Skip to content

Cryptographic entropy: the foundation of digital security

Cryptographic entropy is the measurable unpredictability that every key, nonce and signature depends on. It is a quantity, not a synonym for randomness — and the useful question is not whether a source has it but what evidence exists that it still does, in the system you are running now.

34 articles in 7 sections

Every cryptographic system rests on an assumption that is almost never examined: that the numbers underneath it were unpredictable. The algorithms are public, analysed and generally sound. The entropy they consume is usually taken on trust.

Entropy is a measurement, not a property

Saying a source "has entropy" is like saying a component "has a tolerance". The statement only means something once you say how much, measured how, and conditional on what assumptions. Min-entropy is the measure that matters for cryptography, and an estimate of it is always relative to a model of the source.

That conditionality is not a weakness in the argument. It is the argument. A vendor who states what their estimate depends on has written the model down; one who quotes a bare number has not.

Assurance is the combination of three things

A physical model of where unpredictability originates. Measurement of the quantities that model depends on, while the system runs. And evidence a third party can inspect. Any two without the third leaves a gap: a model without measurement describes the device on the day it was tested, and measurement without evidence is a claim.

Entropy assurance is the name for all three together.

Why local, and why verified

Quantum physics is how the entropy is made. Local and verified are what actually gets bought. Local, because entropy that crosses a trust boundary has to be trusted across it. Verified, because a source that cannot report its own health is a source you are assuming about. That is the whole of LOVE.

Where this pillar leads

Into hardware, through the entropy dependency in key generation. Into post-quantum migration, because new algorithms consume randomness exactly as the old ones did. And into audit, where the questions stop being technical and start being procurement — which is where most readers arrive with a deadline.

Entropy fundamentals

Entropy as a measurable quantity rather than a synonym for randomness, and how much of it cryptography actually consumes.

Assurance and measurement

How min-entropy is measured, what a health test can and cannot detect, and how a source drifts away from its own model.

Locally Verified Entropy

Generating entropy inside the boundary that consumes it, and having the source report its own health while it runs.

Where entropy breaks

Weak randomness in the wild, entropy starvation at boot and in virtual machines, and where a certified module gets its entropy.