def calculate_new_outcome(self, new_voter_preference, voter):
new_preference = self.preferences.copy()
new_preference[voter] = new_voter_preference
new_voting_scheme = VotingScheme(new_preference,
len(new_preference[0]))
new_voting_scheme.execute_voting(self.voting_scheme_option)
new_outcome = new_voting_scheme.get_outcome()
return new_outcome, new_voting_scheme
def change_voter_vote(self, voter, vote):
new_preferences = self.preferences.copy()
possible_strategy = []
true_preferred_cand = self.preferences[voter][0]
opponent = self.vs.get_opponent(true_preferred_cand)
for candidate in self.hate_list:
voter_preferences = self.preferences[voter].copy()
if candidate is voter_preferences[0]:
# It does not make sense making win someone less hated than the person we would like that wins
break
# print()
# Change the preferred voted
c_index_v_pref = voter_preferences.index(candidate)
help = voter_preferences[c_index_v_pref]
voter_preferences[c_index_v_pref] = true_preferred_cand
# print("The voter", voter, "was voting for", voter_preferences[0], "and is gonna try to vote for", help)
voter_preferences[vote] = help
new_preferences[voter] = voter_preferences
# Create a new voting scheme with the new preferences
new_vs = VotingScheme(new_preferences, len(new_preferences[0]))
# Vote a first time with the new preferences
new_vs.execute_voting(self.voting_scheme)
first_outcome = new_vs.get_outcome()
new_opponent = new_vs.get_opponent(true_preferred_cand)
# If the outcome of applying the voting manipulation is the same one then it is not a good manipulation
if collections.Counter(first_outcome[:2]) == collections.Counter(
self.winner_list[:2]):
continue
elif self.hate_list.index(new_opponent) >= self.hate_list.index(
opponent):
continue
# print("The result of the first round is", help_outcome) # Vote a first time
new_vs.second_round(self.preferences)
second_outcome = new_vs.get_outcome()
# If after the second round the winner is the desired one then it is an acceptable way of strategic voting
if new_vs.get_outcome()[0] is self.preferences[voter][0]:
print("For the voter", voter, "that was voting for",
self.preferences[voter][0],
"a new possible strategies is voting for", help)
print("The new outcome in first round would be", first_outcome,
"and in the second", new_vs.get_outcome())
possible_strategy.append(help)
# If the two first candidates are not the same as previously
# and the preferred candidate of the voter is still in top 2
if len(possible_strategy) > 0:
print("For the voter", voter,
"the possible strategies are voting for", possible_strategy)
def test_experiment_borda_voting_voter_5(self):
# Initiate experiment variables
number_of_candidates = 4
voting_scheme = VotingSchemeOption.BORDA
# Preferences Matrix
preferences = {
0: ["C", "A", "D", "B"],
1: ["B", "D", "C", "A"],
2: ["C", "D", "A", "B"],
3: ["B", "D", "C", "A"],
4: ["B", "C", "D", "A"]
}
voting = VotingScheme(preferences, number_of_candidates)
voting.execute_voting(voting_scheme)
model = Model(preferences, voting_scheme)
results = model.calculate(False)
# true outcome should be
self.assertEqual(["C", "B", "D", "A"], results[0])
# Voter 5 (index:4) should bury C
self.assertEqual('This candidate was buried', results[2][4][3]["C"])
# and the new outcome should be, making B win instead of C
self.assertEqual(["B", "C", "D", "A"], results[2][4][1])
# and his new preferences should be
self.assertEqual(["B", "D", "C", "A"], results[2][4][0])
# In the slides example the voter actually swapped C and A candidates instead of
# B, C, D, A he voted B, A, D, C
# In our software his new preferences where B, D, C, A
# The overall happiness is reduced
new_overall_happiness, old_overall_happiness = results[2][4][
2], results[1]
self.assertTrue(new_overall_happiness < old_overall_happiness)
# Happiness of voter 5 should be increased
old_happiness_of_5 = voting.get_happiness_by_voter(4)
