4 and CT6: where most of it happens. The SA groups, the two RAN groups and the plenaries are SA1 to SA5, the two RAN groups, and the plenaries above them.

For what these groups were actually working on when the counts were highest, the two release chapters are Releases 10 to 14: the browser, telepresence and the middle years and Releases 15 to 17: IMS moves onto the 5G core.