Anda di halaman 1dari 20

cQ&

.--
-- CRMPultR
slnNMRos
B
ELSEVIER ComputerStandards& Interfaces 18 (1997) 431-450
IiINIERHlCB

Examining the relevance of paradigms to base OS1


standardisation
Tineke M. Egyedi *
Faculty of Technology and Society, De@ Universiry of Technology, De Vries uan Heystplantsoen 2, 2628 RZ Delji, The Nether/an&

Abstract

It is now customary to think about standardisation in terms of economic and political interests. That analysing ,commiltee
processes from a technological vantage point might be fruitful as well, is often neglected or even denied. This article
explores the extent to which technological paradigms explain the outcome of committee processes. It is argued that,
especially in fields of converging technologies :such as telematic services, paradigms explain patterns of interest negotiation
in standards development. The argument is substantiated with the findings of case studies on three base Open Systems
Interconnection (OSI) standards: the transport, session and message handling protocols. 0 1997 Elsevier Science B.V.

Keyw~~rds: Technological paradigms; Telematic services; Committee processes; X.25; OS1 transport layer; OS1 session layer; X.400;
Teletex; Standardstrajectories; Converging technologies; Operational trajectories; Paradigm of telephony

1. Introduction incentive for industries to make proprietary informa-


tion public. In the aforementioned case studies the
Over the past decades a shift appears to have
choice between standards options is seldom based on
taken place from a style of standardisation predomi-
technical superiority. More often, what seems to be a
nantly based on technological considerations towards
formal technical discussion turns out to be the nego-
a process with strong political and economic over-
tiation of interests (e.g. Sirbu and Zwimpfer [4]).
tones (e.g. Irmer [I]; Van Rooij [2]). It is now more
Non-technical interests are translated into ;an institu-
or less understood that economic and political inter-
tionally legitimate type of reasoning, that is, a tech-
ests determine the outcome of negotiations in stan-
nical type of reasoning [5]. The conclusion of these
dards committees. Studies of standards processes
studies is clear: technological knowledge is predomi-
buttress this view. They do so by setting it alongside
nantly a resource for strategic action, and not a
the traditional factor of ‘technological k:nowledge’.
decisive factor in itself.
Active contribution and participation in the standards
Although the arguments of and the consensus
debate - a significant determinant of the standards
between these studies are persuasive, a number of
process [3] - requires technological knowledge. But
objections can be made. To note one, which is not
the information which industrial participants in par-
likely to get attention in current work: standardisers
ticular are willing to divulge is limited. There is little
who are partly or wholly motivated by true technical
concerns are not accounted for. A more fundamental
. E-mail: egyedi@~tm.tudelft.nl objection is that, by setting the factor ‘economic and

0920-5489/97/$17.00 G I997 Elsevier Science B.V. All rights reserved.


P/I SO920-5489(97)00001-9
432 TM. Egyedi/ Computer Standards & Interfaces 18 (1997) 431-450

political interests’ against that of ‘technological field determine the outlook of practitioners partici-
knowledge’ an important link between the two is pating in standardisation. Their outlook will impact
severed. On closer inspection, the lack of influence committee contributions and the outcome of stan-
of technology-related views is less evident. In many dards negotiations. (Other, individual factors of in-
of the controversies described in the literature (e.g. fluence on committee standardisation will not be
Sirbu and Zwimpfer [4]; Abbate [6]), standpoints are discussed here.)
characterised by other cognitions than those of an There are three reasons to deviate from the well-
economic, interest-oriented nature. The controversies tried route. Firstly, there is insight to be gained by
show a pattern. The interests are organised according using a different approach as it highlights other
to paradigms and concern technology-related out- (aspects of) events. Secondly, too little attempt has
looks. Their basis of organisation would appear more been made to falsify or discover the ltmits of the
relevant in explaining the standards process than the interest-driven explanation. Without too much ado
economic and political interests themselves. Such most events can be interpreted in terms of economic
interests are symptoms of different paradigms. interests. What is needed are conceptual tools which
Thereby the role of technological knowledge is that explain these interests. A third and related reason is,
of paradigm articulation. as implied, that where paradigms and interests ac-
The term ‘technological paradigm’ refers to the count for the same occurrence, the paradigm ap-
set of technology-related cognitions which structure proach may provide a more fundamental - and thus
and focus the behaviour of practitioners in a field of encompassing - explanation. The paradigm ap-
technology. ’ Shared methods, heuristics and rules proach promises to expose recurring patterns in in-
develop. Exemplars, that is exemplary achievements terest negotiation. I will return to the issue of levels
in the practitioners’ field [7], further focus activities. of explanation in the concluding section (Section 7).
A new theory, method or model may serve as an In this article the extent to which technological
exemplar. These and the beliefs and expectations of paradigms explain the outcome of committee pro-
the practitioner community structure actions. These cessesfor telematic services is explored. Three cases
actions can be understood in terms of structured of Open Systems Interconnection (OSI) standardisa-
search processes and follow patterns of ‘normal’ tion are studied in the period 1979-1988. It is a
problem-solving. This set of cognitions roughly cor- period in which two different technological fields
responds to what Kuhn initially called a paradigm and two formerly disparate economic segments are
[7] in respect to science. In Kuhn’s words, paradigms in the process of integrating 193.Therefore, the stan-
focus on ‘puzzle-solving’. As a result, paradigm dardisation mechanisms involved are foremost held
conform activities show continuity. The string of to be generalisable to other areas of converging
successive activities forms a path, so to speak, which technologies (e.g. telecommunications and broadcast-
in the field of technology dynamics is called a ing).
rrajectory [8]. The cases are situated in the formal standards
The field of interest in this article is that of environment of the International 0rga:nization for
telematic services. The article takes the view that Standardization (ISO) and the ComitC Consultatif
shared heuristics, knowledge, operational experience, International Telegraphique et Te’lephonique of the
beliefs, expectations, interests and exemplars in this International Telecommunication Union
(ITU/CCITT). The difference between formal, grey
(e.g. ATM Forum and IETF) and de facto standardis-
ation and the influence of formal committee proce-
’ Dosi’s definition of technological paradigm “as a ‘model’ dures on the standards process are dealt with else-
and a ‘pattern’ of solutions of selected technological problems, where (i.e. in Egyedi [ 101).
based on selcctrd principles derived from natural sciences and on To start with, different types of standards in the
.w/~cred material technologies” ([13], p. 152) [no emphasis of
field of telematic services are introduced (Section 2).
mine; TE] has a limited scope. It is useful for fundamental
technologies, but it is less appropriate for applied areas of technol-
Next, shared, orienting principles in the field of
ogy such as telematic services. telematic services are explored. Paradigmatic ele-
TM. Egyedi/ Computer Standards & Interfaces 18 11997) 431-450 433

ments in informatics, telecommunications and telem- The third and last category of standards noted
atics are examined (Section 3). They offer clues as to here is that of the reference frameworks, also called
which factors may be relevant in the cases described reference models. These describe the logical struc-
in Section 4. The cases concern three base OS1 ture of a system. There are several: the OS1 manage-
standards: the transport, session and message han- ment framework, the Reference Model for Open
dling protocols. Subsequently the case findings are Distributed Processing, the Reference Model Inte-
analysed. First, the relation between the set of princi- grated Broadband Communication, etc. [ 111.
ples which underlie OS1 standardisation and the OS1 The terms ‘reference model’ and ‘architecture’
standards trajectory is examined (Section 5). Next are sometimes used as synonyms. However, strictly
committee processes and the controversies therein speaking a reference model is an implementation
are analysed from the paradigm perspective (Section independent framework, whereas the word ‘architec-
6). The conclusion answers the question whether a ture’ refers to the functional representation of a
technological paradigm approach is worth pursuing range of concrete product standards. Examples of
when analysing committee processes (Section 7). early communication architectures are the propri-
etary Systems Network Architecture of IBR4 (SNA,
1974) and the Transmission Control Protocol/Inter-
2. Standards in telematic services net Protocol (TCP/IP) architecture on which Inter-
net is based. Development of the latter started in
Telematic services consist of a set of functions, 1973 with the Defense Advanced Research Projects
which are specified in base standards. Base standards Agency (DARPA) project of the US Department of
are the generic ‘building blocks’ of standards-con- Defense. Work on these two architectures preceded
form services. Such standards are issued in pairs of the Open Systems Interconnection Reference Model
concrete protocol definitions and abstract service (OSI-RM), the framework within which the stan-
definitions. The protocol definition contains elements dards discussed further on were developed.
which are essential to the service, indicates the pa- OSZ reference model. The OS1 model stems from
rameters among which one can choose, and often the formal international standards bodies of the IS0
includes some optional features. Thus, the implemen- and the CCITT. The impetus for the OS1 model
tor of a standards conform telematic service mini- came from the International Federation for Informa-
mally implements the essential features of these tion Processing (IFIP) Working Group 6.1 and the
building blocks and further needs to decide which European Computer Manufacturer’s Association
options are required for the service in mind. (ECMA). It was taken up by IS0 Technical Commit-
Options in base standards may cause problems of tee (TC) 97 in 1977 and subsequently by CCIIT
interoperability. Such problems arise between service Study Group VII. The OS1 reference framework was
implementations which are based on different choices adopted by both standards bodies in 1984 (IS0 7498
in base standards. Therefore, a restricted number of and CC1l-I X.200).
sets of specified base standards with fixed options The OS1 model identifies and structures all the
have also been standardised. These are called pro- functions needed for datacommunication. It depicts
files or functional standards. A functional standard these as a set of hierarchically ordered layers (see
is a Fig. 1). These layers represent logically separate
functions of datacommunication, referred to above as
building blocks. They address a more or less circum-
‘(. . .) document which identifies a [base] standard or
scribed area of standardisation. Standardisation of
group of [base] standards, together with options and
layers can therefore take place in parallel. Seven
parameters, necessary to accomplish a function or a
layers are defined within the communicating sys-
set of functions. *’
tems. The framework specifies the service which
each layer should provide. A layer uses the service
’ ECITC Guide to IT&T TestIn: and Certification, BNSSek, of the lower layer and provides a service to the upper
Belgium, 1993. layer. Each layer contains tools, so-called entities.
434 TM. Egyedi / Computer Standards & Interfaces 18 (19971431-450

