The World Cup Memory Machine is an independent attempt to tell ninety-six years of the men's World Cup — 1930 to 2026 — as data, story and play. This page explains who built it, why it is the way it is, where the numbers come from, and how current they are.
It is built and maintained by Oriol Cervantes Grau, an economist and decision-intelligence practitioner whose background runs through international organisations including UNDP, UNICEF and the WHO. The project is a portfolio of interactive data storytelling — an argument that rigour and play are not opposites.
Nothing here is affiliated with FIFA or any governing body. It is one person's reading of the public record, with every claim labelled by how confident that reading is.
Every page is a single self-contained file of vanilla HTML, CSS and JavaScript. No frameworks, no build step, no libraries — not even for the 3D penalty engine or the animated charts. That is a deliberate decision, not a limitation.
No framework to deprecate, no npm tree to rot. A single HTML file opens in any browser, forever — the same property that makes archival web pages survive.
Nothing to download but the page itself. No 300KB of framework before a single pixel renders.
When the work is the portfolio, the craft has to be legible. Nothing hides behind a framework — the structure, the interactions and the editorial calls are the author's own.
Each module is a standalone artefact. It can be embedded, archived, or handed over with no setup.
Different parts of the project age differently, so they are handled differently — and labelled so you always know which kind of number you are looking at.
Results, scorers, hosts and the long economic record are fixed history. They are sourced once, checked, and do not change. Projections to 2050 are clearly marked as modelled.
Every figure is checked against FIFA's own accounts and primary sources, and carries a confidence badge. Contested numbers — like the 2026 carbon footprint — are shown as a range, never as fake-precise single facts.
The project's core discipline is refusing to average things that are not the same. A headline cost that is really national infrastructure is not compared as if it were a stadium bill; an estimate is not dressed up as an official figure. Four tiers run across the whole machine:
FIFA budgets, council decisions, government and watchdog figures. Usable as ground truth.
Media or academic figures, always labelled. Ranges where sources genuinely disagree.
Interactive calculations and trajectory models. They show "what if", not "what is".
Live-layer data, dated. Accurate as of its timestamp, not a permanent claim.
The economic and current-tournament layers draw on primary documents and clearly-attributed reporting. The main ones:
The same discipline that shows a carbon estimate as an estimate applies to how this was built. Two honest notes:
The atmospheric portraits, stadiums and clips are AI-generated illustrations, used as mood and texture. They are never presented as archival footage — the data and the analysis are the substance.
Every page is vanilla HTML/CSS/JS written line by line, developed with AI coding assistants. No framework does the hard part; the architecture and the editorial choices are the author's.
The clearest single expression of this method is The Ledger — who really pays for the World Cup, with every figure sourced and labelled.
Open The Ledger →