Gini Index (Metrics Discussion)

[AngelaKTE]

Description

• Goal: Measure inequality in funding distribution across projects
• Our evaluation makes a statement if a voting rule "flattens" or "amplifies" diverse distribution of funding
• Why here?:
  • We'd like to understand if/how much a voting rule affects the equal distribution of funds
  • Equal/unequal distribution of funding (funding outcome per project) could both be a goal for OP Retro Funding:
  • unequal distribution due to project diversity: there's a wide variety of projects at different stages in their livecycle and different impact created, hence large inequality in funding is good
  • equal distribution: since measuring impact is hard for many projects, we aim for equal funds distribution, where groupthink and popularity do not lead to extremes in funds distribution.
  • Note that we don't explore the diversity of voter preferences with this metric, Gini Index here does not say anything about the diversity of voter preferences (see comment GOV-22)
• We compare:
  • we use the exact same votes, and voting rounds, and compare the results, to see if the voting rule itself affects the equal distribution of funding

Chart:

• Lorenz curve plotting the distribution of funding per voting rule
• x-axis: cumulative population proportion (ascending order)
• y-axis: cumulative funding proportion (average funding from all voting rounds we ran in the simulation)

Open questions:

• none

[linear[bot]]

GOV-22 Gini Index (Metrics Discussion)

#### Description * **Goal:** Measure inequality in funding distribution across projects * **Why here?:** * We'd like to understand if/how much a voting rule affects the equal distribution of funds * Equal/unequal distribution could both be a goal for OP Retro Funding: * unequal distribution due to voter preference diversity: voters have different preferences, and it's good that these variety of preferences is captured in the voting results (question: we would have to measure the diversity of preferences in the voter input to make this statement! What if all voters simply agree that there are projects that deserve little, and others deserve a lot?) * unequal distribution due to project diversity: there's a wide variety of projects at different stages in their livecycle and different impact created, hence large inequality in funding is good * **We compare:** * we use the exact same votes, and voting rounds, and compare the results, to see if the voting rule itself affects the equal distribution of funding **Chart:** * Lorenz curve plotting the distribution of funding per voting rule * x-axis: cumulative population proportion (ascending order) * y-axis: cumulative funding proportion (average funding from all voting rounds we ran in the simulation) **Open questions:** * with "equality" we refer to the voting outcome, the funding a project receives * equality does *not* say anything about the diversity of voter preferences, no? * Shouldn't the simulation compare

[AngelaKTE]

@idrees535 would the Lorenz curves be an option for plotting the Gini Index?
see here:
https://zhiyzuo.github.io/Plot-Lorenz/

[idrees535]

Plotted lorenz curve for gini index in run.ipynb