Positioning the Normalized Systems Theory in a Design Theory

Framework

Philip Huysmans, Gilles Oorts and Peter De Bruyn

Normalized Systems Institute

Department of Management Information Systems, University of Antwerp, Prinsstraat 13, 2000 Antwerp, Belgium

{philip.huysmans, gilles.oorts, peter.debruyn}@ua.ac.be

Keywords:

Normalized Systems, Design Science, Design Theory, Design Theory Anatomy.

Abstract:

Several frameworks have been proposed to deﬁne design science and design theory over the last decades.

For this reason, positioning a research stream within both paradigms has become a difﬁcult exercise. In this

paper, the Normalized System (NS) theory is positioned within design science and design theory, in particular

the design theory framework formulated by Gregor & Jones (2007). Normalized Systems theory has been

proposed as a way to cope with the ever increasingly agile environment to which organizations and their

software applications need to adapt. NS achieves this evolvability by articulating theorems that modular

structures need to comply with in order to be evolvable. The results of positioning NS within the presented

framework for design theories show that NS almost fully incorporates all components of the design theory

anatomy. An application of NS theory in other ﬁelds is also discussed, which conﬁrms the applicability of the

anatomy of Gregor & Jones (2007) within other disciplines. By positioning Normalized System theory within

design science and design theory, we also believe to contribute to the deﬁnition of both ﬁelds in this paper.

1 INTRODUCTION

During the last decades, the design science research

approach has increasingly become of interest in in-

formation systems (IS) research. One of the ﬁrst ac-

knowledgments of this evolution was articulated by

March & Smith (1995), who stated that design re-

search (aimed at developing IT artifacts that solve

relevant IS problems) could be argued to be more

important within IS research than traditional (natu-

ral) science (aimed at understanding IT phenomena).

As the relatively new IS research area took shape,

most researchers agreed with the importance of de-

sign science approach as an alternative to the be-

havioral research approach. On the content of IS

design science however, opinions still differ greatly

(Baskerville, 2008). For this reason, giving a clear

and agreed upon deﬁnition of design science is close

to impossible. However, some fundamental proper-

ties of the discipline seem to exist nevertheless.

The notion of design science was ﬁrst formulated

in 1969 by Simon, who stated that researchers can

achieve their research goals “to the degree that they

can discover a science of design, a body of intellec-

tually tough, analytic, partly formalizable, partly em-

pirical, teachable doctrine about the design process”

(Simon, 1996, p. 113). March & Smith (1995, p. 253)

articulate this idea more speciﬁc for IS research, stat-

ing that “design scientists produce and apply knowl-

edge of tasks or situations in order to create effective

artifacts.” The same authors state that ultimate pur-

pose of IS design science research is the formulation

of IT artifacts that provide utility in solving IS prob-

lems, as opposed to formulating and testing hypothe-

ses in a conceptual framework as in traditional natural

science.

Although most researchers agree on these princi-

ples, one of the discussion points in deﬁning IS design

science is whether or not theory should be included in

the design science paradigm. Based on the literature

review of Venable (2006), one can clearly detect the

polemic between proponents and adversaries regard-

ing the inclusion of theory in design science. Whereas

Hevner, March, Park & Ram (2004) are ambiguous

on the role of theory, March & Smith (1995) clearly

state that theory should be solely reserved for natu-

ral science and should not be part of design science

(Venable, 2006). Many other authors however dis-

agree and argue theory should in fact be an essential

part of design science (Nunamaker, Chen & Purdin,

1990; Walls, Widmeyer & El Saway, 1992; Venable

& Travis, 1999; Markus, Majchrzak & Gasser, 2002;

Huysmans P., Oorts G. and De Bruyn P.

Positioning the Normalized Systems Theory in a Design Theory Framework.

DOI: 10.5220/0004460900330042

In Proceedings of the Second International Symposium on Business Modeling and Software Design (BMSD 2012), pages 33-42

ISBN: 978-989-8565-26-6

Gregor, 2006; Gregor & Jones, 2007).

Recently Normalized Systems (NS) theory has

been proposed as an approach to design evolvable

modular structures in software in a deterministic and

proven way. While the purpose of this paper is not

to deﬁne design science or substantiate whether or

not design theory is a part of design, we will at-

tempt to position the state of art of the NS research

stream within the design science and design theory

paradigms. Indeed, as Normalized Systems theory

prescribes speciﬁc constraints for building IT artifacts

in a purposeful way, the research stream seems to

be at least closely related to design science and the-

ory. Positioning the stream into current design sci-

ence methodologies may thus provide us the oppor-

tunity to check the approach for its theoretical com-

pleteness, while possible arising gaps might identify

future research directions. This method is endorsed

by Walls, Widermeyer & El Sawy (2004), who state

that the notion of an Information System Design The-

ory can be used as guiding framework to structure de-

sign research.

This classiﬁcation is a challenging exercise as

both design science and design theory are not yet de-

cisively deﬁned. In this paper, we will argue that Nor-

malized Systems theory is both a design theory and

design science, as it is a proven and normative way of

building IT modules (design theory) and it formulates

IT artifacts (design science). The point of view on

design science and design theory that will be used in

this paper, is the distinction expressed by Walls et al.