I Information structure
’ Telunatic service

Transport - - - - *
4

Nchvork - - - - ) 3 Network service


DataLink f - - - )

Physical f----C
I
/ Physicalmedia for OS1 ) service
levels

A B
Fig. I. (a) The OS1 reference model (source: X.200 CCITT Recommendation); (b) Levels in electronic services and segments in service
provision (source: De Jong [3 I]).

Together these entities provide a service [33]. In packaged set of data enters system 2 at the lowest
other words, a layer provides services which are layer and is passed on upwards. It is sluccessively
implemented as a function in that specific layer or unwrapped again until the bare set of data emerges at
which derive from its lower layers. (or above) the application layer. To interwork suc-
Services are abstractions. For example, the way cessfully, the two systems must agree how to inter-
data is routed through the network is defined at the pret the ‘wrappings’ at a certain layer level. This is
level of the network layer. This ‘routing service’ can the essence of layer protocols. The software of two
be provided in different ways (e.g. connectionless systems should conform to the whole sta’ck of layer
and connection-oriented, that is, with or without protocols if they are to interwork.
maintaining the connection between end systems In sum, a basic OS1 stack (management, security,
during the transmission). Likewise, there are differ- etc. excluded) comprises six basic protoc’ols and six
ent ways to coordinate the communication session service standards with additional generic and specific
(i.e. connect, maintain and break off the session), a application standards [ 1I]. Neither the basic proto-
service provided by the session layer. cols, nor the service standards specify how they
In order to communicate between different sys- should be implemented. This is up to the service
tems, peer-to-peer protocols are defined. Each layer providers.
has one or more protocols, which indicate how com- The case studies discussed in Section 4 concern
munication between the layer entities of both sys- base OS1 protocols.
tems should proceed. Roughly, one can visualise the
standardised datacommunication process as follows.
In system I a set of data is passed downwards from 3. Paradigms in the field of telematic services
the application layer to the medium. At each layer,
control information is appended which indicates how The field of telematic services is not clearly de-
the data should be processed and transmitted. The fined. It is roughly referred to as an area where
original data is, as it were, successively wrapped in informatics and telecommunications converge. This
different layer-specific control information. The complicates the identification of paradigms which
T.M. Egyedi / Computer Standards & herfaces 18f 1997) 43 I-450 435

are unique for the field of telematics. One or more tion will also be paid to exemplary functionalities
paradigms may be at stake. They may stem from the and successful applications.
two separate feeder disciplines and/or from the In the following sections paradigmatic elements
merger of both fields. Clearly, in order to warrant are explored in the feeder disciplines of telecommu-
separate attention paradigms in the field of telematic nication and informatics (Section 3.1) and in the
services would need to differ meaningfully from the field of telematic services itself (Section 3.3). Stud-
paradigms in informatics and telecommunications. ies of the X.25 datacommunication standard are dis-
In order to avoid an unproductive theoretical dis- cussed in Section 3.2. They point to issues with
cussion on paradigm identification, I focus on immediate relevance for the cases. Furthermore, in
paradigmatic elements. That is, the search is first of view of the cases, the focus is below on the period
all for shared outlooks and orientations which could leading up to the 1980s.
be relevant for standardisation in the fielld of tele-
matic services. 3.1. Outlooks of informatics and telecommunication
There are two leads which - apart from those practitioners
noted by Kuhn - are promising in identifying
paradigmatic elements in the field of telematic ser- “What if computermen invented the telephone?”
vices. The first is the notion of management domain. Whitehouse [33], who embarks on this mental exer-
In the field of information and communication tech- cise, concludes that the telephone would then have
nology components, subsystems and systems can be been a more sophisticated bit of machinery, but that
distinguished. It depends on the angle one takes, incompatibility between these sophisticated handsets
whether an artifact is labelled a component, a subsys- would have prevented the service from ‘becoming
tem or a system. It specifically depends on the widespread. His question refers to the marked differ-
management domain under consideration [12]. A ence in professional culture of informatics and
stand-alone personal computer is viewed as a system telecommunications and its implications for technol-
if it belongs to one management domain, namely the ogy development. The characteristics of the practi-
personal domain. When connected to a Local Area tioners’ communities as well as the context of opera-
Network of an organisational unit with centrally tion and exploitation in both fields differ [14]. Be-
located software and printers, both the computer and low, I focus on a number of central, field-allied
the printer are subsystems of the local area network. operational issues and technical features to sketch
Here the management domain is the organisational the outlooks of informatics and telecommunication
unit. In the case of telematic services, the system practitioners.
consists of the set of subsystems which jointly pro- Telecommunications. In the period of interest, the
vides the service. Subsystems are peripheral comput- monopolies of national Public Telephone Operators
ers, the network, switching equipment, communica- (PTOs) dominated the telecommunications’ scene.
tion facilities, etc. What makes the systern of telem- The telecom administrations demanded universal ser-
atic services interesting is that two or morseoriginally vice. With regard to the operational side of the
disparate management domains of telecommunica- network, PTOs were concerned with issumessuch as
tions and computers, each supplying a subsystem, reliable transmission and maintaining the integrity of
are supposed to provide one system service. In the the network. The operators primarily viewed
process of developing standards, domain-bound telecommunications in terms of centralised network
paradigmatic elements promise to be an important control [15]. Telecom products had a relatively long
characteristic. life-cycle, often spanning more than a decade [9].
The second lead stems from Dosi [I 31. Dosi in- The specific requirements of national PTOs were
cludes economic considerations in his identification catered for by operator-allied providers of equipment
of technological paradigms. This addition is of par- and switching facilities. In the international forum of
ticular relevance in applied fields of tech,nology such CCI’IT, national monopolies facilitated international
as telematic services. Thus, where shared outlooks in cooperation.
the field of telematic services are concerned, atten- Telephony strongly influenced the professional
436 TM. Egyedi/ Clmpurer Standards & Interfores 18 (1997) 431-450

