Getting credit ratings ‘right’ seems vitally important to many professional observers and politicians. The increasingly volatile nature of markets in a post-Bretton Woods world of international capital mobility has created a crisis in relations between the rating agencies and governments, which seek to monitor the performance of the agencies and stimulate ‘reform’ in their procedures and business models, even if the exact purpose of this reform seems to elude them. This process started with the Enron bankruptcy, but the subprime crisis has generated a veritable ‘moral panic’ about agency performance in relation to asset-backed securities. In pursuing improvement in the rating system policy-makers need to appreciate the limits to rating. Our expectations of the agencies are founded on a rationalist or machine-like understanding of the workings of capital markets, and on an exogenous understanding of the causes of financial crisis. This worldview implies a correct rating can be determined, and that finding this correct answer is purely a technical matter. A more accurately social and dynamic view of markets and financial crises makes the challenge of effective rating even more daunting.