(2004). These authors deﬁned design science differ-

ent than other authors such as March & Smith (1995)

and Hevner et al. (2004) by separating design science

from design practice. According to this distinction,

design practice is concerned with the creation of in-

stances of artifacts, whereas design science “should

create the theoretical foundations of design practice”

(i.e. design theories) (Walls et al., 2004, p.48). As

this paper will show, Normalized Systems is such a

design theory that creates the theoretical foundation

for designing evolvable systems.

Positioning Normalized Systems as a design the-

ory will also express the different nature of NS from

most of existing design science. According to March

& Smith (1995, p. 253) design science should be con-

cerned with creating effective artifacts by producing

and applying apply knowledge of tasks or situations.

These authors also specify the aforementioned should

be the primary goal of design science, “rather than

producing general theoretical knowledge” (March &

Smith, 1995, p. 253). Normalized Systems theory

however uses a theoretical foundation to prove that

using Normalized Design Theorems leads to Normal-

ized Systems theory that are stable with respect to

the anticipated changes. This alternative approach of

Normalized Systems to design science will become

clear in the next sections of this paper.

The remainder of this article is structured as fol-

lows: in the following section, we discuss the ori-

gin and structure of Normalized Systems theory. In

the third section, we will position Normalized system

theory within the design theory anatomy suggested by

Gregor & Jones (2007). In the discussion we will take

a look at the results of this evaluation. The anatomy

used in this paper will also be placed within its con-

text and the applications of the evaluation will be dis-

cussed.

2 NORMALIZED SYSTEMS

THEORY

In the current information-intensive and agile busi-

ness environment, organizations as well as their

supporting software applications need to cope with

changes in their structure and functionality. The

way in which an organization can assimilate these

changes, determines its evolvability. On top of these

changes, systems are also faced with an ever increas-

ing size and complexity of their structure and func-

tionality. To tackle these two challenges, modularity

has frequently been suggested to divide complex sys-

tem in easier to handle subsystems and cope with the

evolvability requirement by allowing the modules to

change independently (Baldwin & Clark, 2000). In

fact, modularity is a core concept from systems theory

and has been applied in manydifferent application do-

mains, including software engineering (Parnas, 1972)

were multiple modular primitives have been deﬁned

for representing data (e.g., structures), actions (e.g.,

procedures) or both (e.g., classes) as the core of infor-

mation systems. However, (hidden) coupling between

those modules tends to limit the anticipated evolv-

ability (De Bruyn & Mannaert, 2012). Thoroughly

based on concepts and reasoning from systems the-

ory and thermodynamics, Normalized Systems the-

ory (NS) was recently proposed to strictly guide engi-

neers in devising such evolvable modularity. While it

was originally applied for software architectures, its

relevance for other application domains (e.g., orga-

nizational design) has been demonstrated previously

(Van Nuffel, 2011; Huysmans, 2011).

First, with its primary aim of enabling evolvability

in software architectures, NS deﬁnes the occurrence

of so-called combinatorial effects: changes within the

modular structure of which the impact is dependent

on the size of the system they are applied to (Man-

Second International Symposium on Business Modeling and Software Design

naert, Verelst & Ven, 2011, 2012). Such combinato-

rial effects are clearly undesirable in the context of

evolvability. Indeed, as a system grows throughout

time, a combinatorial effect would imply that the ef-

fort required to implement a same kind of change be-

comes ever more complex as time goes by. Moreover,

the concept of stability from systems theory is highly

related to this reasoning and offers clues in how to

avoid combinatorial effects. In systems theory, stabil-

ity is regarded as an essential property of systems and

implies that a bounded input function should always

result in a bounded output function, even if an un-

limited time period T → ∞ is considered. Applied

to information systems, this means that a bounded

set of changes (selected from the so-called antici-

pated changes such as “add additional data attribute”

or “add additional action entity”) should result in a

bounded impact to the system, even for T → ∞ (i.e.,

an unlimited systems evolution is considered). Con-

sequently, stability reasoning conﬁrms that the impact

of changes should not be dependent on the size of the

system, but only related to the kind of change per-

formed (and hence, combinatorial effects should be

avoided). As such, normalized systems are deﬁned

as systems exhibiting stability and lacking any com-

binatorial effects towards a deﬁned set of anticipated

changes (Mannaert et al., 2011, 2012).

In order to obtain such normalized systems, NS

theory proposes four theorems which should be sys-

tematically adhered to (Mannaert et al., 2012):

• Separation of Concerns, stating that each change

driver (concern) should be separated from other

concerns;

• Action Version Transparency, stating that action

entities should be updateable without impacting

their calling action entities;

• Data Version Transparency, stating that data enti-

ties should be updateable without impacting their

calling action entities ;

• Separation of States, stating actions in workﬂow

should be separated by state (and called in a state-

full way).

For each of these theorems it has been formally

proven that any violation against them at any time

will result in a combinatorial effect (Mannaert et al.,

2012). Consequently, if the aim is to obtain a true

normalized system, these principles for building soft-

ware architectures should all be consistently applied.