culture of PTOs [12]. It determined the operators’ mainframes to mini-computers and micro-computers.
general outlook on telecommunications and service The speed of product development reduced the life-
provision. The telephone was a dedicated terminal, cycle of computer products to no more than a few
that is, fit for one purpose only. Dedicated telecom years. The price of hardware continuously fell. As
provisions were designed to meet this specific pur- the turnover rate of products was high, durability
pose. Telephony was based on circuit-switched and was not perceived to be an important issue. Applica-
connection-oriented transmission. 3 The traffic flow tion development focused on designing software
was more or less predictable (e.g. Schoute [ 161). The modules suitable for reuse. Upward and downward
bandwidth necessary for comprehensible telephone compatible software was needed to keep customers
conversation was known. The heuristics which char- committed to their original choice. A highly competi-
acterised the operation of networks were: fixed over- tive market for computer products developed.
head, fixed bandwidth, physical analogue connec- In the early phase of computerisation mainframes
tions and hierarchical network architecture [17]. Op- were extended with terminals and remote batch sta-
erational and technical principles for telephony were tions. Front-end processors dealt with communica-
passed on to newer areas of telecommunications tion aspects. Such computer systems had a restricted
such as telex, telefax and early datacommunication. scope, usually serving one organisation. Computing
Digitisation and the dissemination of computers facilities were accessible to those stationed outside
changed telecommunications. On the one hand com- the organisation via telephone lines. These early
puters became part of the telecommunication system computer ‘network’ systems were predominantly
[ 181. They were for example deployed to operate based on single vendor solutions.
networks (e.g. Stored Program Control in switching Local Area Network (LAN) developments started
systems). On the other hand, new uses of datacom- with the introduction of personal computers and
munication came within reach, demanding a new work stations. It was a period in which organisations
approach to networking. Next to the public datanet- increasingly used multivendor systems. Processing
works, privately operated multinational networks power, which was formerly provided centrally by
emerged (e.g. GEIS). These datanetworks, Wide Area mainframes, became distributed via high speed local
Networks (WANS), used leased telephone lines to networks. Computing and communication facilities
provide some of the earliest telematic services, such were taken over by network servers and by personal
as banking transactions and flight reservation sys- computers and work stations.
tems. Chronologically the onset of LAN developments
Informatics. Babbage’s concept of a multi-pur- followed that of the earlier mentioned WAN devel-
pose machine, his analytical engine, still applies to opments. LANs, however, also strongly affected the
our present day computer [ 191. The discipline of field of datacommunication. In a sense the LAN
informatics developed within the military environ- environment formed a restricted computer-oriented
ment (computational applications) and within the ‘test site’ for new developments in datacommunica-
business environment (administrative applications). tion. As LAN designers and users of computer and
The focus of computer researchers on analogue com- LAN facilities progressively extended their scope of
puter systems shifted towards digital computers. datacommunication across organisational bsoundaries,
Miniaturisation in micro-electronics decreased the a separate body of knowledge and experience
size of computers and increased their processing and emerged. The accompanying outlook on datacommu-
memory capacity. Hardware demand shifted from nication was based on personal and organisational
needs. User control was emphasised.
Thus, an important difference between telecom-
munications and informatics practitioners lis that the
’ I.e. a connection is established and maintalncd throughout the latter took the personal management domain of com-
communication session. In a connectionless servxe, on the other
puter users and the organisational LAN domain as
hand, each data unit has the destination address attached. Units are
routed independently though the network via different routes. No the starting point of computer communication, while
dedicated connection is made between end-terminals. the outlook of telecommunication practitioners cen-
TM. Egyedi/ Computer Standards & Interfaces 18 (1997) 431-450 437

tred on the network and departed from a public Different concepts of network design or architec-
domain view on datacommunication. The exemplary ture (hierarchical and symmetrical), and the related
function of telephony in telecommunication justifies location of network control also reappear in analyses
the use of the term ‘paradigm of telephony’. The of Sirbu and Zwimpfer [4] and Manse11and Hawkins
difference in orientation affected standardisation pro- [151. The technical features reflect opinions about the
cesses regarding datacommunication and - in its distribution of social and economic control within
wake - telematic services. and between networks. These opinions are voiced by
different groups of participants in standardisation.
3.2. Paradigmatic features in committee standardisa- Generally, controversies arise between the PTOs on
tion the one hand, and the private and experimental net-
work operators, computer manufacturers and com-
Paradigmatic differences between informatics and puter users on the other. Accordingly, the PTO-
telecommunications resurface in studies of standardi- dominated forum for telecommunication standardisa-
sation in the field of datacommunication. For exam- tion (CCITT) and the forum for computer standardis-
ple, Abbate [6] compares ARPA’s TCP/IP and ation (ISO) can be identified as sparring partners.
CCITT’s X.25. The two embed different concepts of In order to distil the paradigmatic features from
internetworking (distinct heterogeneous networks controversies in committees, the case of the com-
versus a single homogeneous network), different puter communication standard X.25 for packet-
ideas about the location of control in networking switching [4] is reviewed. The standard is of particu-
(decentral or central), different expectations about lar interest since Sirbu and Zwimpfer qualify its
future demands on networks (e.g. in relation to the development as a political process. Contrasting views
required reliability and user needs), and different on datacommunication are highlighted and sum-
expectations about the telecommunication market marised in Table 1 together with the differences
(e.g. role of private networks). mentioned earlier.

Table 1
Paradigmatic aspects in the outlook of telephony- and datacommunication-oriented practitioners in the 1970s and early 1980s
Paradigmatic aspects Telephony Datacommunication
Management domain Public Private/personal and organisational
Location of network control Central Decentral
(e.g. error and flow control)
Concepts of network design Hierarchical Symmetrical
Focus and source of revenues Transmission Applications
Technical features Circuit-switched/virtual Packet-switched/datagram
circuit connection-oriented connectionless drgital
analog fixed demand versatile demand integration
diversification trajectory trajectory (multi-purpose
(dedicated lines and terminals) terminals and transmission)
Operational features
(a) operational priority (a) integrity and reliability of network (a) flexibility for users
(b) required network reliability (b) high (b) low
(c) development horizon (c) long-term (c) short-ten-n
(d) tariff principle (d) distance (d) volume of flow
(e) market and -relations (e) homogeneous cooperation (e) heterogeneous competition
between monopolies
Interest groups PTOS Private and expertmental network
operators, computer manufacturers,
computer users
Formal standards fora CCITr IS0
438 T.M. Egyedi/Computer Standards & Interfaces I8 (1997) 431-450

The case of X.25. Development of X.25 took were reluctant to allow for incomplete stacks.
place in CCITT. Controversies centred on the choice Telecommunication operators, on the other hand,
between (connection-oriented) virtual circuit and urged for partial OS1 implementation which would
(connectionless) datagram packet switching. Roughly allow freedom of network operation. In line, the
speaking, the former option was put forward by X.25 recommendation was a interpreted as a so-called
public network operators. The latter was generally access protocol which, as the term implies, controls
preferred by private network operators and com- access to the network (and thus only demands a
puter-oriented participants (e.g. IFIP, ISO, ECMA partial OS1 stack).
and ANSI). They strongly felt the need for a connec- Thus, standardisation in datacommunication
tionless option next to a connection-oriented option. showed two distinct camps. Firstly, there was the
The technical decisions involved choices in net- group of PTOs who, analogous to telephony, started
work control (central or decentral) and network de- out with circuit-switched data networks (e.g. Ger-
sign (hierarchical or symmetrical) [4]. The decisions many and the Scandinavian countries). Secondly,
had functional and financial consequences. The there was the group of computer-oriented. practition-
standpoint of the PTOs was generally voiced in ers who focused on packet-switching. Temporarily a
terms of manageability of the network. Virtual cir- number of PTOs belonged to the second group,
cuit could better guarantee the integrity of the net- namely those which in the late 1970s and early
work. Centralised operation of the network was said 1980s started operating X.25-based data networks.
to allow better management of error control, flow Temporarily, because when the debate thereafter
control and packet sequencing. Control services fur- shifted to new areas the computer-oriented practi-
ther implied revenues for the PTOs. Furthermore, the tioners and the packet-switching PTOs again had
connection-oriented option in the X.25 committee opposing views (e.g. in respect to the question
process matched prior operating procedures of the whether multiplexing should be accommodated in
public telecom operators. the transport layer).
On the other hand, the connectionless mode pro-
vided greater flexibility for users (user control) and 3.3. Common orientations on telematic seruices
promised to be more cost-effective [4]. Users de-
manded tariff principles based on volume of flow Since both the work areas of informatics and
instead of distance [17]. The overhead needed for telecommunications assimilated microelectronics, de-
virtual circuit communication was held to be wasted velopments therein incited similar technological pat-
for simple transactions. Moreover, it was thought terns. Digitisation is a case in point. In the field of
that users would to a large extent protect themselves mass- and telecommunication, it made the earlier
against network errors by duplicating virtual circuit distinctions between modes of information and trans-
service functions in end computers. Computing power port media superfluous. The integration of diverse
had become cheap and users needed total recovery of services came nearer (e.g. Slaa [21]). The same
data in case of a breakdown [20]. integration trajectory was apparent in the field of
Briefly elaborating, IS0 was hardly involved in informatics (e.g. the multimedia programmes).
the X.25 standards process. The X.25 recommenda- In respect to telematic services, two types of
tion was ready before completion of the OS1 refer- focusing mechanisms can be distinguislhed: exem-
ence model. Fitting the X.25 recommendation into plars and integrating concepts.
the OS1 model would later prove somewhat difficult. Exemplars. Exemplary achievements in the field
The fact that computer-oriented practitioners treated of telematic services are of two kinds. Firstly, there
the computer as a unit and that telecommunication are the early - primarily technological - exemplars.
operators treated the network as a unit led to differ- Many of these stem from research and development
ent views on the application of OS1 protocols. The on the ARPA network of the US Department of
IT industry lobbied for the ‘Royal Route’, that is, for Defense (e.g. Terink [34], p. 142). For example,
using the full stack of OS1 protocols with no restric- ARPA’s Computer Based Message Systelm served as
tion on how to implement the stack. Initially they an exemplar for development of electronic mail sys-
T.M. Egyedi/ Computer Standards & Interfaces I8 (1997) 431-450 439

