Content uploaded by Edzo Botjes
Author content
All content in this area was uploaded by Edzo Botjes on Sep 02, 2021
Content may be subject to copyright.
©2021 IEEE 23rd Conference on Business Informatics (CBI)
2378-1971/21/$31.00 ©2021 IEEE
preprint of DOI 10.1109/CBI52690.2021.00017 1
Attributes relevant to antifragile organizations
Edzo Botjes∗, Martin van den Berg†, Bas van Gils‡,§and Hans Mulder§ ∗Xebia, Hilversum, The Netherlands
†Hogeschool Utrecht, Utrecht, The Netherlands ‡Strategy Alliance, Amersfoort, The Netherlands §Antwerp
Management School, Antwerpen, Belgium
Abstract—Organizations operate in a socio-economic context,
and alignment with this context is key for business success. The
rate of change and impact of these changes on the operating
model of the organization appears to be increasing. Major trends
are the aftermath of the financial crisis of 2008, the “VUCA”
aspects (volatility, uncertainty, complexity, ambiguity), and the
Covid-pandemic. The challenge for organizations is to become
resilient or even antifragile to survive (unexpected) external
stressors.
Antifragility refers to a class of systems that do not deteriorate
(fragile) or withstand (robust) stressors, but actually improve as
a result of stressors. Our objective is to find attributes that are
relevant for an organization to be(come) antifragile. The list of
attributes found is based on an extensive survey of available
literature, and is validated with domain experts and leaders of
various organizations. We dub the list of attributes found the
Extended Antifragile Attribute List (EAAL).
Considering the current economic and social impact on organi-
zations and people of the Covid-pandemic, the EAAL is relevant
as well timely. The EAAL turns out to be extensive and holistic.
We consider the EAAL to be a stepping stone in setting the scene
of the domain of antifragility.
We suspect that the EAAL might also be applied to generic
system design including technology infrastructure and software
systems. This exploration is part of future research.
I. INTRODUCTION
The goal of organizations, operating in a socio-economic
context, is to remain significant for its stakeholders. Stakehold-
ers are owners, employees and consumers [1]. In order to do
so, the organization has to stay aligned with its environment.
As such, there is a constant ‘dance’ in which the organization
adapts to changing conditions in the market place, and at the
same time also influences this environment - all in the name
of avoiding strategic drift (see e.g. [2]).
It is often said that the rate at which organizations have
to change increases rapidly and that therefore we live in un-
certain times [3] characterized by high volatility, uncertainty,
complexity, and ambiguity (VUCA ) [3]–[6]. For example,
Anthony et al. state that “We’re entering a period of heightened
volatility for leading companies across a range of industries,
with the next ten years shaping up to be the most potentially
turbulent in modern history” [7]. The effects of the present
Covid-pandemic emphasizes this point further. We use the
term stressor to denote an event in the environment of the
organization that increases the need to adapt/ re-align in
organizations. Typical stressors that organizations are facing
are the Covid-pandemic, ever changing customer demands,
and the effects of digital transformation (e.g. [8]–[10]).
In order to stay relevant/ stay aligned with its environment,
it was long thought that organizations should strive to be
robust with respect to external stressors. A more recent insight
is that robustness is the middle ground between fragility
(the organization deteriorates as a result of stressors) and
antifragility (the organization improves as a result of stressors)
[11], [12]. Therefore, a more ambitious goal is to become
antifragile.
This brings us to the articulation of the objective for orga-
nizations that we hope to tackle in this paper: organizations
have to find a way to stay relevant in the current VUCA world
and be able to survive disruptive black swan stressors events.
Our hypothesis is that this can be achieved if the organization
has antifragile properties. This hypothesis is supported by lit-
erature from the field of risk management, complexity theory/
complex adaptive systems as well as enterprise governance
(e.g. [3], [9], [12]–[16]).
This leads to the following research question: What at-
tributes make an organization antifragile?
The main contribution of this study is that we provide
insight in the characteristics of antifragile organizations and
place relevant attributes in context. For scientists, the list
of attributes extends theories of organizational design. For
practitioners, the list of attributes provides insights which are
useful in designing organizations or parts of it.
To answer the research question, we will first frame our
research in more detail by clarifying key terms and our way
of looking at organizations (Section II). In Section III, we then
present our research methodology which follows the lines of
the snowball [17] approach for the literature review. In the
remainder of the article, we present our results (Section IV)
and conclusions (Section V), including a critical discussion of
the results and future research.
The main result of our research is the Extended Antifragile
Attribute List (EAAL), which is validated by domain experts
and practitioners. The EAAL is a synthesis of the attributes
found in the selected literature, and is constructed from the
concepts found in that same literature.
II. FRAMING OUR RESEARCH
In this section, we will clarify how we frame our research.
We start by clarifying the motivation for the term organiza-
tion (compared to similar terms such as enterprise that are
sometimes used). We will also motivate why we consider
organizations through the lens of systems theory, similar to
the work by Morgan [18], while recognizing that there are
different schools of thought in this area (see e.g. [19]–[21]).
A. Organization
The first concept that we will consider is organization.
Myriad definitions exist – some colloquial, others more formal
2
(e.g. from the realm of organization theory). In our view, there
are two key aspects that should be considered:
•The first interpretation of the term organization is: that
which results from the act of organizing. To clarify:
suppose we have a collection of Lego bricks. These can
be organized in different ways: by randomly tossing them
on the table top or, for example, by neatly sorting them
by size and color. The fact that our subjective assessment
may be that the former process leads to something that
is rather ‘disorganized’ is besides the point here. The
same is true for the way people ‘get organized’: there
are different methods for organizing people, processes,
data, and other resources to achieve a certain objective.
•The second interpretation of the term organization refers
to organizations(plural), recognizing them as a legal
entity (see e.g. the introductory chapters of [16]). In the
Dutch language, the category of Person (Dutch: Persoon)
is specialized in two subcategories: Natural person (Natu-
urlijk persoon) and Legal person (Rechtspersoon ). This
further emphasizes the fact that organizations-as-legal-
entities can be studied as independent entities.
Note that organizations in this second interpretation are the
result of specific acts of organizing, i.e. the first interpretation.
