1 Why anybody writes a standard down

1.1 What happens when nobody writes it down

Two machines built by people who never met have to work together the first time they are switched on. A handset and a base station. A browser and a web server. Neither box can ask the other what it was expecting.

There are only three ways out of that. One company owns both ends and sells them as a matched pair. Or everybody guesses, and most pairings fail. Or somebody writes down, in advance, exactly where one side stops and the other starts, and both sides build to the writing.

The third way is a standard.

1.2 What a standard is, in the words of the people who write them

The IETF is one of the two organisations this course follows, and it states its purpose in a single sentence:

Its founding statement is blunter about what a standard actually is: a description carrying a unique name, which says that if you want to do this thing, this is how to do it. It obliges nobody to use it. The benefit is only that products built from it work together RFC 3935.

3GPP — the 3rd Generation Partnership Project, which writes the mobile specifications — gives a more hurried reason. Its documents are implemented by industry almost in parallel with the writing of them, and that demands strict and fast procedures for handling changes §Introduction.

There is a second reason in the same paragraph: changes have to be documented and controlled, so the text stays consistent and any change can be traced backwards years later §Introduction.

1.3 A standard is not a law

Neither organisation can make anybody do anything. A 3GPP document imposes no obligation of its own; an obligation only ever arrives through legislation or a contract somebody signed ^[2].

That is precisely why the wording is picked so carefully. In 3GPP text, "shall" marks a requirement and "should" marks a recommendation ^[3]. The word "must" is not used at all, so nobody can read a specification as legislation.

The IETF does the same job with capital letters. MUST is an absolute requirement, SHOULD allows a reasoned exception, MAY is truly optional RFC 2119.

1.4 How much of this there is

The pile is larger than most people expect. Counted on this machine, the catalogue holds 3,928 3GPP specifications ^[4] and the archive 9,735 RFCs ^[5].

The 3GPP pile splits two ways. 2,149 of those documents are Technical Specifications ^[6] and 1,778 are Technical Reports ^[7]. A Technical Specification is the standard itself. A Technical Report is usually the study that came first — a feasibility study that says whether the thing is worth specifying at all §6.0.2.

That last point is the one worth carrying into the next chapter. Nobody writes a standard alone, and the interesting part is who is in the room: Who writes them — two organisations, two habits.