Does success breed success?

Introduction

Most readers will answer yes to the question asked in the title. Indeed, very often we see that success of a human organisation breeds success. Until, I mean, the machine that looked like it could produce results forever “seizes”. But don’t look elsewhere for the causes of failure: it’s not the machine, the causes are humans inside and/or outside.

In an age when things move fast and change, MPEG has been in operation for three decades. Its standards have achieved and continue achieving enormous success serving billions of human beings: consumers, service providers and manufacturers.

This article makes some considerations on the best way for MPEG success to breed success – unless success to breed failure is the goal. Apparently unrelated considerations are made in The Imperial Diet is facing a problem.

Recalling the MPEG story

MPEG started in 1988 as an “experts group” with the task to develop video coding standards for storage media at a rate of about 1.5 Mbit/s, like the compact disc (CD). This was because, in the second half of the 1980’s, the Consumer Electronics and telco industries imagined that interactive video – local or via the network – was a killing application.

Within 6 months MPEG had already started working on audio coding because – it looks obvious now, but it was not so obvious at that time – if you have video you also need audio and, if you do not compress stereo audio at 1.41 Mbit/s, the output bitrate of a CD, there will be no space left for video. In another 6 months MPEG had started working on “systems” aspects, those allowing a receiver to reproduce synchronised audio and video information.

These were the first steps in the MPEG drive to make standards that had no “holes” for implementors. Thanks to these efforts, the scope of use of MPEG standards, still within the scope of “coding of moving pictures and audio”, have expanded like wildfire: starting from coding of moving pictures at 1.5 Mbit/s and expanded to more video, audio, transport, protocols, API and more. With its standards, MPEG is handling all technologies that facilitate enhanced use of digital media.

The MPEG expansion is a joyous phenomenon that has created an expanding global brotherhood of digital media researchers – in industry and academia – for which MPEG and its standards are the motivation for more research. If research results are good, they can make their way into some MPEG standard.

MPEG needs a structure

Clearly you cannot have hundreds of people discussing such a broad scope of technologies at the same time and place. You can split the work because technologies can be considered independent up to a point. Eventually, however, like in a puzzle, all pieces have to find a place in the global picture. The MPEG structure has been implemented to allow the creation of ever more complex puzzles.

In its 31 years of activity MPEG has developed a unique organisation capable of channeling the efforts of thousands of researchers working at any one time on MPEG standards – only a fraction of which actually show up at MPEG meetings – into the suites of integrated standards that industry uses to churn out products and services worth trillions of USD a year.

The figure below depicts the MPEG structure from the viewpoint of the standard development workflow.

The MPEG workflow

Typically, new ideas come from members’ contributions, but can also be generated from inside MPEG. The Requirements group assesses and develops ideas and may go as far as to request “evidence” of existence and performance of technologies (Calls for Evidence – CfE) or actual “proposals” for fully documented technologies (Call for Proposals – CfP).

MPEG has never had a “constituency” because it develops horizontal standards cutting across industries. It has established liaisons with tens of industries and communities through their standards committees or trade associations. We call many of them as “client industries” in the sense that they provide their requirements to MPEG against which MPEG produces standards. At every meeting, several tens of input liaisons are received and about the same amount of output liaisons are issued.

Many CfPs cover a broad range of technologies that are within the competence of the different MPEG groups. The adequacy of submitted technologies is tested by the Test Group. The submitted proposals and the test results are provided to the appropriate technical groups – Systems, Video, Audio and 3D Graphics.

The Chairs group includes the chairs of all groups. It has the task to assess the progress of work, uncover bottlenecks, identify needs to discuss shared interests between groups and organise joint meetings to resolve issues.

An MPEG week is made of intense days (sometimes continuing until midnight). Coordinated work, however, does not stop when the meeting ends. At that time MPEG establishes tens of ad hoc groups with precise goals for collaborative development to be reported at the next meeting.

The Communication group has the task to keep the world informed of the progress of the work and to produce white papers, investigations and technical notes.