In [16], the leading term is enterprise, which is defined as
“Entities of purposeful human endeavor”. It is also stated that:
In the case of enterprises, the sensible opposite of
doing nothing and the inevitable development of
disorder is organizing: the harmonious ordering and
arrangement of activities and means in view of the
enterprise’ purpose(s). (ibid)
We concur with this view. Yet, making a formal distinction be-
tween enterprise and organization makes this paper needlessly
complex. We will only use the term Organization to signify
organizations as a legal entity which is ‘organized’ to achieve
goals of its associated stakeholders. Our point of view is that
organizations do not ‘have’ goals but stakeholders involved
in the organization do. More formally, when it is stated that
the ‘organization xhas goal y’ then our interpretation is:
stakeholders with regard to organization xagree (to some
extent, at least) on the goal y.
B. System
The term system is notoriously difficult to define. A full
review of the literature on systems theory is beyond the scope
of this article. We base our discussion here on [19]–[23]. In
order to limit the scope of our discussion here further, i.e. to
avoid giving a full overview of systems theory, we pose that
we are mainly interested in using systems theory as a lens on/
to study organizations – as defined in the previous section.
As such, we are not claiming that organizations are a system,
merely that considering them through this lens provides useful
insights in light of our research goal: when looking through
this lense we can identify (systemic) properties of organiza-
tions as systems, and study whether these properties contribute
to the fragility/ robustness/ antifragility of the organization.
In Ashby’s view, a system is considered to have a clear
boundary, and transforms inputs to outputs [22]. One of the
key results of Ashby’s work is the idea that systems are
considered as a black box and correlate inputs with outputs
without understanding the inner construction/ functioning of
the system that is studied. This perspective is too limited for
our purposes.
In the FRISCO framework [23], a system is defined as
A system is a special model, whereby all the things
contained in that model (all the system components)
are transitively coherent, i.e. all of them are directly
or indirectly related to each other from a coherent
whole. A system is conceived as having assigned to
it, as a whole, a specific characterization (the so-
called “systemic properties”).
whereas a model is defined to be “A model is a purposely ab-
stracted, clear, precise, and unambiguous conception.” In other
words, a system is considered to be a mental construct and it
is at least suggested that a system consists of interconnected
parts. This fits well with our notion of organizing parts into a
whole.
The authors of the FRISCO framework remark that (ibid)
“The decision where to draw the boundary of the system
depends on the system viewer”, which further reinforces
the subjective nature of what constitutes a system. We also
subscribe to this view. Particularly when systems become
more complex, it is harder to precisely and objectively de-
termine its boundaries when more than one stakeholder is
involved. In this light, it is useful to consider the work by
Boulding (See a discussion about [24] in [20]) describing a
hierarchy for considering systems on different levels: (1) static
structures and frameworks, (2) clockworks, (3) closed-loop
control mechanisms, (4) open systems with structural self-
maintenance, (5) lower organisms with functional parts and
blueprint growth, (6) animals with a brain to guide behavior,
capable of learning, (7) people with self-consciousness, (8)
socio-cultural systems with roles, communication, and the
transmission of values, and (9) transcendental systems, the
home of the ‘inescapable unknowables’.
From this, we learn that organizations have a high level of
complexity; they are high-up in the hierarchy. This aligns with
the study by Morgan [18] who presents different images on
organization as systems (e.g. the organization as a clockwork/
mechanism, as an organism, etc.).
C. Complexity
One aspect that underlies the above mentioned hierarchy of
Boulding is the fact that higher levels are more complex than
lower levels. This begs the question: what is meant by this
complexity? In our view, the Cynefin framework by Snowden
provides a useful perspective [25]–[28]. This framework is a
sense-making framework which means that it offers guidance
on how to respond in certain situations based on the subjective
assessment of the nature of a specific situation by an actor/
decision maker. Cynefin distinguishes the following modes:
•In the simple domain, causal effects between variables
are apparent, so the proper response is to recognize the
situation and act according to the best practice at hand.
3
•In the complicated domain, causal effects are knowable:
it may take a lot of time and effort but through analysis
they can be found. The proper response is to analyze the
situation and then act according to the findings of the
analysis.
•In the complex domain, causal effects are too complex
for analysis and can only be determined in hindsight.
The response mode is: hypothesize what might work, act
according to the hypothesis and study the effect of the
intervention. If the effect is desirable, the hypothesis was
correct. If not, then the effects should be dampened by
corrective action.
•The chaotic domain is an unordered domain with no con-
straints, and apparently random/ unpredictable, threaten-
ing behavior. It is often seen as the area where immediate
action is required in order to return from utter chaos
(and potentially threatening situations) to one of the other
domains.
•The disorder domain is the ‘catchall’ domain, represent-
ing situations where an actor/decision maker has not yet
come to a conclusion in which of the ‘other’ domains
s/he is.
Here, we are not making any claims about the level of com-
plexity of organizations - even though the Boulding hierarchy
suggests that organizations as defined in this article are likely
to be in the complex domain. In our view, the level of
complexity depends on how much is known about a given
domain and the perspective may be perceived differently from
one person to the next, and it may shift over time: what was
considered to be complex in the early days of the industrial
age, is likely to be complicated or perhaps even simple with
our current understanding of organizations and society. In light
of our objective to understand the antifragility of organizations,
this means that we have to take into account how much
stakeholders (can) know about the organizations that we are
studying.
D. (Anti)fragile, variety, and the learning organization
The term stressor is used for an event from outside the
system that causes stress in the system. For this study we use
the following definition of a stressor: “When systems are per-
forming effectively, they are in a predetermined condition and
conversely when they are not functioning correctly, they are in
an unintended state. An unintended condition can be known
or unknown. Stressors are forces that threaten to transfer a
system from an intended to an unintended condition” [29],
[30]. Three relevant theories for dealing with stressors are the
notion of (anti)fragile, variety, and the learning organization.
1) (Anti)fragile: Fragile is the concept of losing value from
exposure to stressors. Antifragile is the antithesis of fragile:
it is the concept of gaining value from exposure to stressors.
The concept of stressors having no effect on the value is called
robust [12]. Fragile, robust and antifragile form together a triad
[12], [31]–[34]. Figure 1 illustrates the differences in the way
the three main categories deal with stress.
