2024-10-19 event date. Dan Davies’s idea of the “accountability sink” is simple enough to fit into a business anecdote and sharp enough to describe a lot of modern institutions. The supplied essay says an accountability sink is a structure that absorbs or obscures the consequences of a decision so thoroughly that no one can be held directly responsible for it. That may sound abstract, but the example in the source is painfully concrete: a manager trims hotel cleaning staff, the room is not ready at check-in, and the front desk can do little more than offer a voucher.
The point is not just that something went wrong. It is that the person affected by the decision cannot get meaningful feedback to the person who made it. The chain between cause and consequence is broken. Once that happens, accountability disappears into the system rather than resting with anyone inside it. The essay says this can happen in hospitals, airlines, government agencies, and any other organization large enough to hide the origin of its own choices.
The source also brings in a second definition, from Sidney Dekker, that treats accountability as something you must be able to tell about: an account of what happened, why it happened, and who was involved. Put together, the two frameworks suggest that accountability requires both the power to change a decision and the ability to explain it. Without those, people are left with outcomes but no responsible actor.
That is why the essay sees obvious parallels with AI. Delegating decisions to algorithms can become a convenient way to build a sink even faster. An organization can point to the model, the metric, or the automated workflow and treat the result as if it emerged from nowhere. The article argues that this is not a new trick so much as an escalated version of an old organizational habit.
The essay’s real warning is about scale. Corporations and governments already have plenty of ways to diffuse responsibility. AI can make that easier by moving decisions even farther from the people who feel their effects. The source is careful not to claim that every use of algorithms is a sink. Instead, it argues that if a corporation has already been hard to hold accountable, an automated system will probably not make the job easier.
The practical conclusion is sobering: critics of AI accountability need tools stronger than simple complaint channels. If the system has broken the feedback loop, then reporting harm may never reach the person who can fix it. The source ends on that point: we need a new bag of tricks. The article itself is a reminder that technology debates are often really about how power hides.
Seen that way, accountability sinks are less a theoretical flourish than a map of where responsibility goes missing.
That makes the concept useful beyond the essay itself. Once you can name the sink, you can start looking for the broken link in your own systems: the point where feedback stops, where outcomes detach from decision-makers, or where metrics stand in for responsibility. The source’s warning is that AI can make those gaps easier to hide. The more important lesson is that any organization hoping to use automation well will need ways to preserve feedback, explanation and real human accountability.
