What are the limits to measurement? Considering just two examples, GDP and standardized test scores, we have limited ability to measure what matters, but almost unlimited faith in the reliability and importance of our measurements.

GDP is staggeringly imprecise. It doesn’t include the services we provide without payment (such as editing wikipedia or contributing to open source software projects). Likewise, it didn’t include the labour provided by women who looked after children and households until those services were outsourced. It suggests that it is better to work late, get drunk, smash your car and rack up a large hospital bill, than to take the subway home, cook dinner with your family and go to bed after reading bed time stories from library books. Yet, GDP is the de facto measure of economic development (and by proxy success and well-being) and the No. 1 driver of macro economic policy.

The situation is not much better for standardized test scores. They are at best a measure of the student’s ability to do what is expected, and proxy for the development of other more important skills. Thankfully, the schools that are good at preparing students to do well on standardized tests, tend to also be good at providing them with great learning environments. There is no reliable measure of curiosity or creativity, even though those seem to be better guides to a fulfilled and happy (and successful) life than size of vocabulary. Some of the things we care about even disappear when we try to measure them.