How these get made
Each experiment is cheap to make and fun to try. Claude, an AI model, writes the code and the first draft. I choose the questions and make the judgment calls. Then independent AI reviewers, which did not build the page, check the numbers against their sources and hunt for bugs. They find real mistakes, and we publish those too.
The lesson is not that AI builds things. Building is now cheap. Judgment about what is worth building, and the checking that makes it worth reading, are not.
-
India economy dashboard
What it teaches. How to read official statistics: base years, revisions, and what a monthly survey can and cannot tell you. Why the rupee and the RBI’s reserves belong on the same line. And a question it leaves for you: why can’t the RBI hold the rupee steady, set its own interest rates and let money move freely across borders, all at once?
Do better. Split inflation into food, fuel and the rest. Build a proper index of pressure on the rupee. Make a free, machine-readable nowcast of India’s growth this quarter. The full list has six projects.
The code · What we got wrong · Python, no installs
-
India’s Development Equation
What it teaches. How small changes in growth rates become large changes in dates. Why the World Bank’s scoreboard uses nominal dollars, not purchasing power. And that a rank is a claim about a sample: some of India’s recent climb up the table comes from richer countries dropping out of the data.
Do better. Add a purchasing-power view. Replace the single demographic year with the UN’s projections. Rank only countries with data in every year. The full list has five projects.
The code · What we got wrong · HTML and JavaScript, no installs
If you teach
Open a class with a live number. Ask students to explain this month’s inflation figure before they read the chapter that explains it. Give a group one item from a “Do better” list as a project. Or ask them to find our next mistake; the “What we got wrong” files show what that looks like.
If you build
Fork the repository, run it, and make it yours. The code is under the MIT licence and the words under CC BY 4.0, so you can reuse both for any purpose, with credit. The data belongs to its publishers (MoSPI, the RBI, the World Bank and others); follow their terms. If you do better, tell me: ashish@econforeverybody.com.