MPEG is not an empire

From the above, one may think that MPEG is an empire, but it is not. MPEG is a working group, the lowest layer of the ISO hierarchy, in charge of developing digital media standards. It formally reports to a Subcommittee called SC 29 but, as I have explained in Dot the i’s and cross the t’s, SC 29 has ended up with a laissez-faire attitude that has allowed MPEG to autonomously develop strategy, organisational structure and network of client industries. MPEG standards have given client industries the tools to make their analogue infrastructures digital and, subsequently, to leverage successive generations of standard digital media technologies to expand their business. With some success, one could say.

The MPEG organisation is robust. Virtually the same organisation has been in place since – 25 years ago – MPEG had an attendance of 300. Groups have come and gone and the structure currently in operation has been refined multiple times in response to actual needs. Changes have been effected, and there will be more changes in the future. However, they all have been and, as far as I can see, will be incremental adaptations, to perfect one aspect or another of the structure. With this structure, more than 150 standards have been produced, some of which have been wildly successful.

MPEG can count on three assets: the logic of the structure, the experience gained in all those years, its membership and its client industries. With these, MPEG success can breed more success in the years to come.

The Imperial Diet is facing a problem

I said before that MPEG is not an empire. In the imperial context of the Holy Roman Empire, MPEG could be defined as a Margraviate in charge of defending and extending a portion of the frontiers of the Empire. A Margraviate reported to a Kingdom who reported to the Imperial Diet.

Now, let’s suppose that the Imperial Diet has requested the S Kingdom to review the status of its two J and M Margraviates and propose a new arrangement. The main element in the decision is the size of the two Margraviates: 10% of the territory of the S Kingdom for the J Margraviate and 90% for the M Margraviate. Ruling out other fancy ideas, the S Kingdom has two options: request that the M Margraviate be elevated to Kingdom status or create a few smaller Margraviates inside the S Kingdom out of the M Margraviate.

There is a problem, though, if the M Margraviate is cut in smaller Margraviates: the Margraviates of the Holy Roman Empire are not domino game pawns. For decades the M Margraviate has fought hard extending its territory – hence the Holy Roman Empire’s territory – to lands that until then were occupied by unruly tribes. It has been successful in its endeavours because it had large armies with different skills: archers, knights, foot soldiers and more. By skillfully coordinating these specialised troops, the M Margraviate was able to conquer new lands and make them faithful fiefdoms.

But there is another important consideration: there are wild hordes coming from the steppes of Central Asia with a completely new warfare technique. Some armies of M Margraviate are having a hard time dealing with them, even though they are learning a trick or two to fight back.

How could the new armies of the different Margraviates created out of the M Margraviate defend – never mind extend – the frontier, when the S Kingdom does not know the territory, having lived all time in its castle, and has never led an army?

The Holy Roman Empire lasted 1,000 years. There is no doubt that the Imperial Diet would make the M Margraviate a Kingdom keeping its armies and structure unchanged. Warfare is a serious business and the effective defence of the frontiers is the priority.

Conclusions

Fortunately, today there is no Margraviates and Kingdoms anymore, much less the Holy Roman Empire. There are also no new territories to conquer by force of arms and there are no frontiers to defend against rebellious hordes.

I realise now that at the beginning of this article I have promised that I would make some considerations on the best way for MPEG success to breed success and not failure. Maybe I will do that next time.

Posts in this thread

 

 

Dot the i’s and cross the t’s

Introduction

In Book 2 of the Georgics, the Latin poet Virgil writes “Felix, qui potuit rerum cognoscere causas”. This maxim was true in 29 BC when the Georgics was written and remains true 2030 years later. For those who did not have the chance to study Latin, the verse means “Fortunate who could know the causes of things”.

Virgil’s maxim will be used in this article to draw some conclusions on current matters. Before getting to the things, however, I need to talk about the the causes. Those who cannot wait can jump to the conclusions, at their own risk.

The need for standards and standards bodies

At the climax of Belle Époque, Europe realised that a properly functioning industry needed standards. Britain – at that time for sure and, from now on, most likely not really part of Europe – was the first to establish an Engineering Standards Committee (1901).

In 1906 the most culturally advanced industrial field of the time – electrical technologies – was the first to recognise the need, not just for standards, but international ones, and established the International Electrotechnical Commission (IEC). Electrical technologies were second only to telecommunications (actually third, if we consider the Universal Postal Union) in recognising that need. Indeed, 41 years before governments had established the International Telegraph Union. But governments is one thing and private industry quite another.

