When a failure has many contributing causes spread across many actors, the attribution of responsibility becomes a political rather than an analytical problem.
The Distributed Failure Problem
Many significant institutional failures are distributed — they result from the combination of decisions, behaviours, and structural conditions across multiple actors, none of whom individually caused the failure but each of whom contributed to the conditions that made it possible. The financial crisis was not caused by any single actor's decisions; it was produced by the combined effect of mortgage originators, securitisers, rating agencies, investors, regulators, and policymakers, each making locally rational decisions that together produced a systemically catastrophic outcome. The distributed character of such failures makes the attribution of responsibility analytically complex and politically contentious.
The analytical complexity arises from the causal structure: in a distributed failure, removing any single contributing cause would not necessarily have prevented the failure, because the other causes might have combined to produce it anyway. This makes it genuinely difficult to assess how much any specific actor's contribution increased the probability or severity of the outcome. The political contentiousness arises from the incentives: each contributing actor has interests in minimising their attributed share of responsibility, which produces competing accounts of the failure that each allocate responsibility to others.
The Attribution Resolution
The resolution of distributed failure attribution is typically political rather than analytical: responsibility is allocated according to the institutional power of the various contributing actors to shape the official account, not according to their actual causal contribution to the failure. This political resolution produces the specific accountability pattern of distributed failures: the most powerful contributing actors typically bear the least attributed responsibility, while the least powerful bear the most — regardless of the actual causal contributions that an analytical approach would identify.
Distributed failure attribution is the governance problem that institutional accountability processes are least equipped to handle and most need to handle well. The failures that are most systemic — most the product of many actors' combined decisions — are the failures where the attribution process is most political and least analytical. Which means they are the failures least likely to produce the structural changes that would prevent their recurrence.
Discussion