Resilience is a concept that is often mentioned related to
(anti)fragile. Resilience is the ability to recover from or adjust
Fig. 1. Triad of fragile, robust, and antifragile
easily to misfortune or change [13]. Figure 2 illustrates the
generic concept of resilience.
Fig. 2. Generic concept of resilience
There are many definitions of resilience. We choose to adopt
the definition of Martin-Breen [13]. This definition was also
used in the found literature on antifragility [14].
Three sub-types of resilience are distinguished: (1) engi-
neering resilience, (2) systems resilience, and (3) complex
adaptive systems (CAS) resilience. The goal of engineering
resilience is to prevent disruption and changes and to bounce
back to the fixed function/basis [13], [14], [35]. Engineering
resilience can be measured by the following three characteris-
tics: resistance, elasticity and stability [13]. The function and
construction of the system stay the same over time. In case
of systems resilience, the system has the capacity to absorb
disturbance and reorganize while undergoing changes. While
doing this, the system retains essentially the same function,
structure, identity, and feedback, where ‘essential’ is defined
as “something functional and not identical” [13], [36]. In this
situation, the system is able to withstand the impact of any
interruption and recuperate while resuming its operations [37],
the function of the system stays the same over time, and the
construction of the system may change. With CAS resilience,
the system is able to become more resilient and to generate
new system relationships by reorganization [13], [14]. In this
case, the function is maintained, but system structure may
change [13]. This results in the system to being as dynamic
as the world around them, thus a system that is constantly
evolving [14]. The function of the system may change over
time, and the construction of the system may change over
time. Figure 3 illustrates the differences in the way the three
sub-types of resilience deal with stress.
2) Variety: A different way to study complex adaptive
systems is via the concept of variety. Based on the work of
Ashby and Beer there are two types of variety manipulations:
(1) to attenuate variety and (2) to amplify variety [22], [38].
To attenuate variety is reducing the variety in a system. The
absorption of change in the context of systems reduces variety.
To amplify variety is increasing the variety in a system.
To amplify internal variety is about increasing the chance
4
Fig. 3. Three sub-types of resilience, based on Martin-Breen 2011
of a higher entropy and therefore being more capable to
absorb increasing external variety caused by change. Emer-
gence leads to variety amplification. Engineering and systems
resilience are considered to result from attenuating variety.
CAS resilience and antifragile are considered to result from
amplifying variety.
3) Learning organization: A final way to deal with stres-
sors is to adapt concepts from the learning organization as
described by Senge [39] and Garvin [40]. The learning organi-
zation is a way to create resilient organizations which let them
cope with unknown and unpredictable events. “Continuous
improvement requires commitment to learning.” [40].
Garvin provides an actionable approach to creating the
learning organization where Senge provides an holistic view
on which disciplines are needed to create a learning organi-
zation. Therefore we select Senge to provide a framework to
capture the attributes relevant in creating a learning organiza-
tion.
Senge identifies five disciplines that together form the learn-
ing organization: (1) Personal mastery, which is a discipline
of continually clarifying and deepening our personal vision, of
focusing our energies, of developing patience, and of seeing
reality objectively, (2) Shared mental models. Mental models
are deeply ingrained assumptions, generalizations, or even
pictures of images that influence how we understand the world
and how we take action, (3) Building shared vision, which is a
practice of unearthing shared pictures of the future that foster
genuine commitment and enrollment rather than compliance,
(4) Team learning, that starts with ’dialogue’, the capacity of
members of a team to suspend assumptions and enter into
genuine ’thinking together’, and (5) Systems thinking. This
is the fifth discipline that integrates the other four. Systems
thinking needs the disciplines of building shared vision, mental
models, team learning, and personal mastery to realize its
potential.
Building shared vision fosters a commitment to the long
term. Mental models focus on the openness needed to unearth
shortcomings in our present ways of seeing the world. Team
learning develops the skills of groups of people to look for
the larger picture beyond individual perspectives, and personal
mastery fosters the personal motivation to continually learn
how our actions affect our world [39].
E. Conclusion
In summary, we conclude that organizations can be viewed
through the lens of systems theory, and generally have a high
degree of complexity. The way variety is handled and the
degree to which the organization adopts the principles of the
learning organization, largely determines the nature of the
organization - ranging from fragile via robust to antifragile.
III. METHODOLOGY
In order to answer our research question, we need two
things: first, we need to understand which characteristics influ-
ence the antifragility of organizations, when studied through
the lens of organizations as complex systems. Then, we need
to arrange these characteristics in a structured way.
The key input for these two steps is a survey of the available
literature that is validated by experts and practitioners. This
leads to a research approach with the following steps:
1) Search and select literature (search + snowball).
2) Categorize and summarize the literature.
3) Select most relevant literature.
4) Identify system attributes described in the relevant liter-
ature.
5) Create a structured way to categorize system attributes
and use this to design a sorting algorithm (decision tree)
to apply to the selected attributes.
6) Sort the identified system attributes by applying the
decision tree.
7) Validate the results so far with experts and practitioners.
8) Design the final attribute list by mapping the sorted
attributes back on the structured way that was designed
in step 5.
9) Validate the final attribute list with experts and practi-
tioners.
Section III-A describes the characteristics of our literature
survey in more detail. Section III-B explains how we used
triangulation to safeguard the quality of our results. The
combination of research methods applied in this research can
best be described as a post-positivist exploratory qualitative
naturalistic field-study research [15], [41]–[46].
A. Characteristics of validated literature survey
The systematic literature research on antifragility is used to
identify system attributes and it also provided the structure to
order the found attributes.
An initial scan of the available literature on antifragility sug-
gests that the body of scientific literature (i.e. peer-reviewed)
is relatively small, and that there are few case reports that
describe how lessons learned have been applied in practice.
Similarly, the ‘practical’ body of knowledge on antifragility
is relatively small which is unsurprising for such a novel
topic. This impacts our methodology in the sense that a
critical literature survey will likely yield few results. We limit
our search for relevant antifragile attributes by applying the
snowball method (figure 4) [17], [47]. For other (related)
topics (e.g. systems theory, complexity theory), the body of
research is more mature. We have adopted an approach that
is exploratory in nature: we focus on the knowledge question
of which attributes impact the antifragility of organizations,
rather than the design question of how such attributes can be
impacted through interventions [43], [48].
