The Emergence of Ethnocentrism in a Social Network

No preview image

1 collaborator

WHAT IS IT?

Ethnic conflict is an ongoing horror subplot in the story of the human race. From Holocaust remembrance month to the ongoing genocide in Darfur, it is something that is often on our minds and that continuously has to be dealt with. However, it appears to be possible that such tendencies may have actually initially evolved to counteract other evolutionary effects that gave human beings incentives to split into non-cooperative groups and not work as effectively together.

This is a model of an ancient land of knights, split into reds and blues. At first, they are all without lords, but being good knights, they seek out a someone to pledge their loyalty to. However, being good humans, they look out for their own interests, and if they see an opportunity to serve a lord with a greater following, they seize it. They are also capable of prejudice for their own kind, and violence, though these things are decided upon by the observer.

HOW IT WORKS

Each knight walks around aimlessly, looking for a lord to pledge his allegiance to. If he's not committed and runs into another knight, he pledges his allegiance to that knight. If he IS committed and runs into another knight, he compares that knight's following to his current lord's: if it's greater, he switches. This goes on until a king is crowned. However, knights can be prejudiced for following those of their own kind, and only pledge allegiance to them, and lords are capable of violence against other lords they view as potential threats.

HOW TO USE IT

The "percent-red" slider determines the percentage of knights that will be red, and the percentage that will be blue. The "there-can-be-only-one?" switch makes lords (knights with at least one follower) kill other lords they run into with less followers, and the "ethnic-rivalry?" switch makes the "leaders" of each race, or the member of that race with the most followers, kill the leader of the other if they happen to meet and the killer has more followers than the killee. The prejudice sliders set the level of prejudice for following one's own kind, with the number being the probability that each knight will only consider following a member of its own race at any given time.

THINGS TO NOTICE

If the switches are all off and the prejudice sliders are set to 0, notice how a king emerges every time, with every knight as a follower. If the prejudice is high for one race but not the other, that race will tend to produce the king, even if they're a minority. If the prejudice is high for both, they'll tend to split into ethnic groups each with its own leader, but eventually merge unless both prejudices are at 100 percent or only the prejudice for the minority is 100 percent. If prejudice is high for a minority, and the kill rival switches are turned on, far more members of that minority will die than if they are not prejudiced.

THINGS TO TRY

First, run the model with the switches off and prejudice sliders set to zero and see an undisputed ruler peacefully emerge. Next, set the percent-red slider to 40 and the red prejudice 100, and see a red king peacefully emerge. Next, set the percent red slider to 5 and the red prejudice to 100, and see a tiny red group and huge blue group, each with their own leader (except in very rare cases where red get lucky and manage to make a king). Next, set percent red 50 and prejudice for both groups 100, and see two equal, unconnected groups with their own leaders emerge. Next, set the percent red slider to 20, turn on the rival switches, and watch a fairly violent transition to a usually blue single leader occur. Next, keep the settings the same but turn red prejudice up to 100, and watch red either brutally take over Saddam Hussein style or get completely wiped out.

EXTENDING THE MODEL

Try making prejudice levels normally distributed across each knight, and making knights reproduce and die, seeing how prejudice levels evolve over time.

NETLOGO FEATURES

The "make-friends" procedure comes from a model of preferential attachment that is based on turtles having to "economize" on the number and kind of links they have, and that uses the concept of a "marginal friend" (called worst-friend in the procedures) to determine who to cut out of a network if a better opportunity arises.

RELATED MODELS

This is related to the RismanPreferentialAttachment model on Modeling Commons.

CREDITS AND REFERENCES

All of the code written by and ideas produced by Alex Risman, as far as he knows.

Comments and Questions

Please start the discussion about this model! (You'll first need to log in.)

