The Politics of Large Numbers: A History of Statistical Reasoning

"Industrial policy includes so many different facets, and I’m trying to pick one from every lens you might have to use in order to understand it. Statistics is a really important part of understanding industry and the state. You may know this, but the word ‘statistics’ actually comes from the Latin statisticum , of the state, and the Italian word statista , which means statesman. The development of the state and statistics are synonymous with each other. You have to know who a population is. Who are these people—we need a census. What is our land—we need a cadastral survey, we need to actually look at the territory and understand it. What do we own, and how do we subdivide that? How do we create parcels so that people can buy a patch of land and give it to someone else? Alain Desrosières is looking mostly at France, and he is investigating how statistics came to be from around the 1600s into the 1800s. He shows how it reflects the development of the French state at this period of time, which, similar to England and elsewhere in Europe, goes from this rural, feudal society of the Middle Ages into the modernity that we know today. “The development of the state and statistics are synonymous with each other” What he argues is that society transforms into an economy of measurement. One of the most striking early passages from the book is around the sheer scale of measurement in the modern economy. You have accounting, you have zoning, you have taxation, you have weights and measures. You have people who check that the gas pump is actually pumping the liter of petrol that you expect. You have to measure public health, like everything we’ve seen with COVID recently. That’s all a form of measurement. How do we know how many people participate in the economy? And then as we get into the 1900s: how do we measure the GDP? How do we measure employment? We’re obviously focused on inflation these days, but all the inflation data is based around the consumer price index. How do we know that prices are the same for fruit and vegetables and bread compared to last month? He’s really showing that the modern state could not exist without the development of statistics , and vice versa. The evolution of statistics is triggered by the state. As the state needs more systems in place, statistics come to back it up. Econometrics, as an example, came together in the 1930s as a fusion of economics and statistics, and it was triggered by the American New Deal. Suddenly we had all these new programs that the Roosevelt administration was launching, and we had no way of gauging whether any of them worked. So suddenly economists and statisticians said, ‘Okay, we need new methods to see if when we give farmers a certain amount of money in a certain period of time, that leads to positive gains—or not.’ So Desrosières’s whole book is focused on that evolution, and he has an amazing number of historical examples. It’s one of those completely eye-opening works that is really special. There are obviously a lot of books on this. The sociology of quantification has gotten really, really large in the last decade. As we’ve seen the rise of the algorithm and machine learning, a lot of scholars are starting to ask, ‘What do you do about all these numbers and how much they impact our daily lives?’ One of the canonical works is Trust in Numbers . Theodore Porter, writing back in 1986, shows how it was really a choice for science to become the objective field it is today—physics, biology, etc. are all statistical in nature. People asked for objectivity and statistics offered the methods to get there. While statistics and quantification are not necessarily objective—you can massage the statistics, all statistics lie, “lies, damned lies and statistics” (as Disraeli supposedly said)—we’ve accepted as a culture that it feels like quantification is more objective than qualitative judgment. That is arbitrary. It’s a cultural decision we have made as a society. One could easily argue that qualitative judgments are actually stronger and better than quantitative judgments. You’re including more facts because the human brain is better able to synthesize disparate pieces of information than a statistic can. Porter really explains the context and situational development of quantification over the history of science and in public life. Presidents used to be guided by their instincts and by advice from advisors. Now, they look at the GDP numbers and survey polls and ask, ‘Are they up or down?’ Tough and complex decisions get narrowed to a single number that’s enticingly precise and yet limited. We’ve lost something with the movement toward quantification and that’s really important in the context of industrial policy. On the one hand, there are all these stats available—this is why the Amsden book gets a little tough, you can look at all these spreadsheets and try to decide what to do. Or, we can take a step back and say, ‘What are we trying to accomplish as a nation? What are we trying to do?’ When you get into leadership around the economy, there’s this huge challenge of wanting to lead towards the future, but we obviously don’t know what it will be. Is quantum computing next, or will it be the metaverse? I think that’s where critical studies of quantification can become a very important lens to say, ‘Look, sometimes spreadsheets are precise but wrong, and to allocate well you have to qualitatively synthesize what you can and make a decision that affects people in the best possible way.’"