14 Reading the record yourself, and what it cannot tell you

14.1 Why the method matters more than the facts

Everything in this course was read out of files and databases on one machine. None of it came from a person, a briefing or a summary, and all of it can be run again by anybody with the same stores.

That is the useful part. Facts about a moving subject go stale; a method does not. This chapter is the method, followed by the honest list of what the method cannot reach.

14.2 Habit one: search the acronym, never the title

This is the single most valuable thing on the page.

A work item is known by an acronym, and the acronym is what every document carries. The title is free text, written once, and it very often does not contain the word you would search for.

Four examples, all real IMS work, none of them findable by searching titles for the letters IMS:

Acronym	Its actual title
IMS	Provisioning of IP-based multimedia services
IMS-CODEC	Multimedia codecs and protocols for conversational PS services
IMS_SSFDD	IM-SSF Application Server Service Data Descriptions (Stage 2/3)
IMS_Corp2	Transfer of ETSI business trunking specifications

The first row is the founding work item of the whole subject, opened for Release 5, and its title never says IMS.

So the boundary of this course is a rule about acronyms and not about names. An acronym belongs to this subject when it carries the letters IMS with no capital letter in front of them, which takes in 345 work plan rows under 177 different acronyms ^{[1] [2]}.

The lower-case part is not a technicality. 3GPP writes "enhanced" as a small letter in front of a name, so eIMS5G_SBA, eIMS_RP, enIMS and FS_enIMS_ARC are IMS work.

A rule that also forbade a lower-case letter would leave 163 acronyms and throw those four out ^[3] — including eIMS5G_SBA, the acronym with more documents behind it than any other in this register, 543 of them ^[4].

14.3 Habit two: read a work plan row's fields for what they are

A row of the work plan looks like a project record and is not one. Four of its fields mislead if read as a project manager would read them.

• The completion percentage is typed in by hand. 339 of the 345 IMS rows read 100% ^[5], including rows whose acronym begins with the word Deleted. Nothing recalculates it and nothing dates it.

• The planned start and finish are the plan's own bookkeeping. 16 of the 345 rows carry 2009-05-29 as their planned start, across releases years apart, and IMSProtoc3 is one of them IMSProtoc3. Quote them as planned days, never as what happened.

• The release is what the plan says the row is for. It is not a date. One row is marked Rel-5 and the paper that opened it is RP-171635 IMS_CONV_IP_COM-UEConTest RP-171635; both are facts about the record, and neither dates the other.

• The name sometimes carries the word Deleted or STOPPED in front of it — 22 acronyms and 1 respectively ^{[6] [7]}. That prefix is typed into the title by hand. It is the only mark of its kind in the whole work plan and it is not a status field.

14.4 Habit three: date an old paper by its meeting

The document store took a large batch of old files in at once, and the stamp it wrote is the day it took them, not the day they were written.

113 of the 151 papers that the work plan names as the document that opened a work item carry 2017-10-05 as their upload stamp, whatever meeting they were written for.

SP-000216, the paper behind the founding work item, is one of them SP-000216, and so are the papers that opened the first two rounds of protocol alignment CP-060101 CP-080794.

So date an old paper by its meeting — SP-8, CP-31, CP-42 — or do not date it at all. And four meetings in this register have no days and no place on record at all: NP-20, SP-8, SP-9 and SP-15, the oldest ones.

There is a second date trap on the other side of the store. The earliest IMS document carrying a work item acronym went up on 2015-01-23 ^[8], and the newest on 2026-08-04 ^[9].

The first of those two is where the store's labelling begins, not where the work began: the work plan's own Release 5 rows carry planned days from 2000.

Every document count in this course is therefore a floor and never a total, and none of them says anything at all about the first ten years of the subject.

14.5 Habit four: open the method line before leaning on a number

Every count on this site carries the exact question that produced it and the day it was asked. That is not decoration. Two counts that look comparable often answer different questions, and the method line is where you find out.

Three real examples from this course:

• A1. The 241 specifications this work touches ^[10] is a count over a catalogue. It is not a count of specifications any group owns and not a count of what 3GPP has written about IMS altogether.

• A2. A per-year document count and an all-time document count cannot be subtracted from one another, because 237 documents carry no upload time at all ^[11] and appear only in the second.

• A3. A count filtered by work item acronym excludes every document that carries none. The store holds 1629723 meeting documents in total ^[12], and only a part of those carry any acronym.

14.6 Reading a document number

Every document number here is a prefix and a number, and the prefix says where it was filed. That one character before the dash is the difference between a proposal and a decision.

Prefix	Where it was filed	An example from this course
SP-	the SA plenary	SP-260686
CP-	the CT plenary	CP-260044
RP-	the RAN plenary	RP-171635
S2- S4-	SA working groups	S2-2413007
C1- C3- C4-	CT working groups	C4-261402

Meetings are written the same way in reverse — SP-112, CP-111, C4-134, C1-161 — a group and a number. A plenary drops the working-group digit, so CP-111 is the CT plenary and C1-161 is CT1.

14.7 What this record does not hold

What follows is the whole of it, in the register's own terms. Each entry says what to open instead, because a gap with no route out of it is a shrug rather than honesty.

• N1. No date for any release. The register holds release names, version numbers and meeting days, and nowhere the day a release froze. Instead: 3GPP's own release timetable, which is not part of this data.

• N2. No publication day and no approval day for any version. What is stored for 2176 of the 2178 release versions is the day the file went up on the 3GPP server, and 2 have no day at all TS 23.228. Instead: the specification's own page on 3gpp.org.

• N3. No vote, ballot or objection count of any kind. 3GPP works by consensus and no column of any store here records a tally, so nothing in this course was ever carried by a majority. Instead: nothing. There is no such number to find.

