Part 3 - Anomaly detection - when ambidexterity tears organizations apart

In this third part, we present our first set of results addressing the first two critical research questions. We propose to shed light on an anomaly revealed by the different selected cases at Zodiac Aerospace and the comparison between the actual course of action, its controversies and predictions expected from the adaptive, interactive and encapsulated models.

The first chapter, through three different case studies, highlights the difficulties rising from organizational ambidexterity. Generative processes black-boxed in projects perturb the edges of models relying on a non-mutual conditioning between exploration and exploitation as this separation is no longer valid the unknown. We reveal biases and fixations effects deriving from the dichotomy that paradoxically jeopardizes the exploration and the exploitation regimes themselves. It also severs sustained innovation. We find that generative processes are fixated by organization designs, consequently creating tensions for middle/top management and strategic coherence.

These results are a reformulation of three articles addressing these cases with their own specific literature. The Icing Condition detection case, is examined through the lens of decision-making, preferences reversal, and exploration project management in (Le Glatin, Le Masson, and Weil 2017), available here in Publications. The Design Thinking cases hosted by the Airbus Development Team are discussed through the teachings of design theory (C-K) by focusing on the generativity of the method with respect to these exploration projects, see (Le Glatin, Le Masson, and Weil 2016; Le Glatin et al. 2018), available here in Projects. Finally, ADT cases and Business Class seat platform are presented in (Le Glatin, Le Masson, and Weil 2018), see publication. It suggests to consider non-expected utility theories and limitations of organizational ambidexterity.

The second chapter then proposes a baseline for the following modelling part as we need to find an extended model capable of explaining with better clarity the observed phenomena. As we tasked ourselves earlier on, we adopt a behavioural approach used to define modelling requirements based on the limitations observed in our case studies.