Table 2
Estimated running start Formalised
Organisational integration mid- 1980s 1993 ITU-T/JTC 1 Guide for cooperation
Cognitive integration 1979 1984 OS1 reference model

terns. Secondly, there are the economic exemplars. separate traditional organisational and operational
For example, the french concept of TClCtel (also domains of telecommunications and informatics.
known as Minitel) entailed the use of the electronic While the figure which illustrated the OS1 ,model in
telephone book as a trigger service and freely dis- the two standards did not make a distinction between
tributed terminals. These means were applied to lower and upper layers, many textbooks did. ’ Some
attract users in the initial phase of videotex services. drew the line between layer 3/4, others between
The Tc%tel concept guided videotex activities in layer 4/5 [20]. In other words, cognitive integration
many countries. However, as its example primarily started out with the allocation of work areas either to
consisted of applying these two ‘rules’, TCle’tel’s the domain of telecommunications or to the domain
impact on telematics practitioners remained re- of informatics. Integration needed to evolve. Like-
stricted. wise, organisational backing for cognitive integration
Integrating concept. Where different fields con- in standardisation needed to develop. The fields of
tribute in new areas of technology, integrating con- informatics and telecommunications had their own
cepts are needed to forge a common cognitive basis international standards bodies for telematic services:
[22]. The OS1 framework provided such integration IS0 (later ISO/IEC Joint Technical C’ommittee
for standardisation. The aim of the OS1 model was (JTC) 1) and ITU’s CCITT (now ITU-T) respec-
“to provide a common basis for the coordination of tively. Both IS0 TC 97 SC 16 and CCI’FT Study
standards development for the purpose of systems Group VII addressed OS1 standardisation. 6 In the
interconnection.” 4 With its approval as an Intema- late 1970s and early 1980s cooperation between IS0
tional Standard and CCITT recommendation in 1984, and CCITT showed signs of strain. Collaboration
OS1 had the formal support of the computer-oriented between committees of different standards bodies
and telecommunications standards communities re- amounted to division of labour and of responsibili-
spectively. It directed attention, posed specific prob- ties. In respect to the early years of OSI-guided
lems and focused standards work. For example, standardisation, collaboration consisted predomi-
nantly of information exchange. Thereby CCITT and
IS0 committees preserved their autonomy.
‘All CCITT study groups dealing with ISDN and
Concentrating on the time of formalisation, cogni-
associated plethora of new or expanded services are
tive integration seemed to leap almost a decade
guided by OS1 (Wallenstein [23], p. 891.’
ahead of organisational integrative measures.
Whereas the OS1 framework was officially published
The work on OS1 standards displayed the charac- in 1984, more structural forms of coordination be-
teristics of normal puzzle-solving. In this respect OS1 tween the IS0 and the CCITT started in the late
can be termed a paradigm in standardisation. 1980s. It took until 1993 to formalise the cooperation
A second aim of OS1 was that computer applica- (see Table 2).
tions should be as independent as possible from
transmission and communication networks [24]. This
was interpreted in textbooks as a perpetuation of the -E.g. Carter (Ed.), Structure in Data Communications, ICL.
London, 1983; M. Schwartz, Telecommunication Networks: Pro-
tocols, Modelling and Analysis, Addison-Wesley, Reading, 1987.
6 IS0 TC 97 was later superseded by JTC I. Before that, the
4 IS0 Open Systems Interconnection Basic Reference Model, subcommittee structure had already been altered, TC 97 SC 16’s
IS0 7498, IS0 Central Secretariat, Geneva, 1984. work on OS1 was reassigned to TC 97 SC 2 I.
440 TM. Egyedi / Computer Standards & Interfaces 18 (1997) 431-450

Thus, the most significant part of early OS1 stan- include their interest-based analyses. * The cases are
dardisation proceeded in spite of the lack of a firm, analysed in Sections 5 and 6.
integrated institutional basis. ’ It is against this back-
ground that the standards discussed below were de- 4.1. Transport protocols
veloped.
In sum, there were common orientations evident In 1979 IS0 and CCILIT were already working on
in the emerging field of telematic services: techno- Standardisation in the area which was later to be-
logical and economic exemplars of the ARPA net- come assigned to the transport layer of the OS1
work and of T&Ctel, respectively, and the (cognitive) model. Both organisations were aware of each others
integrating concept of the OS1 framework. However, developments, but no coordination took place until
these elements are too weak and unrelated to warrant the matter was taken up in an IS0 working group.
the term ‘paradigm of telematic services’. The CCITT was represented in this working group,
in which also ECMA members participated.
At the time CCITT had high hopes for teletex.
4. Cases of base OS1 standardisation Teletex is a service which interconnects text-based
terminals. ’ A Special Study Group (SG XVIII) was
The early period of OS1 standardisation, the pe- close to completing the teletex Recommendation
riod in which organisational coordination lags behind S.70. It was to be approved in 1980 (ITTJ/CCITT,
cognitive coordination, is an interesting one. One Yellow Book, 1981). The PTOs therefore required a
may presume that limited cooperation will h&e pro- transport protocol which would support the teletex
longed a situation wherein the different ideas preva- standard.
lent in the two standards bodies govern the content Another main issue concerned ECMA’s proposal
of contributions. One would therefore expect that to incorporate multiplexing, a technique to project
paradigmatic elements would influence early work two or more logical connections onto a physical
on OS1 standards and become embedded in stan- connection, in the transport layer Protoc:ol. Multi-
dards. plexing was already part of the pTOs’ X.25-based
Below, the different outlooks in contributions to service (network layer protocol). As each logical
OS1 standardisation are examined between 1979 and connection involved revenues, multiplexing at the
1988. The cases concern standardisation of the trans- level of the transport layer would mean a loss of
port layer, session layer and message handling proto- revenues for PTOs.
cols (OS1 layers 4, 5 and 7 respectively). The devel- The controversies were difficult to solve. Differ-
opment of the transport and session protocols took ences of opinion resurfaced in the compromise which
place in the late 1970s and early 1980s. Standardisa- was conjured. Five transport protocol (TP) classes
tion of message handling started at the same time, were adopted in the standard. The PTOs, concern for
but spans a longer period. (See Table 3.) teletex was accommodated in class 0 (TPO). TPO
Interviews were held with former committee par- provided minimum functionality. It simply passed
ticipants. The case-descriptions below are primarily
based on their account of events and hence also

’ There is an obvious methodological risk in using interviews.


Interviewees arc likely to remember the salient controversies and
7 The following quote illustrates the institutional problem at the not points on which participants agree. Likewise differences of
time: “[ISO] made special efforts to achieve stability of both interest are more conspicuous than paradigmatic differences. Even
Session and Transport Layer standards in tune for the 1984 when prompted for such information. interviewees often have
CCI?T Plenary meetings. In order to amve at mutual texts, difficulties in expressing this tacit knowledge’. Thus, indications
CCITT and IS0 have held joint meetings. In strict terms this is of paradigmatic elements related to telematics are less likely to
not allowed by CCITT rules but meetings have been arranged to become clear.
take place concurrently (...) so that a free Interchange of dele- 9 Teletex is a “service (. .) enabling subscribers to exchange
gates between CCITT and IS0 could occur” (Knightson et al. correspondence on an automatic, memory-to-memory basis via
[251. pp. 18-19). telecommunication networks” (CCITT Recommendation X.200).
T.M. Egyedi/ Computer Standards & Interfaces 18 (1997) 431-450 441

Table 3
The cases of base OS1 standardisation and the period of interest
Case studies Standard Start * End
OS1 1979 1984
Case 1 Transport 1979 1984