new_happiness_of_5 = voting.get_new_happiness_by_voter(
4, ["B", "C", "D", "A"])
self.assertTrue(new_happiness_of_5 > old_happiness_of_5)
class PushOver:
"""
Voting a Candidate that is easy to beat insincerely high in the first round so that the second round is easy to win
"""
def __init__(self, preferences, voting_scheme):
self.preferences = preferences
self.voting_scheme = voting_scheme
self.vs = VotingScheme(preferences, len(preferences[0]))
self.outcome = self.vs.execute_voting(voting_scheme)
self.winner_list = self.vs.get_outcome()
self.happiness = self.vs.calc_happiness()
self.hate = self.calculate_hate()
self.hate_list = self.calculate_hated_candidates()
# self.calculate()
self.change_voter_votes_borda(1)
def calculate(self):
print("Outcome", self.outcome)
print("Winner list", self.winner_list)
# for candidate in self.winner_list:
# print("Candidate", candidate, "has", self.outcome[candidate], " votes")
print("Hated list", self.hate_list)
for voter, voter_preferences in self.preferences.items():
# If the preferred candidate of the voter would win without doing anything
# if voter_preferences[0] is self.winner_list[0] or voter_preferences[0] is self.winner_list[1]:
# Maybe it is interesting to vote for someone else that other voters
# do not like so that the second round is easier
# print()
# print("Voter", voter, "is very HAPPY")
# print(self.preferences[voter])
self.change_voter_votes_plurality(voter)
"""
In plurality voting only one candidate can be voted and it is therefore the only that should be changed.
"""
def change_voter_votes_plurality(self, voter):
self.change_voter_vote(voter, 0)
"""
In voting for two, only the two first candidates in the preference list get a point. Since both of them get a point
it is not important which one is the first or second in the manipulation.
"""
def change_voter_votes_for_two(self, voter):
self.change_voter_vote(voter, 0)
self.change_voter_vote(voter, 1)
"""
In anti-plurality voting, only the least preferred candidate in the preference list does not get a point.
Therefore, only changing the least preferred candidate could affect the outcome of the voting
"""
def change_voter_votes_anti_plurality(self, voter):
self.change_voter_vote(voter, len(self.preferences[voter]))
"""
In borda voting every preference gets a point and the higher it is in the preference list, the more points it gets.
Therefore, all possible combinations of preferences have to be checked
"""
def change_voter_votes_borda(self, voter):
# TODO
i = 0
for possible in itertools.permutations(self.preferences[voter]):
print("A new possibility is", possible)
i += 1
print(i)
# self.change_voter_vote(voter, len(self.preferences[voter]))
return
def change_voter_vote(self, voter, vote):
new_preferences = self.preferences.copy()
possible_strategy = []
true_preferred_cand = self.preferences[voter][0]
opponent = self.vs.get_opponent(true_preferred_cand)
for candidate in self.hate_list:
voter_preferences = self.preferences[voter].copy()
if candidate is voter_preferences[0]:
# It does not make sense making win someone less hated than the person we would like that wins
break
# print()
# Change the preferred voted
c_index_v_pref = voter_preferences.index(candidate)
help = voter_preferences[c_index_v_pref]
voter_preferences[c_index_v_pref] = true_preferred_cand
# print("The voter", voter, "was voting for", voter_preferences[0], "and is gonna try to vote for", help)
voter_preferences[vote] = help
new_preferences[voter] = voter_preferences
# Create a new voting scheme with the new preferences
new_vs = VotingScheme(new_preferences, len(new_preferences[0]))
# Vote a first time with the new preferences
new_vs.execute_voting(self.voting_scheme)
first_outcome = new_vs.get_outcome()
new_opponent = new_vs.get_opponent(true_preferred_cand)
# If the outcome of applying the voting manipulation is the same one then it is not a good manipulation
if collections.Counter(first_outcome[:2]) == collections.Counter(