5
Fig. 4. Snowballing procedure [17]
In order to validate the results of the literature, reviews
are conducted with both experts and practitioners. The practi-
tioners consists of c-level managers responsible for executing
organizational change. The experts consists of architects and
consultants responsible for designing organizational change.
The objective of this validation is to evaluate and improve
the outcome of the literature survey [49]. Since the topic of
complexity theory and complex adaptive systems is mostly
unknown to the various experts and c-level managers, the
naturalistic research approach is the most logical research
method for the validation of the attribute list. Naturalistic
research is described as: “The researcher seeks to make the
research experience as much a part of the subjects’ everyday
environment as possible” [45], and as “A research method to
be set in a natural setting in an attempt to explain or interpret
a certain phenomenon” [15], [46].
B. Triangulation
The combination of literature review, expert review, and
practitioners review is a form of triangulation. Triangulation
is defined as the use of multiple methods mainly qualitative
and quantitative methods in studying the same phenomenon
[50], [51]. Triangulation, as is shown in figure 5, is applied
for the validation of the created EAAL, consisting of the list
of attributes found in the literature survey [50], [51].
Fig. 5. Main research methods [52]
IV. RES ULTS
A. Step 1: search and select literature
The search for literature starts with the following three types
of sources. The literature list in these sources functioned as
the start of the snowballing procedure [17], [47]. As primary
sources for the snowballing procedure we used the references
from the book Antifragile by Taleb [12] and the references
from the Wikipedia pages on ‘Antifragile’ and ‘Antifragility’.
We continued with the snowballing procedure until it stopped
growing in size. For the secondary sources, several academic
search engines are used, namely: (1) Google Scholar, (2)
Bing Academic, (3) Semantic Scholar, (4) ReseachGate, (5)
Citationsy, and the (6) Library of Antwerp Management
School. These are further extended by additional searches
in: (7) Amazon.de, (8) Goodreads, (9) Google Books, (10)
Diva (Sweden), (11) Scripties Online (the Netherlands), (12)
Narcis (the Netherlands), (13) OpenThesis.org (USA), and
(14) OATD (Global).
The following keywords initially where used for the search
queries: antifragile,anti-fragile,antifragility,anti-fragility,
Taleb,Nassim Taleb,antifragile organisations,antifragile orga-
nizations,anti-fragile organisations,anti-fragile organizations.
The literature search was conducted between October 2018
and June 2019. This step resulted in 358 sources1, of which in
total 87 sources2where categorized in the following research
steps.
B. Step 2: categorization and summarize the literature
The 87 sources are labeled according to one or more of
the following six categories: (1) Antifragile, (2) Antifragile &
IT, (3) Organization, (4) Risk and Resilience, (5) Complexity
Science, and (6) Science. These categories emerged during
the creative process of labeling and were validated with
experts and practitioners in the triangulation as described in
Section III-B.
C. Step 3: select relevant literature
A first reading of the 87 categorized sources suggested
that several sources were not as relevant as initially hoped
and expected. To narrow down the set, we used a filter with
the following criteria: must contain listings of attributes that
are linked to antifragile behavior, comprehensiveness, and
relevance. This narrowed the list of sources to nine: [34], [53],
[32], [54], [15], [14], [31], [33], [55].
D. Step 4: select system attributes
The following attributes where provided in the individual
sources.
•Ghasemi and Alizadeh [34]: Absorption, Redundancy,
Introduction of low level stress, Eliminating stress, Non-
monotonicity, Requisite variety, Emergence, Uncoupling.
1https://gitlab.com/edzob/antifragile-research/-/wikis/
EAAL-literature- thesis
2https://gitlab.com/edzob/antifragile-research/-/wikis/
EAAL-literature- selection
6
•Johnson and Gheorghe [53]: Entropy, Emergence, Effi-
ciency vs. risk, Balancing constraints vs. freedom, Cou-
pling (loose/tight), Requisite variety, Stress starvation,
Redundancy, Non-monotonicity, Absorption.
•Kennon et al. [32]: Emergence, Efficiency and risk, Req-
uisite variety, Stress starvation, Redundancy, Absorption,
Induced small stressors, Non-monotonicity.
•Markey-Towler [54]: High in openness, High in consci-
entiousness, Fair degree of extraversion, Moderate degree
of agreeableness, Low in neuroticism.
•Henriksson et al. [15]: Strategy - Design versus emer-
gence, Strategy - Seneca’s barbell strategy, Opportunities
- networks, Opportunities - innovation, Opportunities -
resources, Motivation - mind-set, Motivation - employee
motivation, Motivation - communication
•Kastner [14]: Self-organization, Ownership (result based
system /’skin in the game’), Diversity of cells and or-
ganizational learning, DNA-shared purpose, values and
culture.
•Gorgeon [31]: Simplicity, Skin in the game, Reduce
naive interventions, Optionality, Inject randomness into
the system, Decentralize / develop layered systems.
•Hole [33]: Modularity, Weak links, Redundancy, Diver-
sity, Fail fast, Systemic failure without failed modules,
The need for models.
•O’Reilly [55]: Modularity, Weak links, Redundancy, Di-
versity.
E. Step 5: create sorting algorithm for organizations
In order to transform the list of found attributes into a
condensed and structured list, we needed to develop a sorting
algorithm that helps us to determine which categories we
should use, and which attributes belong in a specific category.
The sorting algorithm was developed in several iterations,
and is based on concepts found in the retrieved literature. As
before, we used triangulation to validate the sorting algorithm
as discussed in Section III-B.
Figure 6 comprises the structure that we designed to catego-
rize system attributes, which is mainly based on the discussion
of (anti)fragile, variety, and the learning organization in Sec-
tion II-D.
Fig. 6. Structure to categorize system attributes
Using this structure, we ended up with the decision tree that
is visualized in figure 7.
