Objectives¶

Adversaries in the data garden are not here for a leisurely stroll. Exposure comes in three kinds, and the distinction is not academic: a determined one can win without ever learning a name. The defences differ by kind, and a release that survives one can still fall to another.

Identity disclosure¶

The prize of the whole dig: matching a pseudonymous record to an actual person. This is the moment an adversary looks at a string of identifiers and says, “Ah yes, that is Bob from Amsterdam.” Once identity is out, privacy stops being theoretical and starts being personal.

Content disclosure¶

Not who a person is, but what is growing in their patch. Content disclosure is the unmasking of the sensitive attribute itself: a medical condition, a political leaning, sexuality, religion, a membership, alongside the flatter kinds of damage like an identity document or a bank detail. It needs no identification at all. Learn that everyone in a small group shares a diagnosis and the diagnosis is out for all of them at once, the names never required.

Link disclosure¶

The roots rather than the flower: who talks to whom, who buys what, who turns up where. Connections give the game away even when identities stay technically hidden, because a network’s shape is its own signature. Recognise the pattern and the label on the node is a formality.

Last reviewed: 2026-07-17.