Case 2 Session 1982 1984


Presentation 1984 1988

Case 3 Message CCI-IT 1979 CCITT 1984 Revised by CCITT/ISO 1988


Handling IS0 1979 IS0 1984

* The start of standardisation activities are estimates

data on to the higher level. TPO assumed that the Us&s of the ARPA network sought compatibility
underlying network was reliable, thus no error facili- with TCP/IP protocols.
ties were included. Class 1 contained error recovery The compromise of five different protocol classes
mechanisms. It was a concession from ECMA in that complicated interworking [26]. In order to alleviate
it did not include multiplexing. The interests of the interworking problems, means were developed which
computer-oriented practitioners were represented in allowed a certain amount of negotiation of protocol
the other three classes. Class 2 supported multiplex- classes. In addition, profiles were developed for
ing. Class 3 combined multiplexing with the error specific applications which defined a fixed OS1 pro-
recovery facilities of class 1 and differed from TPl tocol stack, including the necessary transport proto-
only in the sense that it supported multiplexing. col class. For example, the use of TPO and TPI was
Class 4 of the transport protocol met the require- prescribed for CCIl’T’s message handling recom-
ments of United States’ participants, who sought mendation (X.400). lo All in all: however, the num-
compatibility with ARPA’s TCP. The primary re- ber of classes prevented the transport layer standard
sponsibility for future changes in the protocol classes from being a generic building block for upper layers.
was shared between CCITT and ISO. It was agreed
that CCI’IT would be primarily responsible for TPO 4.2. Session protocols
(simple class) and TPl (error recovery class). IS0
would be in charge of TP2 (multiplexing class), TP3 From 1982 onwards detailed work took place on
(error recovery and multiplexing class) and TP4 the session layer. Again, protocol developrnent took
(error detection and recovery class). Where changes place within an IS0 working group. This time the
were pending, the other organisation would be in- process started out with a gentleman’s agreement to
volved. make the session protocol compatible with CCITT’s
In sum, decisive factors in the standards process teletex S.62 recommendation. ” There were two
were, firstly, the PTOs’ vested interests in and rev- contributions to consider. First of all there were the
enues from X.25 products (in relation to multiplex- session functions of the teletex standard, a solution
ing and error control). A second factor was their which the participating PTOs favoured. Because of
interest and expectations in respect to teletex, a previously developed teletex products, PTOs wanted
dedicated, text- and document-oriented terminal. to be able to claim conformance to the OS1 session
ECMA members, on the other hand, took infor- layer. The second input consisted of the ECMA 75
mation processing in general as the starting point for protocol, which was championed by those with a
discussions on the transport layer. Computer-oriented
participants were mainly driven by their expectation
that several terminals would need to operate simulta-
neously over a single connection [25] and by their lo CCITT later dropped this requirement.
desire for more user control in order to !save costs. ” CCITT S.62 was later renamed CCI’IT T.62
442 T.M. Egyedi/ Computer Standards & Interfaces 18 (1997) 431-450

Table 4 mentors of the session protocol would most likely


Contribution Functional units implement either the ECMA or the CCITT combina-
ECMA Minor synchronisation, major synchronisation. tion of functional units. The choice again invited
resynchronisation interworking problems.
ccn-r Minor synchronisation. activity management
Outcome Minor synchronisation, major synchronisation,
resynchronisation, activity management 4.3. Message handling protocols

Standardisation in the area of message handling


was prompted by an IFIP working party in the late
1970s. The IFIP group formulated user requirements
computer-oriented background. At the time there for what was then called Computer Based Messaging
were no products based on this standard. Systems. Both ECMA and CCITT responded. In
A synthesis of the two standards was sought
parallel the CCITT developed the X.4001 series and
which would be acceptable for CCITT, IS0 and ECMA worked on the Message Interchange Dis-
ECMA. The proposal to adopt encoding and services tributed Applications, a private network-based mes-
of S.62 and the services of ECMA did not work out.
sage handling system [27]. ECMA’s work was fed
The discussion focused on the functional units which into ISO’s development of the Message Oriented
regulated the dialogue (see Table 4). Although the Text Interchange System (MOTIS). MOTIS activi-
services provided by the two sets of functional units ties took place in IS0 TC 97 SC 18. Many IS0
differed, by and large similar functions were offered, participants were ECMA members. After MOTIS
for example: document interrupt and document re- acquired the status of a draft international standard,
sume (activity management in S.62) are a form of
IS0 decided to stop its development and join forces
synchronisation and resynchronisation (ECMA 75). with CCI’IT on further development of the X.400
The evolution of the session layer was a painful series (approved by CCITI in 1984).
process [26]. ECMA’s and CCITT’s work on the Whereas little cooperation existed prior to 1984,
session protocol had taken place in entirely separate thereafter liaisons between CCITT and IS0 strength
environments. The effort vested in their own stan- ened. Joint meetings were held during the CCITT
dards made it difficult to reach consensus. Further- Study Period 1985-1988. These were alternately
more, participating carriers (e.g. Germany and labelled IS0 and CCITI meetings. Parts of ISO’s
Canada) who were members of both the IS0 group draft MOTIS, such as the concept of a message
and the CCITT group also fought for CCITT’s tele- store, were incorporated into the 1988 version of the
tex-based view within IS0 committees. In other X.400 series [32]. Elements of ISO’s (draft which
words, even within IS0 consensus was difficult to would not be assimilated were to be forwarded by
achieve. The standards process stagnated. the X.400 systems without modification (transparent
Despite the overlap between ‘activity manage- transportation). i2
ment’ and ‘major synchronisation and resynchronisa- The process resulted in a single text, processed by
tion’ all functional units were adopted. In other both IS0 and CCITI. On some issues the two
words, there was no compromise. As one intervie- standard groups did not reach agreement. In such
wee put it, cooperation was forced on them by the cases the standard recorded both standpoints, their
OS1 decision mechanisms of IS0 and CCITT. differences and how they were to be applied by IS0
According to participants, lack of willingness to and CCITT. Thus, almost identical standards for
cooperate fully frustrated the development of the message handling systems resulted: the CCITT X.400
session standards (protocol and service). There was
no viable technical reason for the resulting overlap
of functionality. Even when taking into account the
existence of CCITT products, a common technical
solution could have been reached. ‘* Perlee et al. [28], Data Communicatie Atlas, ‘Technical Part.
The consequence of the overlap was that imple- section B, p. 5- 1.
T.M. Egyedi/ Computer Standards & Interfaces 18 (1997) 431-450 443

Table 5
Summary of the main vantage points in the cases of OS1 standardisation
Cases Developed in Vantage points PTOs Vantage points Main outcome
computer-oriented participants
Transport IS0 No multiplexing Multiplexing and error 5 protocol
layer and error control control (user control classes
(safeguard revenues saves costs)
network layer)
Teletex compatibility No dedicated end-systems
TCP compatibility
Session layer IS0 Vested interests in Vested standards work Large overlap in
teletex products (CCITI in ECMA 75 chosen functional units
S.62 compabbility)

MHS (1988): Separately in Downward compatible Compliance to new PRMD/ADMD in name


X.400 series IS0 and CC1T-f with X.400 (1984) OS1 developments
and MOTIS until 1984 (presentation and
application layer)
Central role for No distinction PRMD/ADMD Compatibility with
m-0(ADMD) X.400 (1984)
Simple and consistent More complex functionality
standard for networking

(1988) and the (new) IS0 MOTIS standard, now In ISO’s perspective there was one domain. The
referred to as Message Handling System (MHS). standard adopted the term ‘ADMD’, but gave it the
One of the differences in the 1988 standards was significance of a routing domain (instead of a man-
CCIIT’s distinction between private or business net- agement domain). I4
works (the Private Management Domain, PRMD) Another difference between the CCITT and IS0
and the public network (the Administrative Manage- versions concerned the measure of compliance sought
ment Domain, ADMD). Historically the public net- with the 1984 version of the X.400 series. The
work was the domain of the PTOs. In CCI’TT’s view CCIIT version held that implementations of the
PRMDs were end systems. CCITT X.400 (1988) 1988 recommendation should conform to X.400
treated it accordingly. That is, in the figure which (1984) recommendations. Downward compatibility
CCIIT used to illustrate the relation between was of importance, as many PTOs which imple-
ADMDs and PRMDs, private systems were not di- mented the 1984 version were still operating it in
rectly connected (not: PRMD-PRMD). The figure in 1988. IS0 did not demand downward compatibility.
the recommendation suggested that a public operator From ISO’s point of view it was even preferable to
was necessary as an intermediary (PRMD-ADMD- stick to the 1988 version. The reason was a:sfollows.
PRMD). This applied for PRMD connections within Since 1984, a framework and a number of building
countries as well as for crossborder connections. I3 blocks had been defined for the application layer.

