2 How this subject is cut out of the work plan

2.1 Why the boundary comes before the history

3GPP does not have a folder called IMS. It has a work plan: one long table in which every piece of work anybody has ever opened gets a row, and every row carries an acronym, a title, a release, a group and a set of planned dates.

So "IMS work" is not something the data hands over. It is a line somebody has to draw through that table, and once drawn, every count in every later chapter inherits its edges.

This chapter draws the line and then says plainly what it does and does not take in. A reader who skips it will meet a number four chapters later and have no way of knowing what it counted.

2.2 The rule, in one line

A row of the 3GPP work plan belongs to this subject when its acronym matches the pattern (?:^|[^A-Z])IMS — the three letters IMS, either at the start of the acronym or with something that is not a capital letter in front of them.

That is checkable in a line of code, which is the point of it. It takes in 345 rows of the work plan ^[1], under 177 different acronyms ^[2]. The register entry behind the first of those two records that the work plan holds 9495 rows altogether.

Those 345 rows are spread over 16 releases, Rel-5 to Rel-20 ^[3] — which is why the middle of this course is organised by release and not by anything else.

2.3 Why one acronym is not one work item

The two numbers above are different for a reason that trips people up constantly: a row is one group's share of one work item, not the work item itself.

eIMS5G_SBA, the Rel-16 item that put IMS onto the service-based architecture of the 5G core, has six rows and six unique ids in the plan, across CT1, CT3, CT4 and SA2 eIMS5G_SBA. Somebody saying "one work item" and somebody counting six rows are both right.

At the other end, FS_enIMS_ARC, the Rel-20 architecture study, is one row belonging to SA2 alone FS_enIMS_ARC.

2.4 Why a small letter is allowed in front

The rule forbids a capital letter before the three letters and allows anything else, including a lower-case one. That is deliberate, and it is the one place where the boundary could reasonably have been drawn somewhere else.

3GPP writes "enhanced" as a small e or en in front of a name. So eIMS5G_SBA, eIMS_RP, enIMS and FS_enIMS_ARC are all IMS work with a lower-case letter stuck to the front of it.

Draw the boundary one notch tighter — forbid a letter of any case in front — and 163 acronyms survive instead of 177, with those four among the casualties ^[4]. Two of them are the largest and the newest pieces of work in this whole course.

That count is in the register on purpose. It is there so that anybody who thinks the rule is too loose can see exactly what the tighter rule would have cost, rather than being asked to take the choice on trust.

2.5 Why the title of a row is never tested

The obvious alternative rule is to search the titles for the word IMS. It is wrong, and the register carries four proofs of it.

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

Not one of those four titles contains the letters IMS. The first of them is the founding row of the entire subject, the one that opened IMS in Rel-5 — a title search would have thrown out the beginning of the story it was meant to find.

The reverse mistake is just as easy to make in the other direction, which is why the rule tests the acronym and nothing else. An acronym is an identifier that somebody registered; a title is prose somebody typed.

That difference runs deeper than it looks. An acronym is what every other store on this machine keys against: the meeting-document database has a work item column, and it holds acronyms.

So a boundary drawn on acronyms can be carried from the work plan to the documents to the specifications without anybody matching text twice. A boundary drawn on titles could not be carried anywhere.

2.6 Where a status word hides, and why it is not a field

One more thing lives in the acronym and the title rather than in a field of its own, and it catches people out constantly.

22 IMS acronyms carry the word Deleted written at the front of their own name, and one carries STOPPED ^{[5] [6]}. That prefix was typed in by hand. It is not a status column, nothing validates it, and there is no matching word for "running" or "finished".

The plan does carry a completion percentage, and 339 of the 345 rows stand at 100% ^[7]. That, too, is kept by hand and says nothing about whether anybody is still working.

2.7 What a count over this register is, and is not

Three sentences cover everything a reader needs to hold on to when a number appears in a later chapter.

A count is a count of what the acronym rule reaches. It is never a count of everything 3GPP did about IMS. Work that was done inside a general maintenance item, or under an acronym that never carried the letters, is outside the boundary by construction and cannot be counted from here.

A document count is a floor, never a total. 4939 meeting documents carry one of the 177 acronyms ^[8], out of 1629723 meeting documents the store holds altogether ^[9].

The second entry's method line records that 852617 of those carry any work item acronym at all, so a document whose acronym column was left empty is invisible to every count here, however much it was about IMS.

That column is also messier than it looks. The first entry's method line records that 159 different spellings of it match the 177 acronyms, because the column sometimes carries two work item names at once.

A document can therefore belong to two pieces of work, and adding per-work-item counts together will count it twice.

And the record starts far later than the work does. The earliest IMS document on record went up on 2015-01-23, at meeting C4-68 ^[10]. IMS starts at Rel-5, and the work plan's own Rel-5 rows carry planned days from the year 2000.

So the document counts say nothing whatever about the first stretch of this subject — see Release 5: the founding, and the ring of work around it, where a founding work item has one document to its name and that number means nothing.

2.8 Where the register runs out of words

157 of the 177 acronyms have a work item description on this machine — the few pages a plenary approved that say what the work was for ^[11]. Of those, 36 are quoted in this course in their proposers' own words ^[12].

The register's own list of what it does not hold names the other 20: acronyms for which either no converted description exists, or several do and none of them proves it belongs to that row.

FS_IMSconf, IMS-CCR-Mn, IMS-CCR_Test, IMS-FrWk and IMS-MMTel are among them, and so is IMS2, the row that opened IMS phase 2.

For those 20, nothing may be written about what the work said it would do. Not guessed from the title, not inferred from the acronym, not filled in from what the reader already knows about IMS. Releases 6 and 7: interworking, messaging, and a stream that never stopped runs straight into this with IMS2 and says so on the page rather than quietly writing around it.

2.9 How to check any of this yourself

Every count entry in the register carries a method line: the exact question that produced the number, the file it was asked of, and the day it was asked. That is not decoration. It is the only thing that makes a number on this site worth more than a number in a slide deck.

The entry behind the 345 says it counted rows of the work plan table whose acronym matches the pattern, on 2026-08-05 ^[1]. The entry behind the 177 says it counted distinct values of the acronym field over those same rows ^[2].

Two different questions, two different numbers, both restatable by anybody with the same file.

So the test for any sentence in this course is short: open the entry it cites, read the method line, and ask whether the sentence is true of exactly that question. Where it is true of something wider — all of 3GPP, all of IMS, all time — the sentence is wrong, however good the citation looks.

2.10 Where to read on

The work plan itself is public, and the register entry behind every count in this chapter names the file and the question that produced it. Nothing here was counted by hand, and nothing was chosen by eye.

With the boundary drawn, the rest of the course can walk the releases. It starts where the plan does, at Rel-5 and a single row called IMS, in Release 5: the founding, and the ring of work around it.