What is the "core promise" of differential privacy?

The outcome of an analysis stays essentially the same whether or not any single individual participates — so participation cannot expose you.

The formal promise (from Dwork & Roth's The Algorithmic Foundations of Differential Privacy): "the outcome of a survey will stay the same whether or not you participate in it." Your individual data can't be exposed, because the results would be statistically identical with or without your contribution.

This is why DP gives such strong protection: an attacker comparing the output "with Bob" versus "without Bob" learns nothing distinguishable about Bob. It holds even against adversaries with arbitrary background knowledge — the property older models (k-anonymity, l-diversity) can't guarantee.

Tip: If your presence in the dataset doesn't measurably change any output, there's nothing about you to leak. That's the whole idea.

Go deeper:

• The Algorithmic Foundations of Differential Privacy (Dwork & Roth, PDF) — the source of the "outcome stays the same whether or not you participate" statement.