The rest of the world took a quarter of a century and a war to realise the need for international standards. Finally in 1926 the International Federation of the National Standardising Associations (ISA) started, only to stop 16 years later when governments has other priorities (killing millions of people in WW II). In 1946 the idea was revived and the International Organisation for Standardisation (ISO) was created as a not-for-profit association. National Standards Associations (or National Bodies) – not governments – are represented in ISO.

How can you govern an international organisation that issues standards that are, yes, voluntary but, if you do not conform, you’re in a whole world of hurt? The answer is: hierarchy and scope. In ISO there are 4 layers (actually more, if you want to know how many read Who owns MPEG?): Technical Management Board (TMB), Technical Committees (TC), Subcommittee (SC) and Working Group (WG). Each entity is administered by a secretariat run by a National Body and has a scope that defines what the entity is expected and entitled to do.

Some scopes

Delimitation of territory is one of the most engaging human activities. According to Standards for computers and information processing, by T. B. Steel, Jr, page 103 et seqq. (in Advances in computers, Franz L. Alt and Morris Rubinoff (Editors) Volume 8) in 1967 the scope of TC 97 Data processing was: the standardisation of the terminology, problem definition, programming languages, communication characteristics, and physical (i.e. non electrical) characteristics of computers, and information processing devices, equipment and systems.

According to the same source, TC 97/SC 2 Characters set and coding at that time was about Standardisation of character sets, character meanings, the grouping of character sets into information, coded representation, and the identification of it for the interchange of information between data processing systems and associated equipments…

Typically working groups develop standards. They do so with a major constraint: decisions may only be made by “consensus”, defined as

general agreement where there is no sustained opposition to substantial issues by any important part of the concerned interests, in a process that seeks to take into account the views of all parties concerned.

The definition is supplemented by the note: consensus does not imply unanimity.

Obviously this text can only hint at the complexity of other environments where decisions are made not by consensus but by voting. One can image that these other environments are such that, in comparison, a horse-trading market is a place that boarding school pupils can safely visit.

JPEG and MPEG

In the 1980’s Videotex was a service telcos wanted to offer as a competing service to broadcasters’ Teletext service. One limitation of videotex and teletext, however, was that information could only be displayed with characters and rudimentary graphics (made as combinations of ad hoc characters). Telcos thought that videotex services could be enhanced by pictures transmitted at 64 kbit/s made available by Integrated Services Digital Network (ISDN).

In 1986 a joint group between TC 97 of ISO and SG XVIII of CCITT (ITU-T’s name of the time) was created to develop a compressed image format. As videotext was based on characters, TC 97/SC 2 was the natural place to develop that standard. SC 2 created WG 8 Coded representation of Picture and audio information. WG 8 hosted the Joint Photographic Coding Experts Group (JPEG).

Two years later, WG 8 created the Moving Picture Experts Group (MPEG), not joint with CCITT. In any case, if it had been joint, it would have been joint with SG XV, not with SG XVIII. There was nothing equivalent to the Treaty of Tordesillas, but in CCITT the digital world was divided between SG XVIII for audio and telematic services, and SG XV for video. MPEG – Coding for Moving Pictures and Audio – was a Copernican revolution.

Immediately, MPEG had a skyrocketing attendance: 100 members 18 months after its establishment and 200 members after two more years. That was because MPEG was working on such a high profile standard as digital television (actually only the baseband part of it, but that did not really matter).

Unlike other committees dealing with the “digital television”, who were populated by “advocates” accustomed to use “analogue” arguments to support their proposals, MPEG was populated by technical experts who made their cases with “digital” arguments in the framework of inflexibly digital Core Experiments rules. Some “advocates” did show up in the early MPEG-2 days, but they soon left never to come again.