In reality, the systematic isolation of all concerns pre-

scribed by the theory, result in a very ﬁne-grained

modular structure. While exhibiting a proven degree

of evolvability, the structure in itself may be regarded

as complex in the sense that the system becomes an

aggregation of many ﬁne-grained instantiations of the

modular primitives offered by the employed program-

ming language, many more than in commonly devel-

oped software applications.

Therefore, a set of elements were proposed to

make the realization of normalized systems more fea-

sible. These elements are each a structured aggre-

gation (i.e., encapsulation) of the available software

primitives and together providing the core functional-

ity of information systems, be it in a highly generic

and reusable way. As such, the internal structure

of these elements could be considered to be a set of

design patterns regarding higher-level modular build-

ing blocks, adhering to the above-describedtheorems.

These ﬁve elements are (Mannaert et al., 2011):

• action element, a structured composition of soft-

ware constructs to encapsulate an action construct

into an action element;

• data element, a structured composition of soft-

ware constructs to encapsulate a data construct

into a data element;

• workﬂow element, a structured composition of

software constructs to create an encapsulated

workﬂow element (representing a sequence of ac-

tion elements);

• trigger element, a structured composition of soft-

ware constructs to create an encapsulated trigger

element (controlling for and representing the acti-

vation of a workﬂow or action element)

• connector element, a structured composition of

software constructs to create an encapsulated con-

nector element (allowing the stateful connection

of data elements with external systems).

Each of these elements are described in more detail

in Mannaert et al. (2011) and illustrated to be able

to cope with a set of anticipated changes in a stable

way. Normalized systems are then typically imple-

mented by creating a set of n instantiations of the ﬁve

elements.

Additionally, recent research efforts have demon-

strated that reasoning based on the concept of entropy

from thermodynamics (1) supports and (2) further ex-

tends NS theory (Mannaert, De Bruyn & Verelst,

2012a). More speciﬁcally, the deﬁnition of entropy

as employed in statistical thermodynamics was used

for this purpose, i.e., the number of microstates con-

sistent with the same macrostate (Boltzmann, 1995).

Applied to information systems, microstates are op-

erationalized as binary values expressing the cor-

rect or erroneous processing of a programming lan-

guage construct, while the macrostate is associated

with loggings or database entries representing the cor-

rect or erroneous processing of the software system.

Positioning the Normalized Systems Theory in a Design Theory Framework

Hence, when it is uncertain which microstate conﬁg-

uration brought about the observed macrostate, en-

tropy (uncertainty) is present inthe system. The larger

the number of microstates consistent with the same

macrostate, the higher is the amount of entropy in the

system. In Mannaert et al. (2012a), it was further

illustrated that the aim of minimizing or controlling

entropy in modular system, highly coincides with the

four theorems explained above. Moreover, a set of of

two new theorems were suggested based on this en-

tropy reasoning (Mannaert et al., 2012a):

• Data Instance Traceability, requiring each ver-

sion and values of an instance of a data structure

to be tracked;

• Action Instance Traceability, requiring each ver-

sion and thread of an instance of a processing

structure to be tracked.

Finally, several real-life implementations of NS

applications have been successfully deployed up to

this moment. Some of them have been already brieﬂy

discussed in Mannaert et al. (2011).

3 CLASSIFICATION OF

NORMALIZED SYSTEMS

THEORY

In this section, Normalized Systems theory will be

classiﬁed within a design theory framework. The In-

formation System Design Theory (ISDT) framework

that will be used in this paper is the design theory

anatomy proposed by Gregor & Jones (2007). This

anatomy is an extension of the ISDT framework pro-

posed by Walls et al. (1992), whose sources date back

to the work of Dubin (1978) on theory of the natu-

ral science type and the work of Simon (1996) on

sciences of the artiﬁcial. In their work, Walls et al.

(1992) attempted to formulate a prescriptive theory

that articulates how a design process can effectively

be carried out. Looking back on the formulation of

their ISDT, (Walls et al., 2004) conclude that, al-

though used scarcely, it can be helpful as a guiding

framework that helps in structuring the “how to” of

the design progress based on a theoretical foundation.

As the work of Walls et al. (1992) was just an initial

attempt to deﬁne an InformationSystems Design The-

ory, several reﬂections and reactions have been pub-

lished on their work (Gregor & Jones, 2007; Walls

et al., 2004). For example Gregor & Jones (2007)

clarify that the goal of a design theory can be either a

methodology or a product, which was not yet clearly

deﬁned by Walls et al. (1992) (Gregor & Jones, 2007).

The remainder of this section will show that Normal-

ized System is an example of a design theory deﬁning

the design of a product (e.g. a system). Furthermore

Gregor & Jones (2007) argue that two structural com-

ponents of a theory formulated by Dubin (1978) are

lacking from the anatomy formulated by Walls et al.

(1992), namely “units” and “system states”. These

constructs are deﬁned as the building blocks of theory

and the range of system states that the theory covers

respectively (Gregor & Jones, 2007), constructs that

will be shown to be part of the Normalized System

paradigm in this paper.

