This page introduces the methodology section and links to reproducible demonstrations of identifier graphs, behavioural synchrony, and browser-fingerprint identification.
This section contains concrete experiments and reproducible demonstrations showing what current best-practice methodology can and cannot do. Each investigation is written as a self-contained piece with code or data notes attached where possible.
What this section covers
- Identifier graphs and behavioural synchrony — a synthetic experiment showing why shared identifiers and coordinated behaviour reveal different kinds of account clusters, and why both methods need false-positive controls.
- How vendors turn fingerprints into stable IDs — a synthetic experiment reconstructing browser-fingerprint identification at four levels of sophistication, showing why the “accuracy” vendors publish is only one of two error modes, and why cloned mobile devices set a hard floor on what fingerprinting can achieve.
Code
The supporting code for the first investigation is in src/openbotrisk/experiments/identifier_synchrony/. The second investigation’s code is in src/openbotrisk/experiments/fingerprint_identification/. Both are written as plain Python scripts rather than notebooks so the website pages can stay narrative while the figures remain reproducible.