It's true. However we can evaluate the probability of it happening in a given set of rolls. Definitely some funniness with the statistics of the rng. But it appears to be a known behavior. The real requirement is that the values cannot be predicted. We'll see if there's any flaw in the current setup. Possibly the formula shows that every 10 blocks the rolls within may be correlated in some way. Especially since the only different numbers are block num + client hash (controlled by user). Not for me to break though, I'm terrible at this kind of thing.
Posted using Partiko Android