Fragment
Summary of the thesis
Several recent security incidents show that decision-making on cyber-security can
have consequences reaching far into the future. In a world of further digitalization,
interconnectedness, and increasing activities of cyber-criminals, the question is how the
decision-making needs to adapt to ensure security.
More than 15 years of research has been conducted in the field of security economics on
security investment decision-making. Although the field of security economics already
recognizes static limitations (data quality and invalid inferences), we posit security
investment decision-making is also impacted by dynamic limitations (understanding of
feedback, time delay, accumulation effects in this domain of decision-making). These
limitations may cause decision makers to use heuristics (simple mental rules) for making
decisions in complex, dynamic, and uncertain situations. The use of heuristics can
inadvertently and unconsciously lead to incorrect decisions. Therefore, our research
focusses on obtaining more knowledge and insights on these dynamic limitations.
The main research question of this thesis is: “Which systemic structures drive cyber-security
investment decision-making, and how can security investment decision-making potentially be
improved?”
Currently, many tools and approaches are available to support decision-making in the
field of cyber-security. Examples of these are risk management, benchmarks, standards,
frameworks, and acting after incidents. The key challenge to security investment
decision-making is that these tools and approaches merely address static limitations.
They focus more on policy setting, stakeholders’ goals, and intended output of decisions.
The underlying system that drives this output (performance) is only to a limited extent
captured by these tools. The structure-behaviour-paradigm implies that the operational
structure of any system drives the observed behaviour of that system. As a consequence,
decision makers often lack the means to estimate the long-term consequences of their
decision-making. The increasing costs of security are clear, while the benefits of avoiding
potential future attacks are vague, distant, and seemingly far away. This appearance
favours other business initiatives over security in investment decision-making.
On the basis of a literature study, we observe the following three important systemic
structures that drive security investment decision making: (1) “the attacker-defender
interaction”, (2) “the response of the resilient organisation”, and (3) “the security investment
decision-making financial optimum”. To better understand these systemic structures of
a complex and dynamic cyber-security decision-making system, we have developed a
‘red versus blue’ security awareness game grounded in system dynamics methodology.
This game provides insights into decision-making practices in an empirical setting.
Our game supports the existence of the attacker-defender interaction, the response
of the resilient organisation, and the security investment decision-making financial
optimum systemic structures, including the mutual interactions between these structures.
The analysis of game results confirms that decision-making in complex and dynamic
cyber-security decision-making systems is difficult. Experienced and knowledgeable
participants are not always aware of how their decisions impact the defenders’ future
security state. The game results show the usage of heuristics in complex and challenging
environments.
We tested how the three systemic structures observed in literature appear in different
cyber-security areas. We selected five key security topics closely related to the different
steps of the execution of a cyber-attack (kill chain). This approach allows us to have a
comprehensive empirical overview of the cyber-security field. We used a case study
approach grounded in group model building methodology. The topics of these case studies
were distributed denial of service (DDOS), malware, detection engines usage, insider
threat, and secure software development. Specific equation building techniques (following
system dynamics methodology) allowed us to replicate the observed behaviour of
organisational defence strategies across different domains (threat actor, security function,
business operations and finance). Doing so, we identified additional systemic structures
that drive this observed behaviour and were able to simulate the future effects of different
scenarios and security policy settings using identified controls. The case studies’ policy
analysis confirmed that the “attacker-defender interaction” structure will increase cost and
resources needed while “the response of the resilient organisation” may stabilise or lower
costs.
Building on the insights from our case studies, we were able to identify three core systemic
structures (in line with our literature research) that drive the cost of security, eleven
sub-structures that may act as a game changer (both for attacker or defender), and eight
generic policy options for the chief information security officer (CISO) to influence
the security strategy. The results are finally summarized in a sector specific archetype
describing the cyber-security investment decision-making dynamics. The contribution
to the system dynamics field is an expansion of the research in cybersecurity. In order
to be able to do this research, different academic fields had to be integrated: enterprise
architecture, system dynamics, finance and risk management. The contribution to the
security economics field is a first attempt to integrate different academic fields and
perform a systemic research approach to security investment decision-making.
Through our focus on dynamic limitations, we believe we were able to circumvent the
effects of certain observed static limitations in the field of security economics. However,
future research needs to confirm and deepen our insights. We did not execute any
comparative research on the current tools and approaches supporting the decisionmaking
process. Yet, we perceive this approach is complementary to them. We also found
several limitations in our research that provide possibilities for follow-up research. These
limitations are related to the research assumptions, the used methodology, the scoping,
the serious game, the case study models, and the research re-usability.
×