The parent committee

In the years 1989-90-91, I had supported WG 8 Convenor’s bid to promote WG 8 to SC status (see here for more details) and in April 1990 the SC 2 plenary approved the following resolution:

JTC1/SC2 considering that

  1. The standardisation of the coded representation of picture, audio and multimedia information is considered to be one of the most important areas for standardization in the 1990’s;
  2. The work and scope of SC2/WG8 has expanded substantially beyond the scope of SC2;
  3. The work of SC2/WG8 has developed into a critical mass largely significant to warrant SC status;

Recommeds to JTC1

  1. To establish a new JTC1 Subcommittee for the purpose of developing standards in the area of Coded Representation of Picture, Audio and Multimedia Information;

It took another 18 months for SC 29, the entity WG 8 had morphed into, to hold its inaugural meeting.

A role for SC 29

In ISO a Subcommittee is part of the formal hierarchy. What was SC 29’s role ?

  1. Playground for “advocates”. Having found a hard time in MPEG, one could think that “advocates” should move to SC 29 to find a more consonant “breeding ground”. Indeed, SC decisions are made by voting, but only after a lot of “analogue haggling” in the hallways. This did not happen because, once MPEG had settled the algorithm, the standard was done, save the need to to cross some t’s and to dot some i’s. There could have been room for some “analogue discussions” on some business-related issues as profiles and levels. SC 29, however, was not the right place to hold such discussions because only MPEG experts could handle the technical aspects.
  2. Playground of large company representatives. At that time some ISO committees were populated by some national body representatives who worked for some large companies. They were interested in committees NOT to develop some standards and sent their representatives to act accordingly. But as fate would have it, in the years immediately following the establishment of SC 29 there was a serious economic crisis that forced some large companies to lay off those professional participants to cut “unnecessary” expenses.
  3. Strategic planner. In 1993, the time of John Malone’s “500 channels”, the Italian National Body proposed to investigate standardisation opportunities for content metadata (see here for more details). SC 29 established an ad hoc group to study the needs for users who wanted to find content in those 500 channels. One year later, however, the convenor of the ad hoc group reported that there had been no activity. MPEG then developed the suite of content metadata standards called MPEG-7.

No one should be surprised that, for the next 25 years, SC 29 held yearly meetings to discuss such strategic issues as progression of work items, consolidations and minor revisions, and liaisons. Of course with no “advocates” in sight.

MPEG as a virtual subcommittee

The space left empty by SC 29 was occupied by MPEG. Continuing its initial drive, MPEG developed a modus operandi that has allowed it to produce the integrated digital media standards that have changed and keep on changing the media industry.

The four figures below depict the main elements of MPEG’s modus operandi.

  1. Top-left depicts the adaptation of ISO’s standard development process to acquire technology elements suitable to the development of a standard and to verify that the standard developed matches the original requirements. More on this at How does MPEG actually work?
  2. Top-right depicts the industries contributing technologies (right-hand side), the means to acquire them, the assets accumulated in MPEG standards and the client/implementation industries (bottom of figure). More on this at The MPEG ecosystem.
  3. Bottom-left depicts the unfolding of the MPEG workflow: ad hoc groups; “MPEG week” with its components: plenaries, subgroups, break-out groups, joint meeting and chairs meetings; and creation of new ad hoc groups. More on this at Looking inside an MPEG meeting.
  4. Bottom-right depicts the integrated nature of MPEG standards. The parts are developed by different groups who come to agree on the glue that is needed to keep the parts independent and interworking. Moreon this  in Hamlet in Gothenburg: one or two ad hoc groups?

Conclusions

MPEG has been fortunate to have been able to operate in a paradise island for 31 years.

It has devised strategies and defined work plans. It has sought and established liaison with outside industries. It has added industries as members of the MPEG digital media community. It has called for technologies and integrated them into standards. It has been the ear industries could talk to to have their needs satisfied

All this while MPEG meetings have grown to 600 participants and “advocates” have been kept at bay.

The MPEG digital media community has thrived on a continuous flow of evolving standards with an impact measured in trillions of USD and billions of people.

In Paradise Lost John Milton writes: Better to reign in Hell, than to serve in Heaven.

In Paradise MPEG a reborn John Milton could write: Better to serve in Heaven than to reign in Hell.

SC 29 was kind enough to handle administration. Mindless industry elements should memorise Virgil’s maxim before they engage in their adventures.

Posts in this thread