Considering these missing constructs, Gregor &

Jones (2007) argue that an information system de-

sign theory consists of eight components. The ﬁrst six

components are called the Core Components, as they

are essential to determine how an artifact can/should

be constructed. The other two components can have

a positive inﬂuence on the credibility of the work, but

can be deﬁned later (Gregor & Jones, 2007). The re-

sults of the classiﬁcation of Normalized Systems De-

sign Theory by these components are shown in Table

1. In the next paragraphs, we will discuss the classi-

ﬁcation of the NS theory within these eight compo-

nents.

The ﬁrst component Gregor & Jones (2007) de-

ﬁne is the purpose and scope of a design theory. This

component states “what the system is for, or the set of

meta-requirements or goals that speciﬁes the type of

system to which the theory applies” (Gregor & Jones,

2007, p. 325). According to the same authors, the en-

vironment in which the artifact should operate is an

important factor to consider. To understand the pur-

pose and goal of Normalized Systems, it is indeed

important to keep in mind the agile environment in

which modern systems should operate. As mentioned

earlier, the purpose of Normalized Systems theory

is the elimination of combinatorial effects (towards

a set of anticipated changes), as a means to achieve

evolvability of information systems. Combinatorial

effects however not only appear in information sys-

tems, but can be observed in a very broad spectrum

of domains. Therefore Normalized Systems Design

Theory is not limited to information systems and, as

will be discussed in Section 4.3 of this paper, can ap-

ply to many different natural and artiﬁcial phenom-

ena. This is in agreement to Gregor & Jones (2007)

who state that the applicability of their design theory

anatomy is possibly wider than the IS discipline, as it

is in itself based on sources from other disciplines.

Constructs are “the entities of interest in the the-

ory”, and “are at the most basic level in any theory”

(Gregor & Jones, 2007, p. 325). Other authors, such

as March & Smith (1995, p. 256), deﬁne constructs

Second International Symposium on Business Modeling and Software Design

Table 1: Classiﬁcation of Normalized Systems within the anatomy of Gregor & Jones (2007) and the components of Walls

et al. (1992).

Gregor & Jones (2007) Walls et al. (1992) Normalized Systems theory

Core Components

1. Purpose and scope Meta-requirements Evolvable software architectures by

elimination of combinatorial effects

2. Constructs Combinatorial effects, modularity, ac-

tion/data

3. Principles of form and

function

Meta-description Five Normalized Elements

Four Normalized Design Theorems

4. Artifact mutability Anticipated changes

5. Testable propositions Product hypothesis When theorems are applied, no combi-

natorial effects occur with regard to the

anticipated changes

Process hypothesis

6. Justiﬁcatory knowledge Product kernel theories Systems theory (cf. stability)

Process kernel theories Thermodynamics (cf. entropy)

Additional Components

7. Principles of implementa-

tions

Design method Supporting applications (e.g., “Prime

radiant”)

8. Expository instantiation Real-life NS implementations (by

means of instantiations of the ele-

ments)

as “the vocabulary of a domain [...] used to describe

problems within the domain and to specify their solu-

tions”. Although the combination of these deﬁnitions

gives a clear understanding of a construct, the authors

believe there is still a certain subjectivity in determin-

ing what “vocabulary”or shared knowledge should be

considered a construct and what should not. Within

the Normalized System theory, the authors recognize

three constructs. The ﬁrst and second construct are

generally accepted and known within IT, namely the

concept of modularity and the basic building blocks

of an information systems: data and actions. The

third construct of Normalized Systems theory is com-

binatorial effects which the NS theory is determined

to eliminate, as formulated by Mannaert & Verelst

(2009). These three concepts constitute the main con-

structs of NSDT, although due to the subjectivity of

the deﬁnition of this component it does seems possi-

ble to argue there are other constructs that were not

mentioned.

To avoid combinatorial effects, the NS theory

states that applications should be structured using the

ﬁve elements deﬁned by Mannaert & Verelst (2009),

which are based on the body of thought of the Nor-

malized Design Theorems. As the principles of form

and function are deﬁned as “the principles that deﬁne

the structure, organization and functioning of the de-

sign product” by (Gregor & Jones, 2007, p. 325), it is

clear the Normalized Elements and Normalized De-

sign Theorems make up this component of the design

anatomy. The Normalized design Theorems and El-

ements specify both structural and functional proper-

ties of artifacts by providing guidelines and patterns

for constructing instances of the artifacts. Whereas

the theorems articulates the general principles that

need to be applied, the elements are in fact a de-

sign pattern as they specify a possible way of com-

ply with the Normalized Theorems. For the imposed

internal structure of the ﬁve types of Normalized El-

ements make applications free of combinatorial ef-

fects, the aggregation of these instances of elements

subsequently makes up a Normalized System. The

Normalized Design Theorems on the other hand can

also be considered as a principle of form and function,

as they are the guiding principles for constructing the

Normalized Elements.

The next component has to do with a special na-

ture of an IS artifact that Gregor & Jones (2007,

p. 326) recognize, stating that IS artifacts are very

mutable and constantly evolving. They also believe

that:

“the way in which they [i.e. IT artifacts]

emerge and evolve over time, and how they

become interdependent with socio-economic

contexts and practices, are key unresolved is-

sues for our ﬁeld and ones that will become

even more problematic in these dynamic and

Positioning the Normalized Systems Theory in a Design Theory Framework

innovative times”.