I3 See, e.g. Holler 1291, p. 168. The omtsston of dtrect PRMD


connections in the CCI’IT figure was intended to prevent “open- I4 One of the ISO-initiated discussions centred on the necessity
ing a can of worms”, as one interviewee put it. On ckoser reading. of identifying the ADMD. If a message could be sent from one
PRMD-PRMD connections are allowed withm countries. Where PRMD to another without identifying the ADMD (e.g. by putting
cross-border connections are concerned, the situation depends on in a blank instead of the ADMD name) it might allow systems to
the rules of the governments involved. automatically choose the cheapest way of routing.
444 T.M. Egyedi/Computer Standards & Interfaces 18 (1997) 431-450

The presentation layer protocol had also been com- the OS1 trajectory (i.e. the succession of OSI-related
pleted. Of course, the X.400 (1984) version did not activities). Layer interdependence arose from the
comply with these new OS1 developments. For ex- combination of (a) higher level use of lower level
ample, in the 1984 version the presentation layer was services, (b) the development of base OS1 standards
empty. IS0 advocated adherence to the principles of roughly from the lower layers upwards, and (c) the
OSI, such as using all layer functionalities. In 1988 accommodation of prior non-OS1 standards within
this meant implementing the presentation layer pro- the OS1 framework. Some of the problems encoun-
tocol as well. IS0 further did not want to burden tered and provisionally solved in early standardisa-
new implementors of X.400 (1988) with compliance tion affected higher level OS1 standards and the
to 1984 standards, as compliance to functionalities interoperability of OS1 implementations. These as-
which had since been revised, would make the sys- pects are elaborated below.
tems based on the 1988 version less transparent. One of the aims of the OS1 framework was to
Moreover, for implementors compliance would in- allow “existing standards to be placed into perspec-
volve extra effort with little added functionality. tive within the overall Reference Model”. This
In sum, differences arose from ISO’s emphasis on should cater for the sunk costs in standards which
matters concerning information processing and pri- existed before the OS1 framework and its base stan-
vate networks and CCITT’s emphasis on intercon- dards were completed. CCITT’s X.25, teletex and
nection and ADMD. The PTOs’ main concern was the message handling standard are examples of stan-
the workability of the network, which meant consis- dards which were developed prior or parallel to the
tent, and to a certain extent, simple standards. From OS1 framework. These and ARPA’s TCF’ co-shaped
their point of view, IS0 participants sometimes the background against which negotiations on the
seemed to indulge in personal hobbies, academic ‘empty’ OS1 layers took place.
thinking and superfluous embellishment. Some ideas When projected onto the OS1 model, the X.25
stemming from IS0 were deemed attractive, but too standard for packet-switching (1976) could not be
expensive and difficult to implement. This applied assigned to one of the layers. It combined both data
for example to the freedom of choice in routing link and network functions. In spite of the layer
messages and message circulation. l5 (For a sum- alignment intended by the OS1 model, X.25 was
mary of the above, see Table 5.) accommodated in OS1 layers 2 and 3.
X.25 contained multiplexing facilities. Multiplex-
ing in layer 3 foreshadowed negotiations at the level
5. OS1 and the measure of integration of the transport layer (layer 4). Because X.25 was
accommodated in OSI, the PTOs disputed the need
The OS1 framework, with its own paradigmatic for repeated multiplexing facilities at the transport
characteristics in respect to standardisation, func- level (TP2 and TP3). The point of incorporating TPO
tioned as an integrating concept. The set of princi- and TPI next to TP2 and TP3, which only differed
ples involved focused standards work. As was illus- on the issue of multiplexing, was heavily debated.
trated by the case studies, committee participants In another manner, teletex (1980) ‘fixed’ the
explicitly referred to OS1 principles in standardisa- specifications of the empty OS1 layers. Il. was retro-
tion controversies (e.g. the principle of non-empty spectively accommodated in the application layer,
layers). However, the set of principles itself posed that is OS1 layer 7. However, the teletex service also
some difficulties. Friction existed between the layer- expected the lower layers to provide certain intercon-
aligned, modular approach and the interdependence nection functions. These functions were part of the
of OS1 layers, which would have consequences for teletex standard.

‘( . . .) no transport layer was catered for in the


I5 The idea draws on the way (paper) magazines are sometimes original MHS and Teletex proposals, and this is in
circulated. effect tantamount to using TPO. Indeed TPO was
T.M. Egyedi/Computer Standards & Inte$aces 18 (1997) 431-450 445

designed to be compatible with the (rudimentary) all session layer options. This requires more work
Teletex transport protocol (Marsden [26], p. 333) l6 and more resources (e.g. memory). Implementation
of the complete session protocol, is therefore only
feasible for applications which run on large comput-
The case studies illustrate how earlier defined
ers.
lower layer functions in the teletex standard affected
Minimal implementations of FTAM or X.400,
the development of base standards for the transport
based on either the ISO/ECMA or CCITT session
layer (TPO) and the session layer (e.g. teletex set of
variant, are not suitable for elaboration with applica-
functional units).
tions which require the other session mechanism.
Thus, even before the OS1 framework was ap-
Different session software is concerned. Many
proved, prior implemented standards strongly re-
providers offer services up to and including the
stricted the latitude to define the content of the
presentation layer. Which (groups of) applications
layers. Its consistency was compromised b’ottom-up
can run on the protocol stack, depends on the way
by X.25 and top-down by teletex. The first defeated
the session layer is implemented. Others provide the
the objective of layer alignment; the sec:ond im-
whole stack, including for example X.400. In prac-
pugned 0%‘~ aim to assign each layer a function
tice this means that to exchange Electronic Data
(i.e. TPO). Overlap between layers (multiplexing in
Interchange (EDI) messages between two systems -
the network and the transport layers) and overlap-
one of which runs ED1 on top of the X.400 protocols
ping functionalities within layers (session layer) again
and the other running on FTAM - session layer
thwarted OSI’s objective of conciseness and econ-
software is required which supports both variants. (In
omy of options.
its absence a separate gateway would be needed to
Overlapping functionalities in the session layer
map the protocols onto one another.) Under normal
affected the interoperability of application services.
circumstances overlap in the session layer leads to a
As noted, ISO/ECMA’s and CCITT’s overlapping
bifurcation in OS1 applications.
and incompatible solutions for synchronisation and
In sum, standards developed prior to the OS1
resynchronisation were both adopted in the session
framework led to inconsistencies in the OS1 philoso-
protocol. This had consequences for the interoper-
phy. The OS1 model was initiated to rationalise
ability of, for example, the X.400-based message
standardisation and integrate standards activities in a
handling service and the File Transfer, Access and
merging field. Practice diverged: double multiplex-
Management (FIAM-)based file transfer service de-
ing functions in the network and the transport layer,
veloped in IS0 (in which ECMA also played a role).
parallel protocol classes in the transport layer and
Implementors of FTAM used the ISO/ECMA solu-
strongly overlapping functions within the session
tion for the session service. Conversely, CCITT’s
layer. Two clusters of application layer implementa-
X.400 (1984) best fitted CCITT’s session option.
tions emerged, one based on the IS0 version of the
The dilemma referred to here centres on the way
session protocol, the other based on the: CCIIT
the session layer is implemented. The session layer is
version.
usually not offered as a separate product. To support
OSI-based ex ante standardisation did not cut
both FTAM and MHS one would need to implement
back options as much as was hoped for. Options
were reduced through non-use. For example, TPl
and TP3 are currently seldom implememed. This
I6 The Message Handling standard (X.400 series (1984)) devel-
reduces in retrospect the number of options in the
oped approximately in parallel with the OSI framework. Parts of transport layer.
the standard migrated to OS1 (Marsden 1261, p. 91). A derivative
example of lower layer effects on upper layers is the way the use
of TPO in MHS led to the development of the Reliable Transfer 6. Paradigms in controversies
Service (a checkpoint recovery mechanism) in the application
layer. Because MHS was designed to work on TPO, and “cannot
rely on the integrity of the underlying connection, (. .) additional OS1 contributions developed in the separate envi-
procedures are required” (Marsden[26], p. 333). ronments of CCITT, IS0 and ECMA, whose mem-
446 TM. Egyedi/ Computer Standards & Interfaces I8 (1997) 431-450

