dfsdf Cosmic Sociology - Multi Agent Systems Final Project
Cosmic Sociology
Andrija Curganov | Vladyslav Tomashpolskyi

Conclusion


What has been done?

On the basis of the BDI and Kripke models specified in the 'Problem Description' section we have built a model that allows us to simulate encounters between civilizations with various properties. This model consists of two main parts:

  • Civilization model. The civilizations and their properties are described in this part of the model:
    • Name. A name for the civilization.
    • Attitude. A general outlook towards other societies in space. Malevolent civilizations generally strive to destroy civilizations they encounter, while benevolent ones are much more likely to try and make contact with an alien society.
    • Technical level. A real number in the range of 0-10. This number represents technical prowess of the population. It being a real number constitutes subtle difference even between civilizations of the same scale. During an armed conflict this parameter defines who wins and who dies.
    • Depth of suspicion. An integer number in the range of 0-7. Represents societal arguments about other civilization's attitudes. Is it worth making contact, if another society may be malevolent? Or, maybe, if another society is benevolent, but thinks that we are malevolent and may try to destroy us anyway? During the lifetime of a benevolent civilization this parameter defines when it preemptively strikes, when it hides, or when it tries to make contact with another civilization.
    • State automaton. A representation of the BDI model. It defines what possible actions a civilization can take at any given point in time. The state automaton is slightly different for different types of civilizations. For example, a malevolent society does not have an opportunity to make contact, while a benevolent civilization does not consider a possibility of instant extermination of another society.
  • Statistics gathering engine. To be able to quickly gather statistics of multiple encounters, we have also implemented a way to automate the simulation of any specified number of encounters. By the end of the encounter each of two civilizations either defended itself by destroying other civilization and survived alone, or it prevailed and destroyed the other civilization (thus, also surviving alone), or it hid and thus removed a possibility of destruction or it cooperated with the other civilization. We provide a command-line based text interface that allows user to:
    • Do a manual simulation of one particular encounter.
    • Do an interactive simulation of one particular encounter, where the player assumes the role of one of the societies participating in an encounter.
    • Do a user-specified number of simulations, gather statistics, and present them upon completion.

Encounter Examples


An example of an encounter modeled for two malevolent civilizations. The Reavers detect Klingons during the second rotation and try to destroy them. But, unfortunately for them, Klingons apparently had a quite advanced tech level and were able to completely wipe The Reavers out. Klingons survive.

Modeling encounter for Klingon and The Reavers
[0] Current state of Klingon is: clueless and unsuspecting
[0] Current state of The Reavers is: clueless and unsuspecting
[1] Current state of Klingon is: clueless and unsuspecting
[1] Current state of The Reavers is: discovered another civilization
[2] Current state of Klingon is: clueless and unsuspecting
The Reavers failed to destroy Klingon
[2] Current state of The Reavers is: was destroyed
Klingon civilization finishes. It survives
The Reavers civilization finishes. It dies



An example of an encounter modeled for two benevolent civilizations. Both the Culture and the United Federation detect each other right away. The Culture - quite uncharacteristically - attempts a preemptive strike against UF, sending it to oblivion and beyond.

Modeling encounter for Culture and United Federation
[0] Current state of Culture is: discovered another civilization
[0] Current state of United Federation is: discovered another civilization
[1] Current state of Culture is: attempts a preemptive strikes at another civilization
                                 due to belief of it being malevolent, or expecting a preemptive strike
[1] Current state of United Federation is: was destroyed
Culture civilization finishes. It survives
United Federation civilization finishes. It dies



That one time when the United Federation was far more successful and cooperation was ensured.

Modeling encounter for Browncoats and United Federation
[0] Current state of Browncoats is: discovered another civilization
[0] Current state of United Federation is: discovered another civilization
[1] Current state of Browncoats is: tries to make contact with another civilization
[1] Current state of United Federation is: tries to make contact with another civilization
Browncoats civilization finishes. It cooperates
United Federation civilization finishes. It cooperates



An example of an encounter between a human-controlled and a benevolent civilization. The player has control over simulation and decides that his civilization will surely discover the other one on the first turn. Then the player, being fairly naive, makes a risky move by trying to hail the other civilization to cooperate with them. He succeeds though. And they lived happily ever after.