Evolvabilty and agility are in fact exactly the goal

and purpose of Normalized Systems theory, since

Normalized Systems are highly evolvable and stable

systems based on structured elements that minimize

combinatorial effects. For this reason the component

of artifact mutability, which is formally deﬁned as

“the changes in state of the artifact anticipated in the

theory” (Gregor & Jones, 2007, p. 322) can be cleared

recognized as the anticipated changes of NS.

The next component Gregor & Jones (2007) de-

ﬁne, is testable propositions. These are claims or pre-

dictions about the quality of a system or tool when the

design theory is applied. As such, the testable propo-

sition of NS theory can be formulated as the elim-

ination of combinatorial effects when the principles

of form and function are pursued consistently. Al-

though this component seems clearly deﬁned within

NS theory, the deﬁnition of the component also re-

quires the propositions to be testable. According to

Walls et al. (1992), the assertion that applying a set of

design principles will result in an artifact that achieves

its goal can be veriﬁed by building (an instance of)

the artifact and testing it. Applying this veriﬁcation

method on Normalized Systems, the proposition of

NS theory (the elimination of combinatorial effects)

can be veriﬁed/tested by building a Normalized sys-

tem according to the principles of form and function

and proving that the system is exempt of combinato-

rial effects.

Walls et al. (1992) formulated the idea that kernel

theories should be part of an Information System De-

sign Theory (ISDT). According to these authors, ker-

nel theories govern both the design requirements and

the design process. As Walls et al. (1992) believe,

these two aspects should be separated and therefore

deﬁned both product kernel theories and process ker-

nel theories. Walls et al. (2004) elucidate the impor-

tance of kernel theories for design science, by stat-

ing that the design science process uses these theo-

ries and combines them with existing artifacts to for-

mulate new design theories. Gregor & Jones (2007,

p. 327) however argue that thetwo types of kernel the-

ories (i.e. process and product kernel theoris) are “a

linking mechanism for a number, or all, of the other

aspects of the design theory” and should be consid-

ered as one component, the justiﬁcatory knowledge

that explains why a design works. This symbiosis is

substantiated by the argument that the design process

and design product are mostly founded by a single

kernel theory (e.g. the justiﬁcatory knowledge). They

deﬁne this component as “the underlying knowledge

or theory from the natural or social or design sci-

ences that gives a basis and explanation for the de-

sign” (Gregor & Jones, 2007, p. 322). According to

this deﬁnition, the underlying justiﬁcation for Nor-

malized Systems theory is twofold. First the central

idea of Normalized Systems theory is systems sta-

bility, as formulated in the systems stability theory

which states that a bounded input should always result

in a bounded output (BIBO-principle). In Normalized

Systems theory, this is interpreted as the transforma-

tion of functional requirements into software primi-

tives (Mannaert et al., 2011). Secondly, Normalized

Systems theory shows compatibility with the concept

of entropy, as has been discussed in Section 2. Initial

research efforts largely validate the use of Normal-

ized System principles when studying the NS theory

from the point of view of entropy theory (Mannaert,

De Bruyn & Verelst, 2012b).

The next component of a design theory is its prin-

ciples of implementation. Gregor & Jones (2007) con-

sider this and the next components as additional com-

ponents that are not a no essential part of a design

theory but should be formulated if the credibility of

the theory is to be enhanced. Concerning this compo-

nent, we can refer to the model taxonomy of Winter,

Gericke & Bucher (2009). According to this taxon-

omy, Normalized Systems theory should be classiﬁed

rather as a prescriptive model with result recommen-

dation (“a model”) than a model with activity recom-

mendation (“a method”), referring to the clear prin-

ciples of form and function of NS mentioned earlier.

Although the emphasis within NS theory is on the “re-

sult view” rather than the “activity view”, Winter et al.

(2009) argue that both are views on the same “prob-

lem solving artifact”. This similarity of methods and

models is also apparent in the classiﬁcation used in

this paper, in the form of the similarities between the

principles of form and function and the principles of

implementation. The form or architecture of an arti-

fact can in itself be used as a underlying principle and

target on which a method and guidelines for construc-

tion of the artifact are based. As opposed to clearly

deﬁned principles of form and function, such a for-

mal methodology in the form of a procedure that ex-

plicitly articulates the steps that need to be followed

to construct normalized elements, does not exist. The

formulation of the Normalized elements simply hap-

pens while keeping the theorems at the back of one’s

mind, and is helped by some supporting applications

(e.g. “Prime radient”). These applications are more

than a tool, as they provide guidelinesfor constructing

Normalized elements. For this reason, they could be

considered the principles of implementation of Nor-

malized System theory.

