How to design a blockchain-based provably fair online dice game
In the old times, when we play a traditional dice game, the game fairness is usually endorsed by the game owner's reputation. If an extremely unlucky player loses 10 times in a row, there is no way for him to tell if he is just being unlucky or not.
With the help of the blockchain technology, now we have a way to design an online provably fair dice game technically. Blockchain is a transparent, tamper-resistant online ledger that every one can audit.
We are glad to see dozens of online dice games popping up in the EOS community recently. It's a pity that most of them are not open sourced nor rig resistant, still leaving inapparent potential backdoors for rigged betting, which is risky and can be largely improved. However, these close sourced dice games did help many rookies embrace the blockchain and get to know the beauty of the EOS world. Their efforts should not be forgotten and we appreciate their value added to the EOS ecology.
Hereby our DappPub team would like to share the design for our open sourced FairDice game and enlighten those who are interested to unleash the power of DApps together.
1. Randomness is the key
The key point is that the game has to provide a kind of combined randomness that no single party could control alone.
| "How can I generate random numbers inside a smart contract?" |
|---|
 |
Just like what BM answered in the above question on StackExchange, for a two-party (player-house) dice game, the key point to provide fairness or rig-resistant randomness requires a 3 step process:
Both parties (
player&house) commit hashes of their seeds:
hash(player_seed)&hash(house_seed)Both parties reveal their
hash(player_seed)&hash(house_seed)
Roll=hash(player_seed + house_seed)
The final Roll is random and it's a kind of combined randomness based on both house && player 's input (house_seed & player_seed)
2. Sequence Diagram: a theoretical random dice roll process
Based on BM's answer, we can draw a simple sequence diagram for a theoretical random dice roll process as below:
| Sequence Diagram: a theoretical random dice roll process |
|---|
 |
In step 2, the house creates a new house_seed for each bet and keeps it a secret until step 10.
In step 3, the house calculates the corresponding house_hash = hash(house_seed).
In step 4, the house uploads the house_hash to the blockchain. At this moment, the player still has no idea about what the house_seed is, but he is sure that the house_seed will not change during this bet because he can use the house_hash to verify after the bet. Metaphorically speaking, house_hash is like a house's dark card placed on the table.
In step 5, the player also creates his player_seed.
In step 6, the player calculates the corresponding player_hash = hash(player_seed).
In step 7, the player uploads player_hash to the blockchain. At this moment, the house still has no clue about what player_seed is, but he is sure that the player_seed will not change during this bet because he can use the player_hash to verify later. Metaphorically speaking, player_hash is like a player's dark card placed on the table.
In step 8, the player uploads his player_seed to the blockchain.
In step 9, the open-sourced smart contract deployed on the blockchain verifies the player_seed & player_hash to see if it's a match.
In step 10, the house uploads his house_seed to the blockchain.
In step 11, the smart contract verifies the house_seed & house_hash to see if it's a match.
In step 12, the smart contract calculates the hash(player_seed, house_seed) as the final roll number.
3. Sequence Diagram: optimized "FairDice" random dice roll process
Now we can give it a shot to start building our own blockchain-powered dice game. Compared with other blockchain infrastructure, EOS is preferred here as it's fast and provides good user experience.
In fact, to optimize the whole process, it's unnecessary for player to upload both player_hash & player_seed in two steps.
In other words, since the house has already placed his dark card (house_hash) on the table and not able to change it during the bet, the player could just place his open card (player_seed) on the table directly, instead of placing his dark card (player_hash) on the table and flipping the card (player_seed) in two steps.
Therefore, we have optimized the whole process and improved user experience as below:
| Sequence Diagram: optimized "FairDice" random dice roll process |
|---|
 |
In step 1, since a "referral bonus" is introduced in the game, the referrer along with theROLL under to win will be sent to the house.
In step 2, the house creates a new house_seed for each bet and keeps it a secret until step 9.
In step 3, the house calculates the corresponding house_hash = hash(house_seed).
In step 4, the house creates house_sign = sign(ROLL_under, hash(house_seed), expiration_time, referrer).
In step 5, the house sends house_hash, house_sign, ... to the player.
In step 6, the player creates his player_seed.
In step 7, the player uploads the house_sign & player_seed to the EOS blockchain.
In step 8, the smart contract deployed on EOS checks to prevent player from replay attacking the previous house_seed by cheating in the house_sign.
In step 9, the house uploads the house_seed.
In step 10, the smart contract verifies the house_seed to see if it matches with the house_hash so as to prevent house from swapping the house_seed sneakingly.
In step 11, the smart contract calculates the final roll num based on the player_seed & house_seed.
4. Talk is cheap, show me the code
In this case, since the randomness of the final roll number is dependent on both house side and player side, neither the house nor the player is able to rig the bet alone. In order to be rig resistant and provably fair, undoubtedly, the dice game has to be open sourced:
If you'd like to play our "FairDice" game, here's the corresponding game page:
You could find more info here:
 |
|---|
| DappPub: Unleashing the power of DApps |