Lilla Hortoványi

Strategy Without Templates

Adaptation in Digital Environments

9.3 Causal Opacity: Why Hard to Read

Selective visibility is only part of the problem. Even when firms can observe outcomes, they often cannot explain them with confidence. The third mechanism is, therefore, causal opacity. In algorithmically mediated settings, the causes behind visible outcomes are often difficult to identify because they are technically complex, partly hidden, and distributed across many actors and systems.

One source of opacity is technical complexity. Algorithmic systems may incorporate many variables, changing weights, interaction effects, and dynamic responses. Machine learning systems can generate useful predictions without producing explanations that are meaningful to organizational actors. In practical terms, an algorithm may rank products or allocate visibility effectively while leaving users uncertain about why a particular result occurred.

A second source is proprietary design. Platforms have strong incentives not to disclose the details of ranking and recommendation systems. They seek to protect competitive advantage, reduce gaming, and preserve control over ecosystem behavior. As Pasquale (2015) argues, black-box systems can shape outcomes in important ways while remaining difficult to inspect or contest.

A third source is distributed attribution. Outcomes rarely follow from one isolated cause. They arise from interactions among firm behavior, platform rules, competitors, user responses, and random variation. Bleher et al. (2022) show this clearly in AI-supported clinical decision-making, where outcomes emerge from hybrid processes involving both professionals and algorithmic systems. When many actors and systems shape the result, responsibility and explanation become difficult to assign.

This matters because classical organizational learning assumes some degree of causal legibility. Cybernetic models assume deviations can be traced and corrected. Single-loop learning assumes organizations can observe a performance gap and modify behavior. Double-loop learning assumes they can revisit the assumptions behind action. All of these models presuppose, to different degrees, that the relation between action and outcome can be made intelligible enough for learning. Under algorithmic mediation, that assumption weakens.

Organizations, therefore, often shift from causal explanation to practical inference. They do not always learn by identifying why an outcome occurred. More often, they look for recurring patterns across repeated interactions, test small changes, and build provisional judgments about what tends to work. Möhlmann et al. (2023) show such pattern-based sensemaking among platform workers. Liu et al. (2022) similarly show that ridesharing drivers often distrust algorithmic guidance because they cannot understand the logic behind it. Kronblad et al. (2024) show how multilayered black-boxing in public sector settings compounds opacity and obstructs corrective feedback.

The core claim here is important. The problem is not simply uncertainty. Organizations have always operated under uncertainty. The stronger claim is that digital infrastructures often make causal structure systematically harder to observe. The firm can see effects, but not the mechanism that produced them. That changes what learning looks like.

Tartalomjegyzék

Kiadó: Akadémiai Kiadó

Online megjelenés éve: 2026

ISBN: 978 963 664 204 4

Gazdaság

What happens when understanding comes only after action has already begun?

Traditional strategy rests on the assumption that organizations can understand their environment before deciding how to act. Yet the conditions that once allowed organizations to rely on benchmarking, best practices, and proven strategic templates can no longer be taken for granted. Today, organizations increasingly face situations for which no clear roadmap exists. Established assumptions become less reliable, familiar reference points lose their clarity, and strategic decisions must be made before their consequences can be fully understood.

Strategy Without Templates explores how organizations learn, adapt, and navigate environments in which uncertainty is pervasive and established templates are absent or no longer sufficient. Instead of treating strategy as a process of prediction and planning, the book explores how strategic paths take shape through action, experimentation, adjustment, and learning.

A central insight in the book is that temporary solutions are often necessary. What begins as a practical response to an immediate challenge may gradually shape future possibilities in unexpected ways. Some solutions create new opportunities and sources of advantage. Others become constraints that are difficult to overcome.

Hivatkozás: https://mersz.hu/hortovanyi-strategy-without-templates//

BibTeX EndNote Mendeley Zotero