The ﬁnal component, expository instantiation, has

two functions: it shows the applicability of a design

Second International Symposium on Business Modeling and Software Design

theory and it can be used as a way to explain a design

(Gregor & Jones, 2007). Instantiated artifacts are the

embodiment of this component. Normalized Systems

theory has been used in the development of several

several software applications. According to the def-

inition of Gregor & Jones (2007), the expository in-

stantiation of NS are these applications and the Nor-

malized Elements they consist of. The seven appli-

cations that have been implemented according to the

NS principles range from a system for distribution of

multimedia to a system responsible for the manage-

ment of power units on high-speed railroads. Man-

naert et al. (2012) describe these implementations in

greater detail.

4 DISCUSSION

4.1 Reviewing NS According to the

Anatomy

By positioning NS theory within the design theory

anatomy of Gregor & Jones (2007), it becomes clear

that NS theory incorporates close to all components

of the anatomy. One could argue that there still is

one signiﬁcant gap between the anatomy and NS the-

ory: an explicit method to guide the construction of

Normalized elements (part of the principles of imple-

mentation). However, overall the NS theory shows

great similarities with the design anatomy studied in

this paper. In conclusion to this strong similarity, we

could argue the term Normalized System Design The-

ory is appropriate to describe the presented theory.

4.2 Design Science vs Design Theory

Although there was opted to compare Normalized

System theory to the design theory anatomy of Gregor

& Jones (2007) in this paper, the authors acknowl-

edge several other frameworks exist that attempt to

deﬁne design science and design theory. To put the

used anatomy into context, an overview of the differ-

ent deﬁnitions and point of views on design science

and design theory will be discussed in this section.

Over the last decades, several deﬁnitions and

frameworks have been published that tried to formal-

ize and structure IS design science. These publica-

tions however show some notable differences in be-

liefs and deﬁnition of IS design science and theory.

The work of March & Smith (1995) mainly focuses

on the processes and research outputs of IS design sci-

ence, and emphasizes that theory should not be part

of design science. In their opinion, theories are re-

served for natural sciences, and design science should

“strive to create models, methods and implementa-

tions” (March & Smith, 1995, p. 254) On the other

hand, authors such as Nunamaker et al. (1990), Walls

et al. (1992), Gregor (2006) tried to deﬁne this very

notion of an IS design theory. It should be mentioned

that these authors distinguish a design theory form a

theory in natural sciences. According to their deﬁ-

nition, a design theory is as a prescriptive theory that

“says howa design process can be carried out in a way

which is both effective and feasible”, in contrast to the

“explanatory and predictive theories found in natural

or physical sciences” (Walls et al., 1992, p. 37).

Another point of disagreement seems to be the

outcome of design science. Some academics (March

& Smith, 1995; Hevner et al., 2004; Baskerville,

2008) believe the core of design science is “directed

toward understanding and improving the search

among potential components in order to construct

an artifact that is intended to solve a problem”

(Baskerville, 2008, p.441). According to this point

of view, design science should therefore be limited to

deﬁning artifacts. The work of Walls et al. (1992),

Walls et al. (2004) and Gregor & Jones (2007) how-

ever suggests design science should also be concerned

with building design knowledge in the form of de-

sign theories, something that is simply not mentioned

by the aforementioned authors. Walls et al. (2004)

clearly indicate this difference by distinguishing be-

tween design science and design practice. Whereas

design practice actually creates instances of artifacts,

design science “should create the theoretical founda-

tions of design practice” (i.e. design theories) (Walls

et al., 2004, p.48).

Previous differences clearly show that there is an

agreement that design science can not be equated with

design theory (Baskerville, 2008). To the authors

knowledge and feeling, there are however no publi-

cations that explicitly and decisively specify the re-

lationship between design theory and design science,

nor has there been a discussion between the propo-

nents of both points of view. Therefore, other than

claiming NS is both design science and design the-

ory, precisely positioning Normalized Systems the-

ory within design science is simply impossible at this

point. The nature of Normalized Systems theory also

makes it very difﬁcult to relate NS to the generally

accepted conception of IS design science formulated

by Hevner et al. (2004). As discussed earlier, Nor-

malized Systems theory is based on proven theorems

that deterministically deﬁne that combinatorial effects

will be eliminated when the four theorems are system-

atically applied to the construction of elements. Ac-

cording to Hevner et al. (2004) however, design sci-

ence is typically concerned with designing artifacts

Positioning the Normalized Systems Theory in a Design Theory Framework

that are evaluated and improved upon and which out-

come is a ﬁnal design artifact that performs better than

other artifacts. Although it is clear that Normalized

systems theory is in contrast with this design process,

NS in its essence still is design science as it is “con-

cerned with devising artifacts to attain goals” (March

& Smith, 1995, p. 253). In the authors opinion, this

also shows that the current conception of design sci-

ence is too stringent to allow design theories such as

Normalized Systems theory to be positioned within

design science. Even the notion of “theory-based de-

sign science” (Baskerville, 2008) does not cover the

basic principles of Normalized Systems theory, as this

type of design science should be concerned with the-

ory testing rather than formulating a theory (which

Normalized Systems does).

