Games
From Telecomix Crypto Munitions Bureau
Contents |
[edit] cooperation among conscious agents
Games of cooperation can also be said to be played out by conscious agents (such as yourself) when coordinating hacktivism within a group. Cooperation with an anonymous stranger can be risky if it turns out that the stranger is working for the pentagon (see the pentagon game below).
The prisoners dilemma is a classical game played by game-theorist. The game is a turn-based game played by two agents, A and B. At each turn, both A and B can choose to either cooperate or defect. If both cooperate, they both gain "much" from it. If one agent defects and the other cooperates, the defecting agent will "gain much more than the cooperating agent". If both defect, they both gains "slightly".
[edit] Game: The prisoners dilemma
Below is one form of the prisoners dilemma.
Rules:
- No agent knows what the other agents strategy is.
- It is secret what choice each agent has made until at the end of each turn. At the end of each turn, the choice of the opponent is revealed to both agents.
- The game continues forever. (Alternatively, for a large but uncertain amount of turns.)
Points:
- A cooperates, B cooperates: both gain 3 points.
- A cooperates, B defects: A gains 1 point, B gains 4 points.
- A defects, B cooperates: A gains 4 points, B gains 1 point.
- A defects, B defects: both gain 2 point.
Goals of the game:
- If the goal of the game is to gain more points than the opponent, the winning move is to always defect.
- If the goal of the game is to gain as many points as possible, the winning strategy is to cooperate at the first turn, and for every other turn do what the opponent did at the previous turn. That is, the winning strategy is to first be nice, and then let the opponent play against itself. "To master the game, one has to become the game."
The essence of this game can be played out in more elaborate forms within social groups. The anarchist eco-willage -- a game which assumes that every agent is playing against themselves -- relies on the agents to cooperate with each other, as each agent strives towards gaining the maximum amount of points (or happiness). In a war however, the objective is often to win against the opponent. One could crudely say that war and the anarchist eco-willage has the same fundamental rules, but differ in objectives.
In psychological tests with humans that play the above game and is allowed to walk away from the test with as many dollars as they has gained points, they often select to defect in order to win a few extra bucks. However, if the humans that play this game are already friends, they tend to more often select cooperation as their strategy. This irrational difference in choice of strategies for friends vs. non-friends can not be easily explained.
[edit] Game: The pentagon-game
- Elaborate on what "death" mean, it could mean something else (cooperating with a torrent client that RIAA runs can result in un-payable fines, or forced censorship through HADOPI. In totalitarian countries like Iran and Burma it has however really meant death or torture.)
In this game, whistleblowers needs to be protected from being discovered at the same time as the identities of a few activist agents needs to stay hidden. In this game, there are two types of agents: Those that plays the game as if it was war (pentagon agents) and those thay play it in order to gain as many points as possible (activist agents.) What differ the two types of agents is the chosen goal of the game.
Below is the simple model of the game. Bold text represents a difference from the previous game described above.
Rules:
- No agent knows what the other agents strategy is.
- No agent knows what the other agents goal is.
- It is secret what choice each agent has made until at the end of each turn. At the end of each turn, the choice of the opponent is revealed to both agents.
- The game continues forever. (Alternatively, for a large but uncertain amount of turns.)
Points:
- A cooperates, B cooperates: both survive and gains one point each.
- A cooperates, B defects: A dies, B survives.
- A defects, B cooperates: A survives, B dies.
- A defects, B defects: both survive.
Goals of the game:
- If the goal of the game is to kill all activist agents, the winning strategy is to always defect.
- If the goal of the game is to gain as many points as possible, the winning strategy is to not play the game. (Zero is the highest number of points one can gain without risking oneself.)
However, not playing the game is completely unacceptable.
[edit] Game: N-player with common resources
If we scale the previous game to include more than two players we can discuss collective resource management. Consider that a group of agents has a common resource that is evenly divided among all participants of the game. Each agent can choose to deposit their own resources (points) into the collective resource pool ("the pool"). The more points that has been given away to the collective, the more points will everyone have. In this game, we assume that the number of points in the pool is multiplied by a constant at the end of every turn.
