https://www.fernandomartel.com/posts/Recent content in Posts on Fernando Martel GarcíaHugo -- gohugo.ioen-usWed, 25 Feb 2026 17:32:14 -0500https://www.fernandomartel.com/posts/automated-policy-evaluation/Wed, 25 Feb 2026 17:32:14 -0500https://www.fernandomartel.com/posts/automated-policy-evaluation/Today I came across Project APE, for automated policy evaluation (APE). The project uses Claude Code plus public observational datasets to generate end-to-end economics-style policy impact papers, from idea selection and data pulls to estimation, writeup, review, and reproducibility checks. I am both intrigued and baffled by the project. Baffled because Project APE is perhaps aping (pun intended) current methods too much. The process of evaluating policies one journal paper at a time is incredibly inefficient.https://www.fernandomartel.com/posts/microsoft-ai-capital-allocation/Wed, 04 Feb 2026 11:15:23 -0500https://www.fernandomartel.com/posts/microsoft-ai-capital-allocation/Microsoft’s earnings report on January 28th beat market expectations yet the stock fell more than 15% soon after. Wall Street was spooked by two revelations in the earnings. First, 45% of Azure’s Remaining Performance Obligations come from OpenAI, a company with no clear path to profitability. Second, Microsoft is allocating more compute to various first party Copilot providers and internal R&D, at the expense of Azure customers. The OpenAi concentration risk does not seem so troubling in a context where compute is supply-constrained.https://www.fernandomartel.com/posts/fermi-app/Fri, 28 Nov 2025 00:00:00 +0000https://www.fernandomartel.com/posts/fermi-app/Background and motivation The decisions that matter most to a business are often the hardest to analyze. Whether to acquire a competitor, enter a new market, or bet on a risky product line, these are all strategic choices.1 They commit large resources, shape the firm’s trajectory for years, and are not the sort of thing you can A/B test and optimize easily. Even in tech-savvy, data-driven companies, strategic decisions often fall outside the domain of the data science organization.https://www.fernandomartel.com/posts/the-hidden-cost-of-free/Wed, 16 Jul 2025 00:00:00 +0000https://www.fernandomartel.com/posts/the-hidden-cost-of-free/Many Fortune 500 companies unknowingly undermine their profitability by treating owned acquisition channels, such as homepages, or in-product promotions, as “free” resources. This practice results in inefficient resource allocation and discourages effective targeting, as teams have little incentive to optimize when advertising comes at no cost. By failing to account for opportunity costs and economic margins, businesses risk misallocating valuable assets and missing opportunities for growth. This post explores how understanding opportunity costs and economic margins can drive better targeting decisions, maximize the yield on owned channels, and unlock long-term profitability.https://www.fernandomartel.com/posts/mundane-delightful/Tue, 11 Feb 2025 00:00:00 +0000https://www.fernandomartel.com/posts/mundane-delightful/The long tail of undone tasks and unmet needs A previous experiment showed many repetitive tasks could be automated. Today’s experiment is simple—mundane even—but it highlights how AI does more than automate repetitive tasks; it enables the completion of previously impossible ones. These are tasks where the effort to learn them outweighs the benefits and where no third-party solution exists. Often, these tasks remain undone, resulting in an exceptionally long tail of incomplete tasks and unmet needs.https://www.fernandomartel.com/posts/racing-towards-the-future-how-ai-is-transforming-scientific-inquiry/Wed, 22 Jan 2025 00:00:00 +0000https://www.fernandomartel.com/posts/racing-towards-the-future-how-ai-is-transforming-scientific-inquiry/I recently bought a second hand copy of A Handbook of Small Data Sets, a lovely little book packed with 510 data sets. Many are so small they can be printed on half a page, as shown below: Source: Handbook of Small Data Sets, p. 36 As I was lounging on my sofa, flipping through the book, I had the idea to snap a picture of the above dataset with my Android phone and analyze it with ChatGPT 4o.https://www.fernandomartel.com/posts/political-science-as-problem-solving/Tue, 31 Jan 2023 00:00:00 +0000https://www.fernandomartel.com/posts/political-science-as-problem-solving/I recently came across a manuscript titled “Methodologies for ‘Political Science as Problem Solving’” via Twitter: I have a new "methodology big think" essay on the "problem solving" approach to social science. More here, along with a link to a PDF of the essay. I would love to hear what people think.https://t.co/vJI1Looaml — Cyrus Samii (@cdsamii) January 21, 2023 The manuscript is very well written, and a joy to read. The goal of the paper is to explain what “Political Science as Problem Solving (PSPS)” is, and how it