The general scenario is as follows: from a large number of items, a few are to be tested for quality control. This occurs in manufacturing processes, commercial transactions, agricultural inspection, environmental testing, and many other scenarios.
Sometimes the test destroys the item tested, other times the test is very expensive per item tested. How do we decide which few items are to be tested? The usual method is "pick the items at random".For this we need random numbers. How, and by whom, are these numbers to be generated?
Consider the following scenario. A manufacturing plant produces a particular component for, say, computers. The components have serial numbers. Under certain conditions (e.g. when the plant's environmental control systems are not functioning properly) the number of defective parts increases. Complex correlations (or undetected tampering), with the random number generator used for sampling can lead to erroneous estimates of the percentage of faulty parts. Uncertainty about these factors is often addressed by increasing the frequency and complexity of quality control mechanisms. The NIST Randomness Beacon provides a much simpler solution to this problem.
Scenarios analogous to the above occur in many settings.
Industrial sabotage: a bottle of a certain over-the-counter medication is tampered with (for example, the pills could be replaced by counterfeit pills). If the saboteur can get that bottle tested as part of a "random testing regime" then the manufacturer could incur massive losses.
Clandestine Protectionism: a shipment of fruit arrives at a foreign port. The local government tests one crate and finds it to be infected with a dangerous pest. Unloading of the cargo is prevented and the fruit rots. International institutions are informed, and a temporary embargo is issued. Subsequent testing of the shipment finds no other infected fruit. The local government admits that it was tipped by an anonymous caller to test that particular crate. Cases similar to this have happened in the past. Partly in response to this, international trade organizations are revising the protocols that lead to embargo. A trusted, time-stamped, and authenticated source of randomness could potentially allow the various interested parties to distinguish between cargo tested as part of random sampling and cargo tested "for other reasons". Protocols would then be able to take into account such information so as to contain the damage to consumers and producers.
Environmental testing: water samples from a river are to be collected and tested for contamination. Who decides where and when to collect the samples? If the time and location of water sampling is influenced by environmental conditions or by tampering with the random number generator, then faulty readings can lead to underestimates or overestimates of contamination levels. This can lead to undetected environmental damage or cause unwarranted financial loss to industrial plants that use the river.
Return to NIST Randomness Beacon page