Other Writing
(with João Carabetta Fernanda Scovino Frederico Israel Diego Oliveira )
July 2022
Abstract (click to expand): In this paper we explain how the Data Basis platform helps decisively solve the data access problem for different types of users. We describe its core products: a powerful search engine, a freely accessible data lake featuring a unified schema and hundreds of interoperable tables, and APIs in various programming languages. We exemplify the platform’s utility with discussions of three datasets on labor markets, elections, and local public finances in Brazil. The project is extraordinarily cost-effective: dividing a measure of yearly benefits generated by a conservative estimate of yearly costs to run the organization yields a lower bound social return of 74. We conclude by laying out a roadmap to guide the organization's future steps.
September 2020
Abstract (click to expand): Assuming that ideology, or political positions, can be meaningfully partitioned into groups of issues, in this essay I outline a nine-dimensional political label that is (i) much more informative than a one-dimensional left-right divide and (ii) still relatively simple to communicate. I first describe political labels as simplifying devices that reduce dimensionality from the full position space, then introduce the label and, lastly, I discuss how the framework relates to basic questions in politics.
May 2020
Abstract (click to expand): This essay introduces the concept of a belief cloud as a representation of belief structure over claims humans can make. It divides claims into facts, and those others that are still open for debate. The cloud represents the current state of our knowledge, and it also helps visualizing how it grows. Based on this framework, I discuss various applications in communication, policy making, business, and science.
May 2019
Abstract (click to expand): This essay provides coherent definitions of two bedrock concepts in philosophy and statistics: randomness and probability. When constructing the first, I define repeatability, the measurement set, and distinguish between frequency vs. value prediction. The definition of randomness proposed, namely of a random variable not being perfectly value-predictable for any given information set, is stronger than those commonly used in the literature. Second, after defining probability as a theory about a variable's potential frequency distribution, I argue that the dichotomy between frequentist and bayesian interpretations is illusory. I conclude with remarks about knowledge and determinism.
February 2018
Abstract (click to expand): Is economics a science? Answering this question is not only necessary for philosophical clarity, but also crucial for knowing how seriously to take economists’ claims and advice about public policy. Nevertheless, even among practitioners and academics, consensus is nonexistent. This paper resolves the conundrum in two steps. First I discuss some epistemology of science, defining clearly various concepts necessary to the debate. Several fallacies are clarified, such as "a theory may never be proved true, but only not falsified", "a model is useful because it simplifies reality" or "data mining is bad". In light of solid philosophical ground, I then discuss the practice and methodology of modern economics. The answer to the title question is a perhaps disappointing, but realistic, not yet. I conclude with prescriptions for a path towards a more scientific discipline.