self.winner_list[:2]):
continue
elif self.hate_list.index(new_opponent) >= self.hate_list.index(
opponent):
continue
# print("The result of the first round is", help_outcome) # Vote a first time
new_vs.second_round(self.preferences)
second_outcome = new_vs.get_outcome()
# If after the second round the winner is the desired one then it is an acceptable way of strategic voting
if new_vs.get_outcome()[0] is self.preferences[voter][0]:
print("For the voter", voter, "that was voting for",
self.preferences[voter][0],
"a new possible strategies is voting for", help)
print("The new outcome in first round would be", first_outcome,
"and in the second", new_vs.get_outcome())
possible_strategy.append(help)
# If the two first candidates are not the same as previously
# and the preferred candidate of the voter is still in top 2
if len(possible_strategy) > 0:
print("For the voter", voter,
"the possible strategies are voting for", possible_strategy)
def calculate_hate(self):
hate = {}
for voter, voter_preferences in self.preferences.items():
# Since the preferences are order being the most preferred in the position zero and the less liked
# in the position n, it can also be understood as a measure of not liking or dislike
for dislike, candidate in enumerate(voter_preferences):
try:
hate[candidate] += dislike
except:
hate[candidate] = dislike
return hate
def calculate_hated_candidates(self):
return sorted(self.hate, key=self.hate.get, reverse=True)
class Model:
COMPROMISE = "Compromised in favor of this candidate"
BURYING = "This candidate was buried"
BULLET_VOTING = "There was bulet voting in favor of this candidate"
"""
Voting a Candidate that is easy to beat insincerely high in the first round so that the second round is easy to win
"""
def __init__(self, preferences, voting_scheme_option):
self.preferences = preferences
self.voting_scheme_option = voting_scheme_option
self.voting_scheme = VotingScheme(preferences, len(preferences[0]))
self.points_outcome = self.voting_scheme.execute_voting(
voting_scheme_option)
self.outcome = self.voting_scheme.get_outcome()
self.happiness = self.voting_scheme.calc_happiness(self.outcome)
self.overall_happiness = self.voting_scheme.calc_overall_happiness(
self.outcome)
"""
Calculates for every voter the possible strategic-voting result
output possibly empty strategic-voting option
"""
def calculate(self, bullet_voting_allowed):
strategic_voting_option = []
self.n_strategic_options = 0
for voter in range(len(self.preferences)):
voter_max_hap = self.voting_scheme.calc_happiness(
self.outcome)[voter]
voter_original_hap = voter_max_hap
voter_strategic_votes = {}
candidates = self.preferences[voter]
if self.voting_scheme_option != VotingSchemeOption.PLURALITY_VOTING and bullet_voting_allowed:
self.calculate_bullet_voting(candidates, voter, voter_max_hap,
voter_strategic_votes,
voter_original_hap)
elif self.voting_scheme_option == VotingSchemeOption.PLURALITY_VOTING:
self.calculate_compromising_and_burying_plurality(
voter, voter_max_hap, voter_strategic_votes,
voter_original_hap)
else:
self.calculate_compromising_and_burying(
voter, voter_max_hap, voter_strategic_votes,
voter_original_hap)
strategic_voting_option.append(
self.evaluate_outcome(voter, voter_strategic_votes))
risk = self.n_strategic_options / len(self.preferences)
return self.outcome, self.overall_happiness, strategic_voting_option, risk
def calculate_compromising_and_burying_plurality(self, voter,
voter_max_hap,
voter_strategic_votes,
voter_original_hap):
for index, candidate in enumerate(self.preferences[voter]):
new_voter = voter
new_voter_preferences = self.preferences[voter].copy()
# Change the preferred voted
new_voter_preferences[index] = new_voter_preferences[0]
new_voter_preferences[0] = candidate
# print("The voter", voter, "was voting for", voter_preferences[0], "and is gonna try to vote for", help)
new_outcome, new_voting_scheme = self.calculate_new_outcome(