4.3 Applications of NS

The positioning of NS in the presented framework is

an important basis for further research in the con-

text of the NS theory. Within the design science

methodology, the framework allows the positioning

of at least two research directions. First, existing gaps

between the current research results and the frame-

work components can be considered, as discussed in

Section 4.1. Since NS was not developed by follow-

ing a speciﬁc design science methodology, it is pos-

sible that certain components are missing or insuf-

ﬁciently described. Consequently, this research di-

rection would focus on the original domain of NS,

i.e., software. Second, different domains can be re-

searched using the NS theory. As a motivation for

this research approach, consider the use of the sys-

tems theoretic concept of stability and the thermody-

namic concept of entropy in the justiﬁcatory knowl-

edge component in Table 1. The interpretation of

such fundamental concepts indicates that the NS the-

ory could possibly be applied to other research ﬁelds

as well: in itself, these concepts do not originate from

the software domain. As discussed, the application

of successful solutions from related research ﬁelds is

an important goal of the design science methodology.

As an example of such a research project, we men-

tion how the NS theory has already been applied to

the research ﬁeld of Business Process Management

(Van Nuffel, 2011). In a research project in this re-

search direction, one ﬁrst has to validate whether the

purpose and scope of the NS theory can be applied to

that research ﬁeld as well. Consequently, it is crucial

to consider the research ﬁeld as a modular structure,

where combinatorial effects between the modules are

a relevant issue. Put differently, one has to apply the

constructs deﬁned by the NS theory in the speciﬁc re-

search ﬁeld. For the Business Process Management

ﬁeld, Van Nuffel (2011, p. 89) explicitly mentions: “a

business process essentially denotes a modular struc-

ture, of which the building blocks should be loosely

coupled in order to avoid the described combinatorial

business process effects”. Next, the framework pre-

scribes that principles of form and function should be

described. In the software domain, NS describes four

theorems, which provide the guidelines for the design

of ﬁve software elements. In the Business Process

Management domain, the four principles have been

applied to the design of business processes. This has

resulted in the formulation of 25 guidelines which de-

scribe necessary modular structures to avoid combi-

natorial business process effects. An example of such

a guideline is the “Notifying Stakeholders” guideline

(Van Nuffel, 2011, p. 143): “Because notifying, or

communicating a message to, stakeholders constitutes

an often recurring functionality in business processes,

a designated business process will perform the re-

quired notiﬁcation”. Indeed, omitting to separate this

process would result in a combinatorial effect when

changes to the notiﬁcation process need to be applied.

Similar to NS on the software level, this insight is not

necessarily new: other authors, such as Erl (2008),

Papazoglou & Van Den Heuvel (2006) and Cohen

(2007) provide similar guidelines. However, these au-

thors do not provide a theoretical framework to moti-

vate the formulation of such guidelines. In order to

mature the ﬁeld towards a design science, the authors

of this paper believe that such a theoretical framework

is vital. Consequently, the proposed framework is im-

portant as a methodological guide for future research.

Moreover, the framework allows not only to position

current research results, but also to identify missing

elements. It can be noted that these guidelines can be

positioned between the four NS principles and the NS

elements: they are an application of the NS principles,

but are not sufﬁcient to completely specify process el-

ements. Therefore, additional research is required to

further the insight on these guidelines, and arrive at

such elements. Consistent with the approach provided

in the framework in Table 1, anticipated changes will

need to be formulated (as prescribed in the artifact

mutability component), and the absence of combina-

torial effects with regard to these anticipated changes

needs to be demonstrated (as prescribed in the testable

propositions component).

5 CONCLUSIONS

In this paper, we made an attempt to position Nor-

malized System theory within design science and de-

Second International Symposium on Business Modeling and Software Design

sign theory frameworks. Although we argued that

NS theory in its essence is design science, we also

showed that it does not completely ﬁt in existing de-

sign science frameworks. Positioning Normalized

Systems within the design theory anatomy of Gregor

& Jones (2007) however showed Normalized Systems

strongly resembles a design theory. The similarities

are to the extent that NS can be formulated as the Nor-

malized Systems Design Theory.

Positioning NS in the anatomy of Gregor & Jones

(2007) has shown to present several contributions in

this paper.

First, it endorses the applicability of the frame-

work by showing that a theory such Normalized Sys-

tems can indeed be soundly positioned within the

framework. Furthermore the positioning ratiﬁes the

premise of Gregor & Jones (2007) who stated that

the possibility of applying the anatomy in other dis-

ciplines could be a question for further research. In-

deed, this paper showed that it is initially possibly to

position the application of NS within Business Pro-

cess Management within the anatomy. This proves

that the design theory anatomy of Gregor & Jones

(2007) can indeed be applied in other disciplines than

IS design science.

A second contribution is that positioning NS

within the anatomy gave the opportunity to check NS

for its theoretical completeness, which is consistent

with the view of Walls et al. (2004) who state that

an ISDT framework can be used as guiding frame-

work to structure design research. The gaps that were

identiﬁed according to the application of NS within

Business Process Management also indicated future

research directions.

Finally the authors believe that this papers con-

tributes to the discussion on design science and de-

sign theory. It has been shown that Normalized Sys-

tems theory is an apparent example of a design theory.

As Normalized System theory should however also

be considered design science, the authors believe to

have contributed to both the deﬁnition of design sci-

ence and the elucidation of the relationship between

design science and design theory.

