But the real bias inherent in algorithms is that they are, by nature, reductive. They are intended to sift through complicated, seemingly discrete information and make some sort of sense of it, which is the definition of reductive. But it goes further: the infiltration of algorithms into everyday life has brought us to a place where metrics tend to rule. This is true for education, medicine, finance, retailing, employment, and the creative arts. Are we puppets in a wired world, Sue Halpern, 'New York Review of Books', dated 7 NovemberThe potential problem is that the metrics we use in policymaking might not correlate with societal well-being. Unfortunately, the alternative to a coherent, explicit, considered use of metrics in national policymaking is our current system, which features the unsystematic and de facto use of incoherent metrics that are too narrow and short term in their scope to bring about a rational allocation of resources. Applying broad, meaningful metrics to the health sector, say, is going to be far more efficient and welfare-enhancing than targeting a particular disease, just because the scope for efficiency gains is far bigger when resources can shift between different activities according to where they will be most cost effective.
The other problem with narrow metrics, however well meaning, is that they can easily be gamed. Thus:
Yet indicators of maternal health [in Laos] are worse than in Cambodia ... and levels of malnutrition are atrociously high. To make things look not quite as bad, NGO types say, the government deliberately went around feeding children in villages monitored by the UN for the Millennium Development Goals—until it was found out. The future of Laos: a bleak landscape, 'The Economist', 26 October