hers had an influential role in ISO. Cooperation X.400 (1988) with X.400 (1984) services would
between the distinct standards committees proceeded ensure continuity, an option which most IS0 partici-
haltingly. Unchecked by prior consultation, field- pants did not favour. From ISO’s point of view it
specific viewpoints and interests entered the stan- was preferable to stick to the 1988 version since in
dards process. To what extent to do paradigmatic the meantime the OS1 presentation layer protocol
differences lie at the root of controversies? had been completed. IS0 advocated adherence to the
Many paradigmatic elements show up in the case principles of OSI, and therefore also implementation
studies. Most salient in the early days, was the of the presentation layer protocol. The differences
PTOs’ adherence to the paradigm of telephony, which were symptomatic for the divergent time scales which
they applied to standardisation in a new field of the two practitioner communities considered.
communication. For example, the PTOs’ high expec- An important issue with regard to standardisation
tations of teletex made them persist in their demands of the transport layer was where to locate control of
for teletex-compatible OS1 transport and session pro- the network. Telephony-oriented practitioners had a
tocols. Teletex fitted the PTOs’ general focus on the tradition of network control and set great store on the
written and spoken word. Like the telephone, the integrity of the network. Accordingly, they for exam-
teletex terminal was a dedicated terminal. The tech- ple argued against decentralised error control on the
nical and operational heuristics underlying its devel- level of the transport layer, which in addition would
opment followed those of telephony. The idea of mean loss of revenues. Conversely, lower costs of
teletex was at odds with the multi-purpose computer datacommunication, next to flexible use of network
concept and would not have been conceivable within facilities, motivated computer-oriented practitioners
the informatics community. The only way to under- to demand user control.
stand its impact on OS1 standardisation is by relating The issue of network control reappeared in dis-
it to the paradigm of telephony. cussions concerning the message handling protocols.
In contributions of computer-oriented practition- The PTOs in the CCITT defined private networks
ers, a well-defined paradigm is less identifiable. A (PRMD) as end systems in order to retain authority
clear focusing mechanism (exemplar) is lacking in the traditional domain of the public networks
which could specify the way certain recurring ele- (ADMD) and thereby safeguard the accompanying
ments tie together. However, the practitioners did revenues. IS0 participants, on the other hand, inter-
express a distinct, shared outlook. It consisted of preted the ADMD-PRMD domain distinction in terms
profession- and management domain-related ideas. of routing domains and thereby circumvented the
For example, they viewed datacommunication as an intended restriction of work area. The two bodies
extension of the personal and organisational domain, tacitly agreed to let the different interpretations go in
strived for flexibility in and user control of network order not to jeopardise correspondence between
facilities, and typically emphasised the multi-purpose CCIlT’s and ISO’s document on message handling.
function of terminals and the versatile demands which With regard to overall developments in OS1 stan-
this involved with regard to datacommunication. dardisation, the case findings indicate that the signif-
More specific, in three case studies computer-ori- icance of paradigmatic differences decline during the
ented practitioners were inclined to exploit the poten- second half of the 1980s. Gradually integration of
tial of the computer. During the X.400 (1988) pro- work fields and coordination of standards work within
cess they came up with some sophisticated ideas, the OS1 framework start to become manifest.
which the PTOs rejected. The PTOs emphasised A number of factors facilitated the transition. The
reliable networks and therefore to a certain extent most controversial issues had been tackled. The on-
simple network solutions. set of liberalisation in telecommunications changed
The PTOs’ view on future developments was the PTOs’ expectations and loosened the operational
based on continuity in service provision and was basis for the telephony paradigm. The teletex hurdle
interwoven with vested interests in X.25, teletex and had been cleared. (Regrettably for OSI standardisa-
X.400 (1984) products. Past developments had been tion, lack of demand for teletex was recognised
of an incremental kind. Downward compatibility of rather late.) Moreover, by the mid 1980s OS1 solu-
TM. Egyedi / Com,puter Standards & Interfaces I8 (I 997) 431-450 447

tions were being publicly acclaimed by users, gov- computer- and telephony-oriented practitioners to
ernments and groups of manufacturers [IO]. They base OS1 standardisation. Group-specific outlooks of
drew strong political support from the European CCITT and ISO/ECMA affected the standards con-
Commission, which created incentives for OSI-con- tent. Paradigmatic elements were, for example, ap-
form product demand and provided a favourable parent in the five transport protocol classes in which
regulatory environment. This supportive environment both central and decentral network control were rep-
also reinforced cooperation on OSI-related standardi- resented; in the teletex-compatible transport and ses-
sation and the development of a common outlook. sion protocols; and in the message handling proto-
In the OSI-TCP/IP debate, OS1 developments cols for which, for example, downward compatibility
have been much criticised - and often for the wrong of later versions was an issue. Compromises were
reasons [30]. Although the case studies add to such forged and unclear formulations were upheld in or-
criticism, it must be stressed that the aims of the OS1 der to allow PTOs and computer-oriented practition-
framework were ambitious. Summarising, two re- ers to continue with ‘normal problem solving’ in
lated factors hampered OS1 developments. The first their respective areas of interest.
was the mismatch between the degree of cognitive Next to paradigmatic influences, economic and
integration embedded in OS1 and the degree of or- political considerations (revenues, costs of datacom-
ganisational rapprochement between CCITI and ISO. munication and market control) entered committee
Early lack of cooperation slowed down OS1 stan- negotiations. The cases show how paradigms may
dardisation. Furthermore, OS1 committees had to contextualise such economic and political interests.
cope with two practitioner communities, which ad- For example, the PTO distinction between the ad-
dressed different operational environments (e.g. ministrative and the private management domain can,
area-specific government regulation, dissimilar ex- indeed, be interpreted as a means to stake off their
pectations about service demand, and different ideas operational domain against intruders; however, the
about the pricing of services). distinction may equally be understood as a symptom
The second factor was the allowances made for of the difficulties which ‘telephony’-oriented practi-
standards developed prior to the OS1 framework and tioners have in visualising the way networking should
the consensus demands of formal standardisation. take place without hierarchic, central network con-
Base OS1 standardisation took into account previ- trol. As was noted earlier, such interpretations are
ously developed services, working alternatives and scarce. Usually committee processes are one-sidedly
paper alternatives. The standards codified existing interpreted in terms of interests. No efforts have been
diversity and thus prevented OS1 from achieving the made to explore other grounds. Not surprisingly,
measure of economy and interoperability associated therefore, standardisation literature has never at-
with ex ante standardisation (i.e. standardise before tempted to relate interests to technological paradigms.
products have been developed and marketed). The Relationship between interests and technological
‘ex ante’ OS1 effort partly became an ‘ex post’ paradigms. A number of remarks should be made
answer (i.e. standardisation applied to products or concerning this relationship. Firstly, regardless of the
services which are available on the market). ‘technical’ associations which the term ‘technologi-
Despite these problems, OS1 prevented further cal paradigm’ may evoke, the first scholars which
unproductive variety. To my knowledge, no cursive used the approach took economic considerations to
new emphasis developments were initiated which be part of paradigm development.
competed with OS1 standardisation. Secondly, technology-related group interests - of
whichever nature - are shared cognitions in the same
way as shared technological heuristics, shared expec-
7. Conclusion and discussion tations, etc. They too add a focusing element to the
activities of practitioners.
The paradigm approach can to a large extent Thirdly, certain interests can be accommodated
account for the outcomes in OS1 standardisation. within the paradigm concept in a more embedded
There is a marked difference in the contributions of way. The reasoning here is as follows. Technological
448 TM. Egyedi/Computer Standards & lnterfoces 18 (1997) 431-450

heuristics develop in interaction with their opera- private and experimental network operators and to
tional environment. Practitioner communities in the collect the revenues which are involved in providing
field of telematic services, for example, assess devel- accompanying network control services. The
opments in service demand differently, take the paradigm perspective, on the other hand, clarifies the
turnover rate of business equipment into account in context in which such interests are put forward and
their ideas, favour different architectural designs, etc. explains the ‘naturalness’ of centralised network
depending on the market segment they operate in. control. The idea of central control was embedded in
Especially in applied fields of technology, paradigms technical provisions, in the expertise and operational
embody operational and economic heuristics. For experience gained in the public service environment.
example, the emphasis on network integrity and The PTOs’ emphasis on reliable networks, and thus
central network control is instrumental to universal network integrity in combination with central con-
service provision. Paradigm-based technology devel- trol, was not an improvised argument in. discussions
opment presumes a stable operational environment - on X.25. Nor - taking the example of teletex - can
an operational trajectory. Stability is in the interest of the ‘dedicated service approach’ be seen in isolation.
practitioners as changes in the operational environ- The latter is consistent with PTOs’ handling of tele-
ment could demand a change of paradigm and phony. Although, indeed, centralised network control
paradigms are not easily discarded. Paradigm adher- had beneficial economic consequences for the PTOs’
ents will do what they can to maintain the status quo and although the extension of the concept of dedi-
and will oppose changes. In other words, interest cated ‘end terminals’ may be interpreted as a means
negotiation in standards committees is partly to be to preserve the old order of network design,
understood as a means to preserve the operational paradigms better explain why in such matters certain
trajectory which supports the paradigm-based tech- technical options are chosen and not others, which
nological trajectory. also could have furthered their ‘economic and politi-
Thus, the paradigm approach offers leads to ac- cal interests’. All committee events can in retrospect
commodate technology-related interests. Technologi- be analysed as or reduced to interest-based occur-
cal paradigms ‘organise’ cognitions. They contextu- rences. However, certain controversies will not be
alise technology-related knowledge and interests. noted, and other occasions will too one-sidedly be
(Perhaps needless to say, one cannot always relate interpreted as interest-driven, if the influence of
interests to technological paradigms in a straightfor- paradigms is left out.
ward way. One probably should not try to do so Are there other arguments for applying the
where, for example, national or purely commercial paradigm approach? The paradigm approach ac-
interests in standards are evident. Other perspectives counts for standards contributions of those motivated
can aid in contextualising committee negotiations to participate because of their interest in the technol-
based on such interests.) ogy concerned. Furthermore, the approach explicitly
Controversies in standards committees which in- relates standardisation to technology development,
volve both interests and paradigms - are they better an area of research which by and large has remained
explained by the paradigm approach? The paradigm neglected. In particular in respect to converging tech-
approach may in some cases provide a deeper under- nologies the paradigm approach promises to be fruit-
standing of the committee process than an interest- ful.
based analysis. An example will illustrate what I
mean. The example concerns the development of the
X.25 datacommunication recommendation in CCITI’. Acknowledgements
It is chosen because this standards process is in
literature described as a political process while I I am strongly indebted to Cor de Jong and Arie
described it in terms of paradigms (Section 3.2). Rip for their coaching on the original chapter and to
From the political perspective, the PTOs’ adher- the anonymous reviewers for their suggestions on the
ence to centralised network control would be inter- final draft. The article is based on a chapter in [lo] of
preted as a means to fortify their position against which the overall content and summary are accessi-
TM. Egyedi / Computer Standards d Interfaces 18 (1997) 431-450 449