Click to Run Model

globals [
  king
  smith-leader
  john-leader
  ]

directed-link-breed [friendships friendship]

breed [smiths smith]

breed [johns john]

breed [does doe]

breed [janes jane]

turtles-own [
  in-friends
  out-friends
  total-friend-set
  total-friends
  my-leader
  extended-network
  wealth
  age
  racist
]

to setup
  clear-all
  ask patches [set pcolor green]
  create-smiths 25
  create-johns 25
  create-janes 25
  create-does 25
  ask turtles [
     move-to one-of patches
     set wealth 1000
     set my-leader self
     set racist ifelse-value(random 100 >= initial-racism) [0] [1] 
   ]
  do-breeds
end    

to go
   ask turtles[
  wiggle
  set out-friends count out-friendship-neighbors
  set in-friends count in-friendship-neighbors
  set total-friend-set (turtle-set in-friendship-neighbors out-friendship-neighbors self)
  set total-friends count total-friend-set
  maintain-max
  make-friends
  graves
  if count turtles < 101
  [
    immigrate
    babies
  ]
  ]
  update-turtles
  if ethnic-rivarly?
  [kill-ethnic-rival]
  if there-can-only-be-one?
  [kill-all-rivals]
  
  gravesmore
  update-globals
  do-plots
  do-breeds
  tick
end  

to do-breeds
  ask turtles [
    set shape "person"
    if breed = smiths
    [set color red]
    if breed = johns
    [set color blue]
    if breed = janes
    [set color orange]
    if breed = does
    [set color yellow]
  ]
end 

to wiggle 
  rt random-float 360 fd 1
end 

to make-friends
 let prospect one-of other turtles
  if prospect != nobody
  [
    if out-friends < 1
    [ifelse hard-racism? 
      [ifelse [racist] of self = 0
        [create-friendship-to prospect]
        [if [breed] of prospect = [breed] of self
          [create-friendship-to prospect]
        ]
        ]
      [create-friendship-to prospect]
    ]
  if out-friends = 1
  [let worst-friend min-one-of out-friendship-neighbors [total-friends]
    if worst-friend != nobody
    [ifelse [racist] of self = 1
      [if [breed] of self != [breed] of my-leader and [breed] of self = [breed] of prospect
        [ask out-friendship-to worst-friend [die] 
         create-friendship-to prospect]
       if ([breed] of self = [breed] of my-leader and [breed] of self = [breed] of prospect) or ([breed] of self != [breed] of my-leader and [breed] of self != [breed] of prospect)
       [if [total-friends] of prospect > [total-friends] of worst-friend
         [ask out-friendship-to worst-friend [die] 
           create-friendship-to prospect
         ]]
      ]
      [if [total-friends] of prospect > [total-friends] of worst-friend
         [ask out-friendship-to worst-friend [die] 
           create-friendship-to prospect
         ]]
    ]
  ]
  ]
end 

to maintain-max
  let worst-out-friend min-one-of out-friendship-neighbors [total-friends]
  let worst-in-friend  min-one-of in-friendship-neighbors [total-friends]
  if worst-out-friend != nobody
  [
    if out-friends > 1
  [ask out-friendship-to worst-out-friend [die] ]
  ]
  if worst-in-friend != nobody  
    [if in-friends > 100
    [ask in-friendship-from worst-in-friend [die] ]
    ]
end 

to update-globals
  set king one-of turtles with [total-friends = count turtles - 1]
  if count smiths > 0
  [
  ifelse king = nobody and [in-friends] of one-of smiths > 0 
  [set smith-leader max-one-of smiths [total-friends]]
  [set smith-leader nobody]
  ]
  if count johns > 0
  [  
  ifelse king = nobody and [in-friends] of one-of johns > 0 and count johns > 0
  [set john-leader max-one-of johns [total-friends]]
  [set john-leader nobody] 
  ] 
end   

to update-turtles
  ask turtles
  [set my-leader max-one-of (turtle-set out-friendship-neighbors self) [total-friends]
   set wealth ifelse-value (accumulate-wealth?) [wealth] [0] + ([total-friends] of my-leader) * ((100 - relative-wealth-from-power)+ ifelse-value (my-leader = self) 
     [ifelse-value ([total-friends] of self > 2) [relative-wealth-from-power] [0]] [0] +
      ifelse-value ([breed] of my-leader = [breed] of self) [ifelse-value ([total-friends] of my-leader > 2)[ethnic-benefit] [0]][0]) 
      
 
  set age age + 1
;  ifelse [breed] of my-leader != [breed] of self
;  [if reactionary-tendency > random 100
;    [set racist 1]] 
;  [if reactionary-tendency > random 100
;    [set racist 0]] 
]
end 