is done (including carefully curated references to the most relevant and recent methodological advances for this purpose).https://www.fernandomartel.com/posts/maximizing-okrs-balancing-ambition-and-reality/Sun, 15 Jan 2023 00:00:00 +0000https://www.fernandomartel.com/posts/maximizing-okrs-balancing-ambition-and-reality/[This blog post has been co-authored with ChatGPT and DALL•E 2. Please see below for a discussion of how this process went.] In today’s fast-paced business world, many companies are looking for ways to stay competitive and achieve their goals. One approach that has gained popularity in recent years is management by objectives, specifically Objectives and Key Results, or OKRs. OKRs are a way for organizations to set and track their goals and objectives.https://www.fernandomartel.com/posts/organizational-maturity-and-digital-transformation/Sat, 28 Mar 2020 00:00:00 +0000https://www.fernandomartel.com/posts/organizational-maturity-and-digital-transformation/Brick-and-mortar, Fortune 2000 corporations undertaking “digital transformations” face a common problem: Where to start? I will get to that in a moment but first, semantics. What is digital transformation Different corporations mean different things by digital transformation. Yet, when all is said and done, we can boil it down to a two step iterative process, as outlined below: The Digital Transformation Redux Increase direct sales through digital channels; and Enable a “flywheel effect”, where the flywheel effect involves combining digital data, other data, and “artificial intelligence” to provide customers with “what they want, when they want it”.https://www.fernandomartel.com/posts/stratified-social-distancing/Sun, 15 Mar 2020 00:00:00 +0000https://www.fernandomartel.com/posts/stratified-social-distancing/Consider the following objectives for dealing with COVID-19: Achieve herd immunity as quickly as possible (e.g. something like 60 percent or more of the population resistant after infection); Minimize number of death; and Minimize economic disruption. Strategy to meet objectives If so, one possible strategy to achieve those aims is stratified social distancing. Here are the 3 simple rules that would govern this option: If you are >=45 years old, then maximize social distance, hand washing, etc.https://www.fernandomartel.com/posts/automation-without-tears/Tue, 11 Jun 2019 00:00:00 +0000https://www.fernandomartel.com/posts/automation-without-tears/Recent advances in Machine Learning — Artificial Intelligence for the marketing folk — are all over the news. Dreary headlines like “A.I. Expert Says Automation Could Replace 40% of Jobs in 15 Years” have employees fearing for their jobs. Meanwhile, Wall Street analysts and businessmen (yes, they are mostly men) sharpen their pencils — and knives — as they prepare to make millions of dollars through massive job cuts. These existential fears and profit calculation are not only wrong but dangerous — to employees, customers, and the bottom line.https://www.fernandomartel.com/posts/beware-the-propensity-score/Wed, 27 Jun 2018 00:00:00 +0000https://www.fernandomartel.com/posts/beware-the-propensity-score/Propensity score matching (PSM) has become a popular workhorse in observational causal inference ever since its introduction by Rosenbaum and Rubin in 1983. However, a recent manuscript by King and Nielsen argues that PSM increases “imbalance, inefficiency, model dependence, and bias”. In their view, propensity scores should not be used for matching. This is a strong, important, and potentially very troubling claim. My purpose is to reframe their claims in the modern language of directed acyclic graphs.https://www.fernandomartel.com/posts/10000-missed-opportunities-a-year/Wed, 21 Mar 2018 00:00:00 +0000https://www.fernandomartel.com/posts/10000-missed-opportunities-a-year/Today, most large internet companies run 10,000 or more experiments a year to improve things like user engagement, ad conversions, and the like. In comparison, how many experiments are done each year in the US studying the effectiveness of the elementary math curriculum? Is this satisfactory, considering the US has some 90,000 primary schools, and 33 million elementary students? Arguably, the present status quo in testing and innovation is simply immoral.https://www.fernandomartel.com/posts/__001/Mon, 29 Aug 2016 00:00:00 +0000https://www.fernandomartel.com/posts/__001/On Friday, December 13, 1799 George Washington awoke with a severe sore throat. He was visited the next day by three doctors, who quickly prescribed bloodletting: The withdrawal of blood to restore “humoral balance”. Within hours they removed half his blood. By evening, George Washington was dead. What I find most shocking about Mr. Washington’s untimely demise is how a practice so harmful to patients remained the most common surgical procedure for almost 2,000 years – from the time of the Pharaohs, to the end of the 19th century.