ble via Internet (candl.let.ruu.nl/www/staff/loef- or innovative technology? North-Holland, Amsterdam, 1989,


fen/cv/cvlO.htm). pp. 97-106.
1171 R.J. Solomon, New paradigms for future standards, Commu-
nications and Strategies 2 (2) (1991) 51-90.
[18] C. de Jong, Telecommunicatie: Techniek en nieuwe diensten,
References in: Verschuivende Toekomst, Studium Genetale, Delft Uni-
versity of Technology, 1990b (in Dutch).
[l] Th. Inner, The single global market of telecommunication [19] M. Campbell-Kelly, Geschiedenis van Informatieverwerk-
standards, 1992 Single Market Communications Review 2 ende Machines, De Informatiemaatschappij, Jubileumuitgave
(3) (1990) 56-58. van Natuur en Techniek, Maastricht, 1983, pp. 30-49 (in
[2] L. van Rooij, Geschiedenis van de intemationale normal- Dutch).
isatie, Lecture for Histechnica, Delft, 26th October 1991, [20] J.M. Lasso Peiia, Gpen systems interconnection (OSI) refer-
published in: De Ingenieur (1992) (in Dutch). ence model: Several perspectives of the telecommunication
[3] M.B.H. Weiss, M. Sirbu, Technological choice in voluntary and computer worlds, Graduation thesis, Delft University of
standards committees: An empirical analysis, Economics of Technology Faculty of Electrotechnology, 1992.
Innovation and New Technology 1 (1990) 111-133. 1211 P. %a, Naar een actorperspectief op technologie, Kennis en
[4] M.A. Sirbu, L.E. Zwimpfer, Standards setting for computer Methode 1 (1988) 22-42 (in Dutch).
communication: The case of X.25, IEEE Communications [22] A. Rip, A cognitive approach to science policy, Research
Magazine 23 (3) (1985) 35-45. Policy 10 (1981) 294-311.
[5] S.K. Schmidt, R. Werle, Technical controversy in intema- 1231 G.D. Wallenstein, Setting Global Telecommunications Stan-
tional standardization, MPIFG discussion paper (93/5), Max dards: The Stakes, the Players and the Process, Artech
Planck-lnstitut fur Gesellschafsforschung, Kb;ln, 1993. House, London, 1990.
[6] J. Abbate, From ARPANET to INTERNET: A history of [24] Th. Bruins, Inleiding OSI, in: C. de Jong, E.F. Michiels,
ARPA-sponsored computer networks, 1966-1988, Disserta- J.A.M. Nijhof, P. van der Vlist (Eds.), Handboek Telematica,
tion, University of Pennsylvania, 1994. Samson, Alphen aan den Rijn, 1990, pp. l-6 (in Dutch).
[7] T.S. Kuhn, The Structure of Scientific Revolutions, III, [25] K.G., Knightson, T. Knowles, J. Larmouth, Standards for
University of Chicago Press, 2nd ed., Chicago, 1970 (origin. Open Systems Interconnection, McGraw-Hill, New York,
1962). 1987.
181 R.R. Nelson, S.G. Winter, In search of useful theory of [26] B.W. Marsden, Communication Network Protocols: OS1 ex-
innovation, Research Policy 6 (1977) 36-76. plained, 3rd edition, Chartwell-Bran, Bromley, UK, 1991.
[9] J.C. Ambak, T.M. Egyedi, Development of telematic ser- [27] P.H.M. Vervest, Innovation in Electronic Mail, Towards
vices, in: J.S. Duisterhout, A. Hasman, R. Salamon @Is..), Open Information Network Perspectives on Innovation
Telematics in Medicine, North-Holland, Amsterdam, 1991, Policy, North-Holland, Amsterdam, 1987.
pp. 3-17. [28] J.A. Perlee, M.F.M. Born, H. Krijgsman-Heersink, Datacom-
[lo] T.M. Egyedi, Shaping standardization: A study of standards municatie Atlas, Expertise Centrum, ‘s Gravenhage, 1990 (in
processes and standards policies in the field of telematic Dutch).
services, Ph.D. thesis, Delft University Press, Delhi, the [29] H. Holler, Kommunikationssysteme: Normung und soziale
Netherlands, 1996. Akzeptanz, Vieweg, Braunschweig/Wiesbaden, 1993 (in
[I 11 Th. Bruins, Open systemen, PTT Research, the Netherlands, German).
1993 (in Dutch). [30] T.M. Egyedi, Grey fora of standardization: A comparison of
1121 C. de Jong, Telecommunicatie: Strategi&n voor openbare JTC 1 and Internet. KPN Research, Leidschendam, the
structuren, Collegedictaat, Delft University of Technology, Netherlands, 1994.
Faculty of Technology and Society, September 1991 (in [31] C. de Jong, Openbare netten: Inleiding, in: C. de Jong, E.F.
Dutch). Michiels, J.A.M. Nijhof. P. van der Vlist (Eds.). Handboek
[ 131 G. Dosi, Technological paradigms and technological trajecto- Telematica, Samson, Alphen aan den Rijn, 1990a, pp. l-13
ries, Research Policy 11 (1982) 147-162. (in Dutch).
[14] J.C. Arnbak, Many voices, one structure, Inaugural speech, [32] J.A.J. Kerkers, J.E.P.M. van der Aalst, Message handling
Delhi University of Technology, Delft, 1986. systems, in: C. de Jong, E.F. Michiels, J.A.M. Nijhof, P. van
[15] R. ManseB, R. Hawkins, Old roads and new signposts: Trade der Vlist @Is.), Handboek Telematica, Samson, Alphen aan
policy objectives in telecommunication standards, in: F. den Rijn, Samson, 1992, pp. 1-19 (in Dutch).
Klaver, P. Slaa @Is.), Telecommunication, New Signposts [33] B. Whitehouse, Communications International, August 1990,
to Old Roads, 10s Pmss, Amsterdam, 1992, pp. 45-54. p. 3.
[16] F.C. Schoute, Mixed traffic patterns and traffic capacity in [34] G.J. Terink, Telematicadiensten en apparatuur in Nederland,
ISDN, in: Arnbak (Ed.), ISDN in Europe, Innovative services Alphen aan den Rijn: Samson, 1988.
450 T.M. Egyedi/Computer StAndArdS & Interfaces 18 (1997) 431-450

Tineke Mirjam Egyedi was employed


from 1990 to 1994 by the Delft Univer-
sity of Technology as a Ph.D. researcher
on standardisation of telematic services.
As a consultant for KPN Research, she
compared JTC 1 and Internet standardis-
ation. Subsequently she was engaged by
the Royal Institute of Technology in
Stockholm to study the role of contain-
ers in the transportation infrastructure
from a standardisation perspective. She
received her Ph.D. degree in 1996. As
an emulovee of the Delft Universitv of
Technology she wrote two policy documents for the Dutch &in-
istry of Transport and Water Works. Currently, she is studying the
effect of different transmission protocols on the use of multimedia
by researchers cooperating from several locations. Her main inter-
est concerns standardisation and the development of infrastruc-
tures.

Anda mungkin juga menyukai