3 ms·

I wonder if a provably unbiased distributed gambling application cannot be made so as to guard against the potential for such attacks. Generating random number

I wonder if a provably unbiased distributed gambling application cannot be made so as to guard against the potential for such attacks. Generating random numbers in a distributed and uncheatable way is already possible, I wonder what is holding this back.

patio11 18y ago
The problem isn't bias -- the problem is one actor having more information than the others, earlier than the others. There is no evidence that the RNG was anything other than fair. A related attack on poker is just stuffing a game with colluding players. You don't have to know player A's cards to beat him in the long run if you know the cards of B, C, D, and E (and, if you want to get really tricky -- they can know yours, too!). Plus, given that this is online poker, you can have players B, C, D, and E be bots, programmed to talk to your server to get the odds, throw enough hands to look reasonable, and bet big only once in a while when their collusion means The Syndicate rakes in the moolah. Example: the flop is 5, 6, 7 with 3 different suits. The best hand from The Syndicate is pocket 7s, which is (poker amateur hour commences) pretty strong but not unbeatable. Everybody is out but one player. The next card is shown, and it is a 3. There are two worlds the game could be in now: the world where there are fours available, in which case the adversary might have one. Then there is the world with no fours available, in which case The Syndicate has the best possible hand and will only lose by a freak accident. The adversary goes all in. The Syndicate checks with the other TEN cards from the deck they saw, notes that they've seen three of the 4s already, and calls, with a smile on his face. (Again, amateur poker player but I'm thinking this is a 95%+ win for the syndicate.)
- alexandros 18y ago
  Your point is valid, translating existing games which are built on a set of assumptions (physical presence -> unique non-communicating players, physical deck + shuffling -> unique cards and true randomness) may not be translatable to an online experience in a sustainable manner. I was refering to the general problem of not being able to trust the single point of control, which applies to all types of online gambling. In this context, I was wondering why games that can be gambled on have not appeared that work in a fully distributed manner.
  - njharman 18y ago
    Because it doesn't solve the cheating described in the post you replied to. Distribute all you want but if 5 of 6 players are colluding player #6 is gonna be screwed.
    - alexandros 18y ago
      there is a way to derrive an unbiased random number, even if everyone else is cheating. take this: http://everything2.com/title/Coin%2520tossing%2520protocol http://everything2.com/title/Coin%2520tossing%2520protocol and extend for N participants and results from 1 to N. Given this, a lot can be done.

patio11 18y ago

The problem isn't bias -- the problem is one actor having more information than the others, earlier than the others. There is no evidence that the RNG was anything other than fair. A related attack on poker is just stuffing a game with colluding players. You don't have to know player A's cards to beat him in the long run if you know the cards of B, C, D, and E (and, if you want to get really tricky -- they can know yours, too!). Plus, given that this is online poker, you can have players B, C, D, and E be bots, programmed to talk to your server to get the odds, throw enough hands to look reasonable, and bet big only once in a while when their collusion means The Syndicate rakes in the moolah. Example: the flop is 5, 6, 7 with 3 different suits. The best hand from The Syndicate is pocket 7s, which is (poker amateur hour commences) pretty strong but not unbeatable. Everybody is out but one player. The next card is shown, and it is a 3. There are two worlds the game could be in now: the world where there are fours available, in which case the adversary might have one. Then there is the world with no fours available, in which case The Syndicate has the best possible hand and will only lose by a freak accident. The adversary goes all in. The Syndicate checks with the other TEN cards from the deck they saw, notes that they've seen three of the 4s already, and calls, with a smile on his face. (Again, amateur poker player but I'm thinking this is a 95%+ win for the syndicate.)

alexandros 18y ago
Your point is valid, translating existing games which are built on a set of assumptions (physical presence -> unique non-communicating players, physical deck + shuffling -> unique cards and true randomness) may not be translatable to an online experience in a sustainable manner. I was refering to the general problem of not being able to trust the single point of control, which applies to all types of online gambling. In this context, I was wondering why games that can be gambled on have not appeared that work in a fully distributed manner.
- njharman 18y ago
  Because it doesn't solve the cheating described in the post you replied to. Distribute all you want but if 5 of 6 players are colluding player #6 is gonna be screwed.
  - alexandros 18y ago
    there is a way to derrive an unbiased random number, even if everyone else is cheating. take this: http://everything2.com/title/Coin%2520tossing%2520protocol http://everything2.com/title/Coin%2520tossing%2520protocol and extend for N participants and results from 1 to N. Given this, a lot can be done.