• N4. No reason for any decision. The stores hold the outcome word a document was given — 379 IMS documents carry approved ^[14] — and never why the room settled on it. Instead: the meeting report, which this school does not carry.

• N5. No status field for a work item. Nothing says whether work is running, stopped or finished. Instead: read the newest documents under the acronym and the current version of the specification, which is what What is running now: Releases 18 to 20 does.

• N6. No document before 2015-01-23 carries a work item acronym in this store ^[8]. Instead: for the first ten years, the work plan rows and the work item descriptions, which do reach back.

• N7. No expansion of any abbreviation. The work plan carries an acronym and a title and nothing that says what either stands for. Instead: the definitions clause of the specification that uses the name.

• N8. No title for 32 of the 151 papers the work plan names as the document that opened a work item, SP-000216 among them SP-000216. They may be cited and not named. Instead: open the file itself; the register carries its address.

• N9. No quotable work item description for 20 of the 177 acronyms ^[2]. Either the folder holds none, or it holds several and none of them proves it belongs to that row. IMS2, VoIMS and IMS-MMTel are among them IMS2 VoIMS IMS-MMTel, and nothing may be written about what those work items said they would do. Instead: the 3GPP work item pages, which the register links to.

• N10. No parsed text for 72 of the 241 specifications this work touches ^[10]. TR 23.794 and TR 22.823 are among them TR 23.794 TR 22.823: they may be named and counted, never quoted, and none of their clauses may be cited. Instead: the published document.

• N11. No registration count for 31 of the meetings in this register. The attendance store only reaches back a little way. Instead: nothing. Do not say how many people were in the room.

• N12. No technical content beyond the sentences quoted here. The register holds titles, scopes, definitions, objectives and counts — no procedure, no message, no parameter, no architecture. Instead: the specifications themselves, as listed in The specification family: 241 documents, and the two that are IMS.

• N13. No comparison with any other area of 3GPP work. Nothing here counts positioning, ambient IoT or sensing as a subject of its own. So IMS is never the largest, oldest or most argued-over subject in 3GPP on this evidence, however much it may feel like it. Instead: the other tracks of this school, each built the same way from its own register.

157 of the 177 acronyms have a description on this machine ^[15], and 36 work items are quoted here in their proposers' own words ^[16].

Those two numbers are the ceiling on how much of this subject can be told in anybody's own words rather than counted.

14.8 What a quoted sentence in this course actually is

Every quotation on this site is the opening sentences of a section, not the whole of it. A scope, a definition, an objective, a justification: the register takes the first sentences and stops.

That is why the Release 20 architecture study's scope reads, in full as the register holds it, "This study aims to:" and nothing after it ^[17].

The list that follows the colon in the real document is not on this machine, so this course does not say what that study aims to do in its own scope's words.

Each quote entry also carries the file it was copied from and how that file was proved to belong to its work item — usually that the file name begins with the document number the work plan points at.

Read that method line before leaning on a quotation. A work item folder holding several papers and no proof is exactly the case the register refuses to quote at all.

14.9 Ten minutes, from nothing

Somebody says "IMS" in a meeting and you have ten minutes before you need an opinion. This is the order that works, and none of it needs a specification open.

• M1. Search the document store for the acronym, not the word. If you do not know the acronym, start from the 345 work plan rows that carry IMS as a token ^[1].

• M2. Sort by newest and look at the last handful of documents. That tells you which group is working on it now, because the meeting code is in the number.

• M3. Open the paper the work plan points at and read the justification. It is one page and it is the shortest honest statement of why the work exists — 36 of them are quoted in this course ^[16].

• M4. Look up the version of the specification it targets. Below 1.0.0 is a draft and nothing in it is settled.

• M5. Count the change requests against TS 24.229 under that acronym TS 24.229. Edits landing on the protocol mean the work has reached the wire; none means it has not.

Five steps, no specification text read, and at the end you know how finished a piece of work is — which is usually the question the room is actually asking.

14.10 What the record is very good at

Having said all of that, four questions this data answers well, and it answers them better than any summary would.

• G1. Which document to open for which question — the whole of The specification family: 241 documents, and the two that are IMS.

• G2. How much work landed where, with every number's origin restated. The scale is 241 specifications ^[10] and 4939 meeting documents ^[18].

• G3. What each work item said it was for, wherever a description proves it belongs to that row — 36 of them, word for word ^[16].

• G4. What is being written at this end of the record, and by which group — the whole of What is running now: Releases 18 to 20.

That is a map of a subject. It is not the subject, and this chapter exists so that nobody mistakes one for the other.

14.11 Reading order

For somebody starting from nothing and wanting to be useful in a conversation about IMS within a day:

• B1. TS 22.228 TS 22.228 — what the system is supposed to do at all, from the user's and the operator's side. Short, and it defines the words.

• B2. TS 22.173 TS 22.173 — telephony itself, and the supplementary services people actually notice.

• B3. TS 23.228 TS 23.228 — how the system is put together. This is the document everything else refers back to.

• B4. TS 24.229 TS 24.229 — what goes on the wire. It is the largest single target in this record, named by 1680 IMS documents ^[13].

• B5. TR 23.794 TR 23.794 — the study on putting IMS on the 5G core, then the work item that implemented it eIMS5G_SBA. This is the pair that explains the modern shape of the subject.

14.12 Where to go next

If this course was too much detail, the same subject is written twice more: the quick-start course is an hour, and the overview course is a morning. Neither is this course with paragraphs deleted; both were written from the same register at a different altitude.

The sources page of this course lists every register entry cited anywhere in it, each with what it is, where it was read from and the day it was read. That page, not this one, is the thing to check a sentence against.