Isn't it a bad implementation though? They generate the 30 initial numbers, without your client seed, and they can generate what ever they want, and you can't verify that they cheated with the inital generation. So while it is technically provably fair, because of how the initial shuffle is generated, they could create a higher house edge by predicting what the gambler likes to do (ie over 7) and generate the inital deck so it is more likely to get under 7? They should just get rid of the initial generation and play with a fair deck (5 ones, 5 twos, 5 threes, e.t.c)
I can't render an opinion on Pocketdice, but it looks like NLNico and RHavar have seen obvious defects which would preclude it from being provably fair.
The way I see it, "provably fair" (the concept) is marketed by casino operators with several claims. Here's a sample (emphasis added in
bold):
Sorry to hear that you are skeptical about the site.
All of our games are provably fair. Therefore, it is impossible for us to change the outcome depending on the bet amount. Big bets and small bets have the same chance of winning.
We cannot influence the odds of any wagers you place in bitZino, other than simply changing the rules to the game in question, which you will know (for example, if we paid out 6 to 5 on a blackjack instead of 3 to 2, you'd be able to easily notice this).
Provably fair systems allow players to independently verify every single wager they make - usually immediately after the wager is complete. Effectively, this is a zero-trust system: players don't have to trust third-party licensing providers to be confident they're getting a fair game.
When presented with a viable weakness, however, casino operators seem to resort to a promise (which requires trust) or marketing (emphasis commentary added in
bold):
Thanks for brining this to our attention.
We absolutely always use a random number for every single game, and would never even consider cheating our loyal customers.
(trust me)After reviewing the post in question, I don't think this is an even accurate criticism anyhow.
It would take 17+ hours to generate a single seed to influence a single round of play, but our casino allows you to play as many games per second as you would like.
(didn't read the method correctly)For example, in May there were more than 25,000,000 games played on our site. That works out to about ten games per second.
Regardless, our casino has much much better odds than anything in Vegas.
(marketing)Cheers
...you are right about our Mersenne Twister (MT) truncates to 32-bits. However, our dealer shuffle doesn't just use MT and Fisher-Yates, it also uses the Java RNG and therefore the process as a whole has enough randomness. (trust me)
From my perspective, if the provably fair system isn't "provable" then it can only degenerate into a simple promise of fairness. A promise requires that you trust the casino, which is contradictory to the casino's claims.
Many of these casinos have been around for some time, so they tend to be a little dismissive when an attack is presented. It's natural: they have established a base of players, some of whom might play regardless of provable fairness, some who don't verify anymore, some who may amplify claims, and so on. And they've likely received more than their fair share of claims from players about cheating.
Truly interesting responses. Thanks for the question, DarkStar_!