Rules:
- Each agent begins with 2 points.
- We do not know how many other agents there are playing the game. No agent knows how many points any other agent has given to the pool. An agent only know how many points that was received from the pool the previous turn.
- At each turn, every agent can make a choice of how many points should be given away to the pool.
- The game continues forever. (Alternatively, for a large but uncertain amount of turns.)
Points:
- The number of points in the pool is multiplied by 1.1 and is then evenly distributed among all the participants at the end of every turn (it is possible to receive fractions of points from the pool). Points earned from the pool can be used the next turn.
- The number of points that was not given away to the pool are saved to the next turn.
Goals of the game:
- Greedy agents wish to gain more points than the everyone else. The winning strategy is to never give away points.
- Non-greedy agents wish to gain the most points possible. The winning strategy is to give away at least one point at the first turn, and then keep giving away the same amount of points that one just received from the pool. The average number of points the cooperating agents are willing to give away at the first turn will determine how quickly the collective resources approach infinity, if at all.
Corporations are bound by law to seek to increase their stock value. Stock value is their relative value in comparison with other actors of the market. Elaborate here. How can greedy agents come to influence the systems overall behavior? How can we overcome this problem?
[edit] The winning game
(i think that this piece of text needs complete a rewrite) Since it is the activist agents that are the sole creators of their game, they could modify the rules. In cipherspace, activists can manufacture identities at any speed they want, or select to not have any identity at all. Such manufactured identities can be either just random numbers (a public key is a random number), or pseudonyms together with some sort of method for identification (again, a public key). A manufactured identity in is in this sense anonymous, unless the agent wish to be identifiable under a pseudonym. Punishment for playing the non-defection game can only be delivered if authorities has knowledge about a form of identification tied to an address, a social security number, IP address, or a name. Through exploiting the laws of mathematics (and especially its subfield, cryptography) one can create a space where the laws of human legal systems can not operate. Its agents can not play the defecting move. Ideally, this is how it should be at least. In reality the protocols used often has flaws. Nothing is perfect.
Three new games, three new strategies:
- The simplest solution for avoiding loosing the game is to refuse to play the game with anyone that has not proved to be a cooperative (activist) agent. Dropping anonymity to a trusted group of peers is one solution to gain the trust needed to play the game with other trusted agents. This strategy puts a limit on how large the group can become, as ones friends friends might be "defecting" friends. (The game above is more-or-less still played.)
- Another solution is to never reveal any identity of any participating agent. Whistleblowers can use cipherspace to send information to an anonymous publisher, or publish the information themselves. The value of the information published will however be difficult to determine, as its orign can not be not be fully trusted. It is generally not possible to trust something that one has no knowledge about. (There is no game.)
- Yet another strategy is to rely on manufactured identities; pseudonyms that has no connection to ones body. Creating a pseudonym in cipherspace is as easy as selecting a name and instruct the software to build a new public key pair. Using the signature of ones cipher, one becomes nearly impossible to impersonate while at the same time the identity is nearly impossible to trace back to the agents physical body (which is subject to punishment). Cooperation can then be safely carried out as long as one does not comply with a request to reveal ones real identity. Instead of trusting each other, they can put trust in each others pseudonyms and cipher keys. (This is a completely new game.)
(this text needs to be more well written) If we examine the completely new game (number 3), we have a new set of rules. First off, we know that no agent can be "killed" unless they reveal their "real" identity. Example scenario: The whistleblower has anonymously uploaded a piece of information to a location and this information needs to be processed by agents before it is published. The agents running the publishing site will need to verify that it is genuine before they can use their (trusted?) signatures to show to the world that they think the information is genuine. People that trust the signatures because they have been trustworthy in the past, can continue to trust the validity of the leaked information. In order to gain trust, one can help other activists by validating the informations truthfulness. Cooperation requires no sacrifice and no danger for anyone, as long as they do not themselves reveal their identities. The "winning game" of socio-cyphernetic black-box activism is likely to be a game where there is no such thing as a defective move.
With computer software and communications networks, we can create and play this winning game. Total world dominance is only reachable if we shed our need for authorities.