F. Step 6: sort the identified system attributes
While we initially hoped that classifying the attributes with
our decision tree would be an almost mechanical process,
Fig. 7. Decision tree to order the attributes
this turned out to be not feasible. The overall context of the
description of the attribute in the selected works and expert
judgment were needed to come to a sort. Therefore, we ended
up following a creative process in the sorting of the attributes.
The following examples are illustrative for this process:
•Redundancy was identified in [32]–[34], [53], [55]. It
does not describe an aspect of the learning organization
and its main goal is to maintain the function of a system
when a sub-system fails. This attenuates the variety.
The use of sub-systems to maintain the functionality is
described by Martin-Breen [13] as attribute of Systems
Resilience.
•Another example is Efficiency vs. risk as discussed in
[53]. This attribute does not fit the learning organization.
Efficiency does attenuate variety and is described similar
to engineering resilience by Martin-Breen [13], and risk
is described in line with freedom of employees and fits
more Martin-Breen’s description of CAS Resilience than
Taleb’s description of Seneca’s barbell.
The result of this step is the EAAL that will be explained later
in this Section.
G. Step 7: reviews
1) Expert review: The summary of the found literature in
the form of the EAAL was validated by 18 experts. Validation
took place in one-by-one sessions (10 persons) and in two
group sessions (10 and 3 persons). There was a small overlap
designed in this setup to add extra independent validation of
the outcome of these sessions.
The experts had different backgrounds and points-of-view.
The expertise topic of the experts are ‘at least’ one of the
following list: antifragility, enterprise architecture, enterprise
engineering, organizational design, organizational change.
The two questions asked to the experts were: (1) Does it
make sense what I am telling you? and (2) Do you see any
big mistakes, blind spots or contrary statements?
The one-by-one sessions were semi-structured interviews
with a duration of 30 minutes to 90 minutes per interview. The
group sessions were in the form of a presentation concluded
with an open discussion. The duration of these sessions was
120 minutes.
7
2) Practitioners review: To review the applicability and
relevance of the EAAL, seven C-level managers (CFO, CIO,
CTO, COO) of seven different organizations were interviewed.
These managers are responsible for executing change. Table
I contains the demographics of the interviewees. The method
applied was that of semi-structured interviews with a maxi-
mum duration of one hour.
The objective of the interviews was to (1) check the the-
oretical background, (2) explore the concepts, and (3) check
the relevance and application of the EAAL.
TABLE I
DEMOGRAPHICS PRACTITIONERS
Role Type of organization FTE
CIO Aviation company 70.000 - 90.000
CIO Aviation company 1.000 - 5.000
CIO Governmental
organization
10 - 200
COO Non-profit organization 10 - 200
CIO University 200 - 400
CFO University 1.000 - 5.000
CTO Retail 200 - 400
H. Step 8: The EAAL as validated summary
The EAAL is the result of all previous steps as discussed
in this article. In essense, this meant that we had to plot the
attributes that resulted from step 5 back on the main structure
that was presented in figure 6. Figure 8 shows the result of this
analysis, and the final content of the EAAL. Table II defines
the different attributes of the EAAL.
Fig. 8. The EAAL
I. Step 9: reviews
1) Expert review: The overall feedback from the one-
on-one expert reviews was that the EAAL model sparked
inspiration and was relevant to their individual expertise. The
main feedback that resulted into (re)designing the EAAL
model where the following feedback points: (1) all the aspects
of a learning organization as described by Senge (1990) are
always relevant [39], (2) it is impossible to find out based on
observation why something is working. In this case, the expert
was referring to the function and construction dilemma and to
the issue of determining causal relations based on observation.
The overall feedback from the group validations was also
that the EAAL model sparked inspiration and is relevant for
the future. The group of 10 experts also provided feedback
that the topic was pretty complex and would need examples
on how to apply this in their day-to-day work.
Based on the feedback of the experts, the main adjustment
in the EAAL model was that we positioned the learning
organization across “attenuate variety and amplify variety”. In
the first version of the EAAL model, the learning organization
was positioned only across “amplify variety”.
2) Practitioner review: The overall feedback from the
practitioners was that the EAAL model sparked inspiration but
it can not be used in the boardroom in this form. Instead, the
EAAL model should be used in the department that advises
the board during the analysis and design process. Based on
the EAAL model, ‘smart’ questions can be created and asked
in the boardroom.
Details of the feedback from the interviews are:
•“For the first time I see a holistic overview of the field
of antifragility, resilience and agile”.
•“This is a powerful tool to think about the future design
of the company”.
•“This does not resonate with me. I do not understand
what you are trying to tell me”.
•“This all sounds very logical and clear to me. Great to
have it in one image”.
•“After your talk with our CTO he is researching holacracy
and the possibilities in the transformation of our organi-
zation”.
The feedback of the practitioners did not lead to adjustments
of the EAAL.
V. CONCLUSION
A. Conclusion
The literature search regarding antifragile and the subse-
quent expert and practitioner reviews led to the development of
a structured list (model), called Extended Antifragile Attribute
List (EAAL). The objective of the EAAL is to help organiza-
tions achieve antifragile properties. The main contribution of
this study is that we provide insight in the characteristics of
resilient and antifragile organizations. The lens on resilience
is that from the point of view of achieving antifragility. For
scientists, the EAAL extends theories of organizational design.
For practitioners, the EAAL provides attributes which are
useful in designing organizations or parts of it.
B. Discussion
The applicability of the EAAL needs further consideration.
Looking from a traditional perspective on systems think-
ing towards an antifragile world leads to a non-productive
discussion, because the mindset and paradigms seem to be
completely opposites.
Even though antifragile has advantages (in the sense that
it helps organizations to more effectively deal with the ripple
effects of changes from inside and outside the organization),
we argue that antifragile is neither a goal in itself, nor does
8
it replace current methodologies for strategic management,
governance and enterprise architecture. Instead, our position
is that it is a factor to be considered in these discussions,
and the EAAL is a useful tool for structuring the debate.
EAAL, considering the current economic and social impact
on organizations and people of the Covid-pandemic, is a
relevant as well a timely list of attributes which offers a new
perspective, next to traditional systems thinking and systems
design [18]–[21].
