So, cryptography and physics have a joint endeavor: keeping all our devices, systems, and communications systems safe. To guarantee this, there are various standards that define how you must build entropy sources and random number generators for cryptographic use, such as the AIS 31 standard from the German BSI or the NIST SP800-90 suite from the National Institute of Standards and Technologies (NIST) in the US.
Quite interestingly, a very important aspect of these standards, and of randomness generation in general, is to assess the quality of such entropy sources. For many years, people relied on statistical tests to do so. The rationale was “if I pass the statistical tests, then it means my device is random”. However, this is fundamentally flawed. It was really interesting to see that, quite recently, NIST updated their own standard (one of the most relevant in the whole industry) to clarify that “they are rejecting its use [of the statistical tests] for assessing cryptographic random number generators”.