new_voter_preferences, voter)
if new_outcome[0] is self.outcome[0]:
continue
new_happiness = self.voting_scheme.get_new_happiness_by_voter(
voter, new_outcome)
if new_happiness > voter_original_hap:
self.n_strategic_options += 1
if new_happiness <= voter_max_hap:
continue
voter_max_hap = new_happiness
voter_strategic_votes[new_voting_scheme] = new_outcome
def calculate_compromising_and_burying(self, voter, voter_max_hap,
voter_strategic_votes,
voter_original_hap):
for new_voter_preference in itertools.permutations(
self.preferences[voter]):
new_voter_preference = list(new_voter_preference)
new_outcome, new_voting_scheme = self.calculate_new_outcome(
new_voter_preference, voter)
if new_outcome[0] is self.outcome[0]:
continue
new_happiness = self.voting_scheme.get_new_happiness_by_voter(
voter, new_outcome)
if new_happiness > voter_original_hap:
self.n_strategic_options += 1
if new_happiness <= voter_max_hap:
continue
voter_max_hap = new_happiness
voter_strategic_votes[new_voting_scheme] = new_outcome
def calculate_bullet_voting(self, candidates, voter, voter_max_hap,
voter_strategic_votes, voter_original_hap):
# to perform bullet voting as strategic voting we have to generate the preference with only the first one
for candidate in candidates:
new_voter_preference = ["" for i in candidates]
new_voter_preference[0] = candidate
new_outcome, new_voting_scheme = self.calculate_new_outcome(
new_voter_preference, voter)
if new_outcome[0] is self.outcome[0]:
continue
new_happiness = self.voting_scheme.get_new_happiness_by_voter(
voter, new_outcome)
if new_happiness > voter_original_hap:
self.n_strategic_options += 1
if new_happiness <= voter_max_hap:
continue
voter_strategic_votes[new_voting_scheme] = new_outcome
"""
Evaluate the input strategic vote for the input voter and return the kind of strategic vote applied and to which candidate.
input @voter
input @voter_strategic_votes
output (dict/list/array?) changes
"""
def evaluate_outcome(self, voter, voter_strategic_votes):
best_happiness = self.voting_scheme.get_new_happiness_by_voter(
voter, self.outcome)
changes = {}
# for bullet prove change the preference in the array with ""
# before checking if the position is greater/smaller than the original check if the position is ""
new_preferences = []
new_final_outcome = []
new_overall_happiness = 0
for new_voting_scheme, new_outcome in voter_strategic_votes.items():
if self.voting_scheme.get_new_happiness_by_voter(
voter, new_outcome) <= best_happiness:
continue
best_happiness = self.voting_scheme.get_new_happiness_by_voter(
voter, new_outcome)
changes = {}
if new_voting_scheme.preferences[voter][1] is "":
candidate = new_voting_scheme.preferences[voter][0]
changes[candidate] = self.BULLET_VOTING
new_preferences = new_voting_scheme.preferences[voter]
new_final_outcome = new_outcome
new_overall_happiness = self.voting_scheme.calc_overall_happiness(
new_outcome)
continue
# Position is the index in the array and candidate the value
for old_position, candidate in enumerate(self.outcome):
if new_outcome.index(candidate) < old_position:
# Compromising
changes[candidate] = self.COMPROMISE
elif new_outcome.index(candidate) > old_position:
# Burying
changes[candidate] = self.BURYING
new_preferences = new_voting_scheme.preferences[voter]
new_final_outcome = new_outcome
new_overall_happiness = self.voting_scheme.calc_overall_happiness(
new_outcome)
return new_preferences, new_final_outcome, new_overall_happiness, changes
def calculate_new_outcome(self, new_voter_preference, voter):
new_preference = self.preferences.copy()
new_preference[voter] = new_voter_preference
new_voting_scheme = VotingScheme(new_preference,
len(new_preference[0]))
new_voting_scheme.execute_voting(self.voting_scheme_option)
new_outcome = new_voting_scheme.get_outcome()
return new_outcome, new_voting_scheme