Delft experiment did not prove randomness

I posted about the Delft quantum experiment, but I skipped the randomness news. Here is ScienceDaily's description:

The experiment gives the strongest refutation to date of Albert Einstein's principle of "local realism," which says that the universe obeys laws, not chance, and that there is no communication faster than light.

As described in Hanson's group web the Delft experiment first "entangled" two electrons trapped inside two different diamond crystals, and then measured the electrons' orientations. In quantum theory entanglement is powerful and mysterious: mathematically the two electrons are described by a single "wave-function" that only specifies whether they agree or disagree, not which direction either spin points. In a mathematical sense, they lose their identities. "Local realism" attempts to explain the same phenomena with less mystery, saying that the particles must be pointing somewhere, we just don't know their directions until we measure them.

This is a pretty good description. The so-called (and misnamed) "local realism" seemed less mysterious to Einstein and others by assuming that electrons are little spinning particles with definite positions, momenta, and spins, even when we cannot see them. Quantum mechanics could then be seen as describing our uncertainty about reality.

We have known since about 1930 that this Einstein local realism is wrong. Quantum mechanics teaches that electrons have wave-like properties that prohibit such definiteness.

The Delft experiment is just the latest in 90 years of work supporting quantum principles. Part of the acclaim for this experiment is in its approach to randomness:

This amazing experiment called for extremely fast, unpredictable decisions about how to measure the electron orientations. If the measurements had been predictable, the electrons could have agreed in advance which way to point, simulating communications where there wasn't really any, a gap in the experimental proof known as a "loophole." To close this loophole, the Delft team turned to ICFO, who hold the record for the fastest quantum random number generators. ICFO designed a pair of "quantum dice" for the experiment: a special version of their patented random number generation technology, including very fast "randomness extraction" electronics. This produced one extremely pure random bit for each measurement made in the Delft experiment. The bits were produced in about 100 ns, the time it takes light to travel just 30 meters, not nearly enough time for the electrons to communicate. "Delft asked us to go beyond the state of the art in random number generation. Never before has an experiment required such good random numbers in such a short time." Says Carlos Abellán, a PhD student at ICFO and a co-author of the Delft study.

This is quackery, as there is no proof that there is any such thing as randomness. I have explained it on this blog many times, such as here and in my 2015 FQXi essay.

For the ICFO team, the participation in the Delft experiment was more than a chance to contribute to fundamental physics. Prof. Morgan Mitchell comments: "Working on this experiment pushed us to develop technologies that we can now apply to improve communications security and high-performance computing, other areas that require high-speed and high-quality random numbers."

If you want high-speed and high-quality random numbers, just toss a coin 512 times, and apply SHA-512 repeatedly. There is no known way to do better than that.

With the help of ICFO's quantum random number generators, the Delft experiment gives a nearly perfect disproof of Einstein's world-view, in which "nothing travels faster than light" and "God does not play dice." At least one of these statements must be wrong. The laws that govern the Universe may indeed be a throw of the dice.

This is confusing. They have disproved Einstein's view of local realism being hidden variables. They certainly have not found anything going faster than light, and they have not proved that God plays dice. They used a random number generator to get results that look random. Such an experiment cannot possible prove the existence of random numbers.