Modeling encounter for player and Browncoats
Oh wise ruler! What should we do?
    0. discover?
    1. newly born?
What will it be, wise one? 0
[0] Current state of player is: discovered another civilization
[0] Current state of Browncoats is: discovered another civilization
Oh wise ruler! What should we do?
    0. destroy?
    1. contact?
    2. hide?
What will it be, wise one? 1
We will contact other civilization immediately. Keep your fingers crossed so they don't destroy us.
[1] Current state of player is: tries to make contact with another civilization
[1] Current state of Browncoats is: tries to make contact with another civilization
player civilization finishes. It cooperates
Browncoats civilization finishes. It cooperates



Another example of a player-driven encounter. This time the computer-controlled civilization naively contacts the player with dreams of cooperation but those get shuttered right away when player perform Exterminatus against the fair Asari.

Modeling encounter for player and Asari
Oh wise ruler! What should we do?
    0. discover?
    1. newly born?
What will it be, wise one? 1
[0] Current state of player is: clueless and unsuspecting
[0] Current state of Asari is: discovered another civilization
Oh wise ruler! What should we do?
    0. discover?
    1. newly born?
What will it be, wise one? 1
[1] Current state of player is: clueless and unsuspecting
We received a transmission! We have two options:
    0. destroy?
    1. cooperate?
How shall we proceed? 0
We shall strike at those pesky creatures, and use all of their resources!
Yes, oh wise one. We prevailed and began the process of assimilation
[1] Current state of Asari is: was destroyed
player civilization finishes. It prevails
Asari civilization finishes. It dies

Statistics and analysis


We ran a number of simulated encounters to gather statistics. Encounters were executed one by one, without taking into account previous experiences. As the focus of this project is a resolution of a single encounter we were more interested in more secluded environments.

======================================================================
             1000000 simulations completed in 64.69 seconds
======================================================================
Total benevolent civilizations: 999793 (297610 died)
Total malevolent civilizations: 1000207 (443611 died)
Malevolent destroys benevolent: 221758
Malevolent destroys malevolent: 232493
Malevolent hides from malevolent: 17557
Malevolent hides from benevolent: 18771
Benevolent defends against malevolent: 162791
Benevolent successfully makes a preemptive strike against malevolent: 48327
Benevolent hides from malevolent: 48460
Benevolent destroys benevolent: 50507
Benevolent defends against a preemptive strike benevolent: 25345
Benevolent hides from benevolent: 52107
Benevolent cooperates with benevolent: 121884
Avg CoS depth resulting in death: 3.72
Avg CoS depth resulting in survival: 3.46
======================================================================

First of all it is evident that approximately 74.12% (741221 out of 1000000) of all encounters end in the destruction of one of participating civilizations. This number is bad news for any society trying to make contact in the Big Universe. Approximately two thirds of those are deaths of malevolent civilizations. This is understandable, as malevolent civilizations always go for an armed confrontation and have a 50% chance of having a higher tech level. Benevolent civilizations only make preemptive strikes or try to defend themselves. When they are trying to defend during an encounter, they again have a 50% chance of succeeding. Other deaths can be explained by preemptive strikes.

We distinguish a number of possible strategies for each civilization:

  • Destruction. "Destroy or be destroyed" as they say. This is a path that only malevolent civilizations walk: persue destruction of another civilization as soon as it is detected. Though it should be taken into account that sometimes another civilizations may prove quite resourceful, hence in 16.27% of the encounters, benevolent civilization manages to repel the assault.
  • Preemptive strikes. Sometimes even benevolent civilizations may be forced to resort to violence. If one civilization expects another one to strike first it may try to make a preemptive strike as a way to prevent this. This kind of action was initiated in 4.832% of benevolent-benevolent encounters.
  • Stealth. Go into hiding is always an option for any civilization. If one just sits put and tries to do everything to stay undetected, it may actually stay unnoticed, and survive. This accounts for 13.69% of all encounters in this particular run.
  • Cooperation. The ultimate goal of any benevolent society. This is dangerous as it implies that this society trusts another one with information about its location and state. This outcome constitutes 12.18% of encounters

It should also be noted that depth of suspicion has great influence over preferred course of actions of a benevolent civilization. The higher it is, the more likely a civilization is to regard another one as being benevolent of nature, therefore the more likely it is to make contact.