We state the EAAL to be a promising model providing
a holistic view of the domain of antifragility. From this
(holistic) perspective, the attributes grouped per behavior type
of the organization, provided by the EAAL, can be utilized as
a grindstone during the (re-)design of strategy, governance,
and enterprise architecture addressing the resilient and/ or
antifragile behavior of the organization. The EAAL turns out
to be more extensive than the narrow focus of a book review
starting with Taleb’s [12].
C. Limitations
The selected literature for the synthesis into the EAAL,
is mostly based on deduction, reasoning, and synthesis. The
EAAL is validated through this lens. The EAAL and the
underlying literature is not (yet) validated in its application.
D. Future research
Future research is needed to (a) improve and validate the
EAAL, and (b) develop tools for its application in the real
world. We will discuss each in turn.
The EAAL can be improved in rigor via replication of
the literature research, and validation by an expert group.
The literature research replication should apply an even more
rigorous manner while including a more extensive explanation
of arbitrary choices, for example by applying the reference-
matrix mentioned by Wohlin [17]. The validation replication
could be established by organizing a program of group support
sessions, in which the same validation script is followed in
a relay mode with C-level practitioners (such as CIO’s or
CFO’s) and domain experts (such as IT and Risk Management)
from different sectors (such as Health Care, and Finance). The
extended validation is needed to verify possible application in
the generic system design field.
We also intend to broaden our perspective with a focus on
the humanities of complex and antifragile organizations. The
statement by Markey-Towler [54] and Kastner [14] that the
antifragile organization fits certain personality types better than
others, leads to the following future research questions. Firstly,
considering that an antifragile organization itself is a system-
of-systems including less resilient sub-systems, diversity is an
attribute of a CAS-resilience system and Seneca’s barbell is
an attribute of an antifragile system. This would result into
the reasoning that a more heterogeneous representation of
the various personality types in an organization is preferable
over a more homogeneous. Secondly, the statement that the
current view on an antifragile organization prefers a certain
type of personality, raises the ethical question of the impact
on the people in the organization when applying organizational
change to improve resilience.
Further research is needed to identify and develop an
antifragile mindset and tool-set based on human behavioral
concepts, such as emotional maturity in agile teams. These
concepts are not well represented in the domains of enterprise
governance and architecture. These fields have their roots in
the reductionist approach and are considered helpful in the
Cynefin domains of complicated and simple/ clear.
Digital transformation, fueled by the Covid-pandemic, and
the agile way of working have increased connections within
and between organizations. This extra connectivity results in
organizations being confronted with a complex and chaotic
context. Human behavior plays a major role in countering this
increased complexity [16], [20], [21], [28], [38].
REFERENCES
[1] M. Op’t Land, E. Proper, M. Waage, J. Cloo, and C. Steghuis, Enterprise
Architecture: creating value by informed governance, ser. The Enterprise
Engineering Series. Berlin, Germany: Springer Science & Business
Media, oct 2008.
[2] B. Wit and R. Meyer, Strategy Synthesis, Revolving Strategy Paradoxes
to Create Competitive Advantage - Concise version. Thomson, 2006.
[3] G. Hutchins, ISO 31000: 2018 Enterprise Risk Management, ser. CERM
Academy Series on Enterprise Risk Management. Certified Enterprise
Risk Manager(R) Academy, nov 2018.
[4] N. Bennett and G. J. Lemoine, “What a difference a word makes: Un-
derstanding threats to performance in a vuca world,” Business Horizons,
vol. 57, no. 3, pp. 311 – 317, may 2014.
[5] ——, “What vuca really means for you.” Harvard Business Review,
vol. 92, no. 1/2, feb 2014.
[6] O. Mack, A. Khare, A. Kr¨
amer, and T. Burgartz, Managing in a VUCA
World. Cham, Switzerland: Springer, jul 2015.
[7] S. D. Anthony, S. P. Viguerie, and A. Waldeck, “Corporate longevity:
Turbulence ahead for large organizations,” Strategy & Innovation,
vol. 14, no. 1, pp. 1–9, 2016.
[8] M. Fitzgerald, N. Kruschwitz, D. Bonnet, and M. Welch, “Embracing
digital technology: A new strategic imperative,” MIT sloan management
review, vol. 55, no. 2, p. 1, 2014.
[9] W. Aghina, K. Ahlb¨
ack, A. De Smet, C. Fahrbach, C. Handscomb,
G. Lackey, J. Woxholth, M. Lurie, and M. Murarka, “The 5 trademarks
of agile organizations,” McKinsey & Company, New York City, pp. 2–5,
dec 2017.
[10] G. C. Kane, A. Nguyen Phillips, J. R. Copulsky, and G. R. Andrus,
The technology fallacy - how people are the real key to digital trans-
formation, ser. Management on the cutting edge series. MIT Press,
2019.
[11] N. N. Taleb and R. Douady, “Mathematical definition, mapping, and
detection of (anti) fragility,” Quantitative Finance, vol. 13, no. 11, pp.
1677–1689, nov 2013.
[12] N. N. Taleb, Antifragile: Things That Gain from Disorder. New York,
NY, USA: Random House, nov 2012.
[13] P. Martin-Breen and J. M. Anderies, “The bellagio initiative, background
paper, resilience: A literature review,” in Resilience: A Literature Review,
Brighton:IDS, nov 2011.
[14] D. Kastner, “Antifragile organisation design: A framework of self-
organisation practices in today’s complex and unpredictable economy.”
Master’s thesis, Central Saint Martins - University of the Arts London,
Master innovation management, London, United Kingdom, jul 2017.
[15] V. Henriksson, M. Spiss, and C. van Houten, “To master disaster:
How sme managers can thrive and benefit from economic crises,”
Master’s thesis, Linnaeus University, School of Business and Economics,
Department of Marketing., V¨
axj¨
o, Sweden, jun 2016.
[16] J. A. Hoogervorst, Foundations of Enterprise Governance and Enter-
prise Engineering. Presenting the Employee-Centric Theory of organi-
sation. Springer, 2017.
[17] C. Wohlin, “Guidelines for snowballing in systematic literature studies
and a replication in software engineering,” in Proceedings of the 18th
international conference on evaluation and assessment in software
engineering, 2014, pp. 1–10.
9
[18] G. Morgan, F. Gregory, and C. Roach, Images of organization. SAGE
Publications, 1997.
