I have also used the diehard suite a while back to convince myself that a random mixer thing I came up with was suitably random. Basically I wrote a thing to read from various sources of entropy, bleach them a bit and then mix them all together in a way that was extremely chaotic before feeding them into the kernel entropy pool. This was because it used to be a problem (back circa 1998 or so) that if you had headless servers serving a lot of ssl sessions they would run out of entropy and block from time to time. At the time, diehard had been implemented in fortran and then compiled to C using a gnu fortran transpiler so it was completely impossible to understand what each of the tests were actually doing.
On the topic of tfa, one of the most practical “RNG in your brain” techniques I heard of was suggested by the poker player Dan Harrington in his books, which is for sampling at random with some probability between some discrete choices. So say you have something you want to do at random 1/3rd of the time and the other 2/3rds of the time you’re going to do another thing[1], the way he says to do this is base it on your watch face. So you decide [2] that if the second hand of your watch is between 1 and 20 then you will do thing A and if it’s between 20 and 60 you’ll do thing B. Then you look and you do whatever thing. The good part about this is clearly even if people know you’re doing it they can’t exploit it because they don’t know the set of choices you’ve given yourself, what the boundaries are where you choose between strategies etc. It doesn’t work if you need to sample rapidly but in the context he gave it (tournament poker) you would have a random amount of time more than a minute between decisions anyway so its fine in that context.
[1] From a game theory point of view, it can be shown that in games with randomness and hidden information like poker, the optimal strategy is a Nash equilibrium in mixed strategies which means you need to sample a probability distribution to choose your actions in this way.
[2] before looking. This is important.