Depends on the reader. For combat stats, typically 100 events is considered bare minimum, and normally 1000 is better. Shroud's logs don't have all the necessary information, currently, though, for all tests. For crafting & harvesting events, it's complicated due to there not being any log entries that show what the percentage was, so you have to video everything, which isn't pleasant. There are fail/success entries for crafting, as long as you know the percentage, that might be ok. However, again, it comes down to the perception of the reader. Some readers will accept 10 samples as indicative, because the cost to perform those 10 attempts is ridiculous. Some readers will demand more than 100 samples, simply because they don't want to believe, and/or are staunch supporters of "whatever", regardless of the cost to the tester. I've seen in some games where you can submit logs for analysis, and they add them to a pool, all in an effort to get the dev team to admit there may be an issue, or even just to review the server side logs to see if there may be an entropy issue. So, with that segue.. "A lack of entropy can have a negative impact on performance and security." (in Shroud's context, security would mean " a good RNG " ) Predictable, repeating, deterministic, 'stretches', 'runs', and similar RNG problems have plagued MMO's since MMO's have existed. One thing is for certain, the more players you have all using the same RNG pool, seed, or similar entropic source? The worse the problem becomes. There were times in EQ1 when PvE combat could be predicted depending on the time of day, because their original RNG seed was time based, and with thousands of concurrent players, became deterministic. If someone is in the 92-95% success range and is seeing more than 5 failures in a row... that's not good, regardless of the underlying RNG.