[19] J. Achterbergh and D. Vriens, Organizations - Social Systems Conduct-
ing Experiments. Springer Verlag, 2009.
[20] M. C. Jackson, Critical Systems Thinking and the Management of
Complexity, 1st ed. Wiley, 2019.
[21] J. R. Turner and R. M. Baker, “Complexity theory: an overview with
potential applications for the social sciences,” Systems, vol. 7, no. 4,
2019.
[22] W. R. Ashby, An Introduction to Cybernetics. London, UK: Chapman
& Hall and University Paperbacks, 1956.
[23] E. D. Falkenberg, W. Hesse, P. Lindgreen, B. E. Nilsson, J. L. H. Oei,
C. Rolland, R. K. Stamper, F. J. M. V. Assche, A. A. Verrijn-Stuart, and
A. Voss, A Framework of Information Systems Concepts. IFIP WG 8.1
Task Group FRISCO, IFIP, 1998.
[24] K. E. Boulding, “General systems theory - the skeleton of science,” in
Modern systems research for the behavioral scientist, W. Buckley, Ed.,
1968, pp. 3–10.
[25] C. F. Kurtz and D. Snowden, “The new dynamics of strategy: Sense-
making in a complex and complicated world,” IBM systems journal,
vol. 42, no. 3, pp. 462–483, 2003.
[26] D. Snowden, “Strategy in the context of uncertainty,” Handbook of
Business strategy, 2005.
[27] D. J. Snowden and M. E. Boone, “A leader’s framework for decision
making,” Harvard Business Review, vol. 11, pp. 68–76, 2007.
[28] D. Snowden and Z. Goh, Cynefin - Weaving sense-making into the fabric
of our world. Cognitive Edge, 2020.
[29] B. L. Turner, R. E. Kasperson, P. A. Matson, J. J. McCarthy, R. W.
Corell, L. Christensen, N. Eckley, J. X. Kasperson, A. Luers, M. L.
Martello, C. Polsky, A. Pulsipher, and A. Schiller, “A framework for
vulnerability analysis in sustainability science.” Proceedings of the
national academy of sciences, vol. 100, no. 14, pp. 8074–8079, jul 2003.
[30] G. P. Chrousos, “Stress and disorders of the stress system.” Nature
reviews endocrinology, vol. 5, no. 7, pp. 374–381, jun 2009.
[31] A. Gorgeon, “Anti-fragile information systems (completed research
paper),” in Anti-fragile Information Systems, Fort Worth, dec 2015.
[32] D. Kennon, C. Schutte, and d. D. Lutters, “An alternative view to as-
sessing antifragility in an organisation: A case study in a manufacutring
sme.” CIRP annals: manufacturing technology, vol. 64, pp. 177–180,
apr 2015.
[33] K. J. Hole, Anti-fragile ICT Systems, ser. Simula SpringerBriefs on
Computing. Cham, Switzerland: Springer International Publishing, mar
2016.
[34] A. Ghasemi and M. Alizadeh, “Evaluating organisational antifragility via
fuzzy logic. the case of an iranian company producing banknotes and
security paper.” Operations Research and Decisions, vol. 1, pp. 21–43,
feb 2017.
[35] C. S. Holling, “Engineering resilience versus ecological resilience,”
Engineering within Ecological Constraints, vol. 31, no. 1996, pp. 31–43,
1996.
[36] B. H. Walker, C. S. Holling, S. R. Carpenter, and A. Kinzig, “Resilience,
adaptability and transformability in social–ecological systems,” Ecology
and society, vol. 9, no. 2, dec 2004.
[37] R. S. Santos, “ ¨
why resilience?¨
a review of literature of resilience and
implications for further educational research.” in Review of Resilience
Research. Claremont, CA: Claremont Graduate University & San Diego
State University, 2012.
[38] S. Beer, The heart of enterprise: the managerial cybernetics of organi-
zation, ser. Managerial cybernetics of organization. Chichester, West
Sussex, UK: John Wiley & Sons, 1979, vol. 2.
[39] P. M. Senge, The Fifth Discipline: The Art and Practice of the Learning
organisation, ser. A Currency book. New York, NY, USA: Double-
day/Currency, mar 1990.
[40] D. A. Garvin, “Building a learning organization,” Harvard business
review, vol. 71, no. 4, pp. 78–91, jul 1993.
[41] N. K. Denzin and Y. S. Lincoln, The SAGE Handbook of Qualitative
Research. Los Angles, USA: SAGE Publications, feb 2017.
[42] A. B. Ryan, “Post-positivist approaches to research,” Researching and
Writing your Thesis: a guide for postgraduate students, pp. 12–26, 2006.
[43] J. Recker, Scientific Research in Information Systems - A Beginner’s
Guide, ser. Progress in IS. Heidelberg, Berlin, Germany: Springer, jul
2012.
[44] C. Emory and D. R. Cooper, Business Research Methods, ser. Kent
International Dimensions of. Homewood, Illinois, USA: Irwin, jan
1991.
[45] J. W. Chesebro and D. J. Borisoff, “What makes qualitative research
qualitative?” Qualitative Research Reports in Communication, vol. 8,
no. 1, pp. 3–14, 2007.
[46] M. Alvesson and K. Sk¨
oldberg, Reflexive Methodology: New Vistas for
Qualitative Research. London, UK: SAGE Publications, sep 2009.
[47] University Library Groningen, “Search methods - information liter-
acy history - libguides at university of groningen,” https://libguides.
rug.nl/c.php?g=470628&p=3218096, jan 2020, (Online; accessed on
05/13/2020).
[48] R. J. Wieringa, Design science methodology for information systems and
software engineering. Springer, 2014.
[49] P. E. Newton, “Does it matter what ’validity’ means. - lecture
given at university of oxford, department of education,”
http://oucea.education.ox.ac.uk/wordpress/wp-content/uploads/2013/
06/2013-Meaning- of-validity-Oxford-v4-slides.pdf, feb 2013, (Online;
accessed on 03/02/2020).
[50] T. D. Jick, “Mixing qualitative and quantitative methods: Triangulation
in action,” Administrative Science Quarterly, vol. 24, no. 4, pp. 602–611,
dec 1979.
