• Nir Kabessa

Social Weighted Voting: Remedying Rational Ignorance in Governance



At an MIT hackathon in 2019, we thought of a simple idea to reduce the effects of rational ignorance on elections, decision-making, and general governance.

"Rational ignorance is refraining from acquiring knowledge when the supposed cost of educating oneself on an issue exceeds the expected potential benefit that the knowledge would provide." (Wikipedia)

In the context of governance, it occurs when voters lose from learning enough about a decision to vote in a way that's best for them, because their vote is inconsequential. They actually maximize utility by remaining ignorant.

So what if everyone's vote wasn't equal, and some had a bigger say than others? That's called weighted voting, and it has existed throughout history. A direct yet unfair example is Sweden's wealth-weighted voting (1866-1914) where “the wealthiest members of the rural communities received as many as 5,000 votes” and “in 10 percent of the districts the weighted votes of just three voters could be decisive." While this sounds highly unequal for national governance, it's very much the norm within DAOs and blockchain consensus mechanisms, referred to as token-weighted voting.

A fairer example of this is the US's electoral college, where individuals vote for representatives who vote on their behalf. In most cases, however, Americans have no knowledge of their electoral representative and are essentially making one decision: who to elect.

Advances in technology allow for expanding this model to a peer-to-peer ecosystem where every voter is an electoral representative. Our concept presented such a model.

Social-Weighted Voting

The basic model we proposed is similar to Alvin E. Roth's two-sided matching market with incomplete information about others' preferences, except the purpose of the result is quite different. In our proposed case, Pareto efficiency is measured in the utility of the outcome of the following decision (election, referendum) that the total population makes (similar to quadratic voting). In simple terms, it adds one step before decision making: each voter ranks every other voter they know based on how much they trust them to vote well. Then, with a ranked list of citizens, votes on decisions are weighed.

Given:

  • There is a finite set of voters: V = {v1,v2,v3...v_n}

  • There is a finite set of binary decisions to vote on: D={d1,d2,d3...d_i}

Steps:

  1. Each voter ranks r_m all other voters by expected utility of their vote

  2. Each voter receives x points for the rank he receives from each user he receives as

3. Voters vote on decisions with a weight of x_n

4. Repeat every referendum

Several advantages:

  • Retain 1-person-1-vote while weighing heavily the votes of those who are trusted to make decisions

  • Increase Pareto efficiency by better reflecting the desires of each ignorant participant

  • Create a reputation mechanism between participants → everyone is both a citizen and a public representative

  • Negative ranking (placing a voter at the bottom of your list) can even out the advantage of popularity/celebrity

Adding Quadratic Voting

Quadratic voting and social weighted voting can function together to achieve something closer to Pareto efficiency, both through how a voter weighs his votes and the tradeoffs for the decisions that they make.

Critical Players

In a model such as this, certain participants become critical players who are susceptible to bribery and collusion. This model does not solve for that, only proposes a new means of unequal distribution of power that remains fair.

© 2019 Yup.io Inc.

  • Black Facebook Icon
  • Black Twitter Icon
  • Black Instagram Icon