Strong encryption is useless if the key can be guessed. And a key can only be as unpredictable as the random numbers used to make it.
Good keys need good randomness
Encryption keys must be unpredictable. If an attacker can work out how a key was generated, even the strongest algorithm won’t help. Computers are designed to be predictable, so producing truly unpredictable numbers is surprisingly hard.
When randomness goes wrong
- ◆Debian OpenSSL (2008): a code change made in 2006 left Debian and Ubuntu systems generating keys from only about 32,000 possibilities. Every affected key had to be replaced.
- ◆Shared primes (2012): researchers scanning the internet were able to compute the private keys of about 0.5% of the HTTPS servers they studied, mostly small devices with poor randomness at start-up.
In both cases the maths was fine; the random numbers were not.
Two ways to make random numbers
Pseudo-random number generators use maths to stretch a small random “seed” into a long stream of random-looking numbers. They are fast and, with a good seed, secure. True random number generators measure an unpredictable physical process, such as electrical noise, to create that seed.
Quantum random number generators (QRNGs) are true random number generators that measure quantum events, such as whether a single photon passes through or bounces off a half-silvered mirror. According to quantum physics, the outcome is fundamentally unpredictable, not just hard to predict.
How randomness is checked
NIST publishes standards for random number generation, including SP 800-90B, which describes how to test entropy sources. Good hardware, quantum or not, should be tested against these standards and monitored continuously, because a faulty source can fail silently.
Do you need a QRNG?
Most modern operating systems and processors already include good random number generators. QRNGs are most useful where randomness must be demonstrably high quality: hardware security modules, key-generation services, lotteries, scientific simulations and small devices that struggle to gather randomness on their own. QRNG chips are now small enough to fit in phones; some Samsung Galaxy Quantum models sold in South Korea include one.
A QRNG does not protect against quantum computers on its own. It makes keys harder to guess, while post-quantum algorithms protect how those keys are used.
task_altKey takeaways
- check_circleWeak randomness has broken real systems, even when the algorithms were strong.
- check_circleQRNGs use quantum events that are fundamentally unpredictable.
- check_circleRandomness sources should be tested against standards such as NIST SP 800-90B.