[51] A. Hussein, “The use of triangulation in social sciences research:
Can qualitative and quantitative methods be combined,” Journal of
comparative social work, vol. 1, no. 8, pp. 1–12, 2009.
[52] M. O. Alassafi, A. Alshdadi, R. J. Walters, and G. B. Wills, “A frame-
work for critical security factors that influence the decision of cloud
adoption by saudi government agencies,” Telematics and Informatics,
Elsevier, vol. 31, no. 7, pp. 996–1010, nov 2017.
[53] J. Johnson and A. V. Gheorghe, “Antifragility analysis and measurement
framework for systems of systems,” International Journal of Disaster
Risk Science, vol. 4, no. 4, pp. 159–168, dec 2013.
[54] B. Markey-Towler, “Antifragility, the black swan and psychology,”
Evolutionary and Institutional Economics Review, vol. 15, no. 2, pp.
367–384, dec 2018.
[55] B. M. O’Reilly, “No more snake oil: Architecting agility through
antifragility,” Procedia Computer Science, vol. 151, pp. 884–890, 2019.
10
TABLE II
EAAL ATTRIBUTES
Attribute Description References
Top-down C&C Top-down command and control applies when an employee does not have the freedom to decide
their own action but has to follow instructions from the organizational hierarchy.
The careful design of the features of an iPhone or a good pen are examples of limited freedom of
movement in the product itself.
[13], [53], [15],
[14], [55]
Micro-management Micro-management entails the freedom in the use of the product. A detailed working instruction
describing a business process, results in no freedom for the employee in the execution of their
job. An example is a Lego building block. It is engineered and fabricated with the greatest detail
resulting in a building block that is almost completely robust. Lego has a very small resilience
behavior through engineering.
[13], [53], [32],
[15], [34], [14]
Redundancy Redundancy is about having not a single point of failure by making use of duplication.
An example is a backup electricity generator. Another example is local government as backup
system of the central government.
[13], [53], [32],
[33], [34], [32], [55]
Modularity Modularity is the degree that components may be separated and recombined, often with the benefit
of flexibility. For example, a car with a standard chassis onto which different components can be
connected creating a unique car.
[13], [31], [33], [55]
Loosely coupled Loosely coupled is the degree of dependency on the exact working of another module. It is important
to understand that there is always some degree of coupling. Loosely coupled is also known as
‘weak links‘, ‘uncoupling‘, ‘loose/tight coupling‘, or ‘a low level of interconnectedness between
components‘. For example, when there are new employees introduced at the finance department
this should not impact the taste of the coffee at the same office.
[13], [53], [31],
[33], [34], [55],
Diversity Diversity is the ability to solve a problem in more than one way with different components.
Optionality, the availability of options, is a specialisation of diversity. An example is that within a
team you want diverse co-workers since other types of people come up with other types of solutions.
[13], [12], [31],
[33], [15], [14], [55]
Non-monotonicity Non-monotonicity is learning from bad experiences. Mistakes and failures can lead to new
information. As new information becomes available it defeats previous thinking, which can result
in new practices and approaches.
[53], [31], [32],
[33], [34]
Emergence When there is little or no traceable relation between micro and macro level output then emergence is
there. This is the situation where random things (unintended states) appear more often and X-events
(black swans) appear. The law or requisite variety applied in this reasoning, makes that internal
emergence counters external emergence, and this subsequently leads to antifragility
[53], [32], [15],
[14], [34]
Self-organization Self-organization is a process where some form of overall order arises from local interactions
between parts of an initially disordered system. For example, students sitting together in the school
cafeteria.
[32], [15], [14]
Insert low-level stress Continuous improvement is achieved by inserting low-level of stress continuously into a learning
system. This will keep the system sharp all the time.
[12], [32], [31], [34]
Network-connections A network is created by connections to other nodes. More connections increase the potential for
optionality for new constructions and also new functionalities.
[53], [31], [15],
[33], [14], [34], [54],
[55]
Fail fast The other combined attributes in this group enable the possibility to execute the strategy “fail fast”. [32], [31], [33], [34]
Resources to invest Opportunities can only be seized when there are resources free to do so. Resources can be money
but also time and labor. To survive, a black swan investment should be possible when required
[12], [31], [14], [15]
Seneca’s barbell To be antifragile a robust sub-system is needed to which 80-90% predictable value with low risk
is situated. The remaining 10-20% should be used for high return on investment activities.
[12], [53], [32], [15]
Insert randomness When insert-low-level stress and fail fast delivers no issues the next step is to insert randomness into
the systems. A great example of this is chaos engineering by Netflix or the HackerOne bug-bounty
system.
[12], [32], [31], [34]
Reduce naive intervention Naive intervention is an intervention based on a model and reductionistic logic which ignores
experience. An example is not listening to the experienced but not so articulate employee, or by
ignoring the balance nature has found in an ecosystem.
[12], [31], [14]
Skin in the game Make certain that the person making the decision and doing the work has a pain and gain relation
with the outcome. This goes beyond having a feedback system in place. This goes beyond having
KPI’s in place. An example is that when working Agile scrum, the product owner should be a
co-worker in the team for whom the solution is build.
[12], [15], [31], [14]
Personal mastery Personal mastery is a discipline of continually clarifying and deepening our personal vision, of
focusing our energies, of developing patience, and of seeing reality objectively.
[39], [31], [15],
[34], [33], [54]
Shared mental models Mental models are deeply ingrained assumptions, generalizations, or even pictures of images that
influence how we understand the world and how we take action.
[39], [33]
Building shared vision Building shared vision - a practice of unearthing shared pictures of the future that foster genuine
commitment and enrollment rather than compliance.
[39], [15], [14]
Team learning Team learning starts with ’dialogue’, the capacity of members of a team to suspend assumptions
and enter into genuine ’thinking together’.
[39], [53], [32],
[15], [33], [31], [33],
[14], [34]
Systems thinking Systems thinking is the Fifth Discipline that integrates the other four. Systems thinking also needs
the disciplines of building shared vision, mental models, team learning, and personal mastery to
realize its potential.
[39], [53], [31],
[32], [33], [34], [14],
[54], [55]