ACKNOWLEDGEMENTS

P.D.B. is supported by a Research Grant of the

Agency for Innovation by Science and Technology in

Flanders (IWT).

REFERENCES

Baldwin, C. Y. & Clark, K. B. (2000). Design Rules: The

Power of Modularity. Cambridge, MA, USA: MIT Press.

Baskerville, R. (2008). What design science is not. Euro-

pean Journal of Information Systems, 17, 441–443.

Boltzmann, L. (1995). Lectures on Gas Theory. Dover Pub-

lications.

Cohen, S. (2007). Ontology and taxonomy of services in a

service-oriented architecture. The Architecture Journal,

11, 10.

De Bruyn, P. & Mannaert, H. (2012). Towards applying

normalized systems concepts to modularity and the sys-

tems engineering process. In Proceedings of the Seventh

International Conference on Systems (ICONS 2012).

Dubin, R. (1978). Theory building, revised edition. London:

Free Press.

Erl, T. (2008). SOA: Principles of Service Design. Pren-

tice Hall Service-Oriented Computer Series. Upper Sad-

dle River, NJ: Prentice Hall.

Gregor, S. (2006). The nature of theory in information sys-

tems. MIS Quarterly, 30(3), 611 – 642.

Gregor, S. & Jones, D. (2007). The anatomy of a design the-

ory. Journal of the Association for Information systems,

8(5), 312–335.

Hevner, A., March, S., Park, J., & Ram, S. (2004). Design

science in information systems research. MIS Quarterly,

28(1), 75–105.

Huysmans, P. (2011). On the Feasibility of Normalized En-

terprises: Applying Normalized Systems Theory to the

High-Level Design of Enterprises. PhD thesis, Univer-

sity of Antwerp.

Mannaert, H., De Bruyn, P., & Verelst, J. (2012a). Explor-

ing entropy in software systems: Towards a precise def-

inition and design rules. In Proceedings of the Seventh

International Conference on Systems (ICONS).

Mannaert, H., De Bruyn, P., & Verelst, J. (2012b). Explor-

ing entropy in software systems: Towards a precise deﬁ-

nition and design rules. In Proceedings of ICONS 2012:

The Seventh International Conference on Systems, (pp.

93 – 99). IARIA.

Mannaert, H. & Verelst, J. (2009). Normalized Systems:

Re-creating Information Technology Based on Laws for

Software Evolvability. Koppa.

Mannaert, H., Verelst, J., & Ven, K. (2011). The transfor-

mation of requirements into software primitives: Study-

ing evolvability based on systems theoretic stability. Sci-

ence of Computer Programming, 76(12), 1210–1222.

Special Issue on Software Evolution, Adaptability and

Variability.

Mannaert, H., Verelst, J., & Ven, K. (2012). Towards evolv-

able software architectures based on systems theoretic

stability. Software: Practice and Experience, 42(1), 89–

116.

March, S. T. & Smith, G. F. (1995). Design and natural

science research on information technology. Decision

Support Systems, 15(4), 251 – 266.

Positioning the Normalized Systems Theory in a Design Theory Framework

Markus, M. L., Majchrzak, A., & Gasser, L. (2002). A de-

sign theory for systems that support emergent knowledge

processes. MIS Quarterly, 26(3), 179 – 212.

Nunamaker, Jr., J. F., Chen, M., & Purdin, T. D. M. (1990).

Systems development in information systems research.

J. Manage. Inf. Syst., 7(3), 89–106.

Papazoglou, M. P. & Van Den Heuvel, W.-J. (2006).

Service-oriented design and development methodology.

International Journal of Web Engineering and Technol-

ogy, 2(4), 412–442.

Parnas, D. L. (1972). On the criteria to be used in decom-

posing systems into modules. Communications of the

ACM, 15(12), 1053–1058.

Simon, H. (1996). The Sciences of the Artiﬁcial (Third edi-

tion ed.). Cambridge, MA: MIT Press.

Van Nuffel, D. (2011). Towards Designing Modular and

Evolvable Business Processes. PhD thesis, University of

Antwerp.

Venable, J. (2006). The role of theory and theorizing in

design science research. In DESRIST Proceedings.

Venable, J. & Travis, J. (1999). Using a group support

system for the distributed application of soft systems

methodology. In Proceedings of the 10th Australasian

Conference on Information Systems, Wellington, New

Zealand, (pp. pp 1105–1117).

Walls, J., Widmeyer, G., & El Saway, O. (1992). Build-

ing an information system design theory for vigilant eis.

information systems research. Information Systems Re-

search, 3(1), 36–59.

Walls, J. G., Widermeyer, G. R., & El Sawy, O. A. (2004).

Assessing information system design theory in perspec-

tive: How useful was our 1992 initial rendition? Jour-

nal of Information Technology Theory and Application

(JITTA), 6(2), 43–58.

Winter, R., Gericke, A., & Bucher, T. (2009). Method ver-

sus model - two sides of the same coin? volume 34 of

Lecture Notes in Business Information Processing, (pp.

1–15). 5th International Workshop on Cooperation and

Interoperability, Architecture and Ontology, Amsterdam,

NETHERLANDS.

Second International Symposium on Business Modeling and Software Design