Bribery Cost (Metrics discussion)

[AngelaKTE]

Description

• Goal: Measure the cost to change the funding outcome for a project (desired_increase)
• Assumption: we assume that changing the allocation of 1 vote (token) costs $1
• Why here?:
  • we'd like to understand how many tokens have to be moved on average to change the funding outcome a_p for a project and compare if the voting rule affects the cost of bribery
  • ideally, the voting rule requires a lot of votes/tokens to be moved to change the outcome
  • even though it is called "bribery", there could be many ways to convince voters to change their vote:
  • bribe
  • collusion
  • software manipulation
  • our metric in its essence tells us how many tokens/votes have to be moved or manipulated to change the outcome
  • it does not matter, if the votes to be moved are held/distributed by one or several different voters (bribing one voter could be more likely logistically, while convincing voters to vote only slightly different could be easier)
• How we model it:
  • in the simulation, we randomly select a desired_increase of the funding outcome for the attacker, and calculate the bribing costs for this target value
  • in the simulation, we randomly select the target project who should receive the desired_increase
• We compare:
  • we compare the voting rules
  • cost of bribery per desired_increase (note to Angela: not average across all desired_increase)

Chart:

• we plot the cost of bribery per desired_increase for all voting rules evaluated
• x-axis: desired_increase
• y-axis: cost of bribery

Open questions:

the code seems not final? def simulate_bribery_median(model, target_project, desired_increase):

… bribery_cost = desired_increase ?

[linear[bot]]

GOV-23 Bribery Cost (Metrics discussion)

[AngelaKTE]

Notes from the formal spec of Evaluation metrics. We are interested in: – The theoretical minimum bribery cost. – The theoretical maximum bribery cost. – The average bribery cost with respect to some distribution. – The expected needed bribery cost for a given dataset.

Show the costs in the simulation chart, based on "the cost to move a token from one project to another is 1".