to create-turtle  ;; patch procedure
  sprout 1 [
    set breed one-of (list smiths johns janes does) 
    set shape "person"
     set wealth 1000
     set my-leader self 
    set racist ifelse-value (50 > random 100)
    [1]
    [0]
    make-friends
    ]  
end 

to immigrate
  if chance-of-immigrant > random 100
  [ask one-of patches [create-turtle]]
end 

to babies
 if member? self max-n-of ((10 / 100) * count turtles) turtles [wealth]  
 [if 100 > random 100 
  [ hatch 1
    [ set wealth 1000
      wiggle
     if mutation-rate > random 100
     [set racist abs(racist - 1)
      set breed one-of (list smiths johns janes does) ]
     make-friends 
    ]
     ]]
end   

to graves
  if member? self min-n-of ((1 / 100) * count turtles) turtles [wealth] 
  [die]
  if 1 > random 5
  [if member? self min-n-of ((10 / 100) * count turtles) turtles [wealth] 
    [die]
  ]
end 

to gravesmore
  if 200 > random 100
  [ask one-of turtles [die]]
end 

to kill-ethnic-rival
  if ticks > 10 ;; without this delay, the model won't be able to run because none of the leaders will be defined
  [let smith-leaders (turtle-set smith-leader)
  let john-leaders (turtle-set john-leader)
  if smith-leader != nobody
  [ask smith-leader
  [ let rival one-of john-leaders with [xcor = [xcor] of self] with [ycor = [ycor] of self]
    if rival != nobody
    [if [total-friends] of self > [total-friends] of rival
      [ask rival
        [die]
      ]
    ]
  ]]
  if john-leader != nobody
  [ask john-leader
  [ let rival one-of smith-leaders with [xcor = [xcor] of self] with [ycor = [ycor] of self]
    if rival != nobody
    [if [total-friends] of self > [total-friends] of rival
      [ask rival
        [die]
      ]
    ]
  ]]
  ]
end 

to kill-all-rivals
  ask turtles
  [if my-leader = self and in-friends > 0 ;; letting rivals have no followers basically makes everyone kill everyone else 
    [let rival one-of other turtles-here with [my-leader = self and in-friends
        > 0] ;; see above comment
      if rival != nobody
    [if [total-friends] of self > [total-friends] of rival
      [ask rival
        [die]
      ]
    ]
    ]
  ]
end 

to do-plots
  set-current-plot "Racism"
  set-current-plot-pen "Reds"
  plot count smiths with [racist = 1]
  set-current-plot-pen "Blues"
  plot count johns with [racist = 1]
  set-current-plot-pen "Total"
  plot count turtles with [racist = 1]
  set-current-plot-pen "Oranges"
  plot count janes with [racist = 1]
  set-current-plot-pen "Yellows"
  plot count does with [racist = 1]
  set-current-plot "Population"
  set-current-plot-pen "Reds"
  plot count smiths
  set-current-plot-pen "Blues"
  plot count johns
  set-current-plot-pen "Total"
  plot count turtles
  set-current-plot-pen "Oranges"
  plot count janes
  set-current-plot-pen "Yellows"
  plot count does
end 

to extra
if self = one-of min-n-of (25 / 100 * count turtles) turtles [wealth] and 20 > random 100
  [die]
if self = one-of max-n-of (25 / 100 * count turtles) turtles [wealth] and 20 > random 100
  [set wealth wealth / 2
    hatch 1
    [wiggle
     create-friendship-to my-leader
     set age 0]
  ]
;   ifelse my-leader = self
;  [set wealth wealth + (wealth-from-network * wealth-from-power * (count [total-friend-set] of my-leader with [breed != [breed] of self]
;    + ethnic-benefit * (count [total-friend-set] of my-leader with [breed = [breed] of self])))]
;  [set wealth wealth + (wealth-from-network * (count [total-friend-set] of my-leader with [breed != [breed] of self]
;    + ethnic-benefit * (count [total-friend-set] of my-leader with [breed = [breed] of self])))]
  if random 100 < mutation-rate
     [ifelse racist = true
       [set racist 0]
       [set racist true]]
end

There is only one version of this model, created almost 11 years ago by Alexander Risman.

Attached files

No files

This model does not have any ancestors.

This model does not have any descendants.

NetLogo