Hunting for Wolves

Jul 25, 2019 /
By Vladislav Ponomarev

Hunting for Wolves - Illustration

Gentlemen's Club

Cryptocurrencies are very volatile - they scare someone, push someone away. However, there are at least 30 million people in the world for whom they have become part of life.

Pretty much like in the real world, people strive for convenience and keep savings in places they trust. Wallets and exchanges are the first to be named.

The number of such places counts for hundreds and thousands. An ordinary crypto owner uses at least a few wallets and has several exchange accounts.

We conceived Mazekine as a gentlemen's club, where market players can easily exchange information and make the life of their users better. It works well when club members are not deceiving anyone and provide trustworthy information, but how can we check it?
What if there are hundreds of such gentlemen? Thousands?

That's where we faced the prisoner's dilemma in its entirety.

Prisoner's Dilemma

Back in 1950 two mathematicians, Merrill Flood and Melvin Dresher had framed the famous prisoner's dilemma.

It shows why two entirely rational individuals might not cooperate, even if it appears that it is in their best interests to do so. We assume that the player ("prisoner") maximizes his gain, without care for the benefit of others.

In the prisoner's dilemma, betrayal strictly dominates cooperation; hence, the only possible balance is the betrayal of both participants. To put it straight, whatever the behavior of another player is, everyone will win more if he betrays. Since in any situation it is more profitable to betray than to cooperate, all rational players will choose betrayal.

Behaving individually rationally, the participants come to an irrational decision: if both betray, they will receive a smaller gain in total than if they would cooperate. This is the dilemma.

In the prisoner's repeated dilemma, the game occurs periodically, and each player can "punish" another for non-cooperation earlier. In such a game, cooperation can become a balance, and the threat of punishment can outweigh the incentive to betray.

You can find a little more about the prisoner's dilemma here.

Hunting for Wolves

When designing our anti-fraud system, we realized that there should be a powerful incentive for club members to share only truthful information and to force others to do the same (Fair Play principle).

At the same time, as Robert Axelrod says, the system as a whole will be stable if the majority (ideally everyone) of the club members follow the same strategy, which is:

  • Nice, i.e., responsible not trying to circumvent the rules;
  • Retaliating, i.e., punishment must follow immediately;
  • Forgiving, i.e., should not prosecute the guilty after the execution of the penalty;
  • Non-envious, i.e., should not try to get more gains than other members.

We decided to base the Wolfhunt system on two elements: the reputation rating ("karma") and compulsory competition to identify cheaters ("hunting for wolves"). We use them in combination to motivate good players and prevent bad ones to poison the club's atmosphere.

Key Principles of Wolfhunt

  • All participants connecting to the Mazekine network are obliged to behave responsibly and not report false data.
  • All participants are required to monitor the quality of the data provided by implementing random checks:
    • The requesting participant requests information on the addresses it manages, at its own expense;
    • In the event of a false response, the participant immediately marks it calling a dedicated method of Mazekine's API and 50 transactions will be blocked on its account until the end of dispute;
    • The wallet suspected of fraud is required within three business days to provide evidence that it manages the specified address;
    • For the period of investigation 50 transactions are blocked on a suspect's balance;
    • If the suspected wallet is unable to provide evidence of managing a disputed address, its karma is reduced by 1 point, and the penalty of 50 transactions is deducted from his balance. Mazekine will then transfer 45 transactions to dispute initiator as an incentive, and five will be written off as Mazekine fee;
    • If the suspect provides evidence, the initiator of the proceedings will receive the same request;
    • If it turns out that the trial itself was a fraud, Mazekine will fine the initiator as described above;
    • If it turns out that several wallets manage the address (via a private key), Mazekine will save a note about such an address, and it will not participate in future disputes;
    • The number of random checks should be at least 0.5% of the total volume of transactions in the Mazekine network, or 50, whichever is greater.

As a result, a badly behaving member can easily lose its karma, whereas its restoration will take significant time and effort.

How karma affects system usage

If you ever tried to run your own business, you know that reputation is the key. That is why we set the following reputational rules to make it matter in the network:

  • Karma <= 80: the participant ceases to be rewarded for details about the addresses as long as it does not restore the required level of trust;
  • Karma <= 70: Broadcast requests to such participants come only in 1/3 of cases;
  • Karma <= 60: Mazekine will suspend the wallet's participation as a data provider in the network until the provision of evidence that measures have been taken to remedy the situation and restore good faith.

We do not pretend Wolfhunt is a perfect system, and we already know a lot of operational difficulties about it. However, it is a good point to start the evolution. Should you want to discuss Wolfhunt, feel free to leave a comment on our social networks.