[Lyrrad]

[ QUOTE ]
im now happy that thermal noise can be a good source of entropy.

but.. lets think about the implementation we're discussing here for a minute... if the thermal noise that pokerstars was using was guaranteed to be completely random then there'd be no point in hashing it with the mousemovement data and other user input to make it somehow EVEN MORE random, dont you agree with that logic? so there must be some reason why theyre doing that that means they dont entirely trust the thermal noise entropy on its own.

they also say on their web site that they utilise a PRNG as well.. so its not clear why they'd do that unless they were somehow not getting the results (performance? quality? unclear what) out of their thermal noise and mousemovements generators. dont you agree that there must be some reason why they do that, that they havnt explained?

im not being paranoid, im just being inquisitive and playing devil's advocate.

the whole thrust of my posts, which no one seems to discuss or question, is that a high quality prng thats well implemented is going to shuffle cards extremely well and as well as a claimed truly random source.. and that no one has visibility into the cards that are shuffled, with either method. thus you shouldnt somehow trust a trng any more than a prng when it comes to a shuffled deck. it makes not the slightest bit of difference to the shuffled deck or the end users who are playing cards wth that deck.


[/ QUOTE ]

Why use two sources of random data? If one of the sources was compromised, the shuffle will still be completely random. Essentially they do the following: random data + random data = random data. However, if some data is slightly not random, then it's random data + non-random data = random data. Essentially a property of random data is that you can combine it with any other data, random or non-random to get random data up to the length of the original random data. (As long as the random data source is from an independent source of the other data)

Adding good pseudo-random sources (hash and stream) makes it so that even if the true randomness breaks down, then there is still some level of unpredictability.

I'd assume that PokerStars has some tools to measure the randomness of the sources and monitors it regularly.

Why PRNGs aren't used exclusively: With TRNG sources and a proper method, even if you know how the cards are generated, it's impossible to predict what cards are going to come next. Without a TRNG in use, then it is theoretically possible to predict the cards if you know the algoritm.