Athena: Building a Governance Platform
Last week, you read about one of the winning Ethereal projects: Attacking Casper, a study into Casper’s security. Another group of students from Blockchain at Berkeley — Saroj Chintakrindi, Jeremiah Andrews, Kevin Chang, Tammy Vu, Avneet Saini, and Federico Kunze — had a different take on the topic of governance: voting mechanisms.
◼ Project Overview and Motivation
Getting shareholders involved in corporate decisions has always been a problem that didn’t have a clear solution. According to CNN, “Just 27% of shareholders bother to vote.” They don’t feel that they can make an impact with the voting power that they have. That’s where Athena comes in. Athena implements a voting scheme (CTT Voting) where shareholders can vote according to their true preferences, based on the amount of shares they own. It also rewards those who participate with interest. This mechanism is socially optimal in the sense that it picks the outcome that benefits everyone the most, and incentivizes honest reporting of each voter’s preferences. Athena implements this modularly, to allow for the mechanisms to be easily switched out.
◼ Protocol and Incentives
Each voter is assigned a certain number of tokens proportional to their share of the company. Then, the administrative board can use the Athena platform to ask a question and present a certain number of options. Each voter stakes a certain number of their coins on each option (0 to 100%). Afterwards, the staked coins are aggregated and a winning option is chosen. Then, each voter is subjected to a Clarke tax, which is computed as follows: for each voter, an alternative tally is calculated in which that voter did not vote. If the alternate winner is different from the original one, then the margin by which the alternate winner won is charged to the voter. This way, each voter is charged based on the influence of their vote. In order to incentivize members to vote, voters all receive a small amount of interest at the end of the phase. In order to keep the game theory mechanisms secure, all the charged tokens are burned. Let’s take a look at an example. Suppose that an organization wants to decide between acquiring Company A and Company B. For the sake of simplicity, let there be three voters, each with a hundred tokens, that stake their tokens as follows:
Clearly, company A wins by 10 votes. Now, let’s compute the Clarke tax for Bob:
When Bob’s stake is removed, we see that company B wins by 20 tokens, indicating that Bob’s vote was indeed decisive and he is duly charged 20 tokens. Similarly, if we remove Eve’s vote, we see that Company B wins by 30, so Eve is taxed 30.
◼ What if you can’t use CTT?
The great part is that if CTT doesn’t work for a particular organization, then the modularity of the Athena smart contract allows the administrative board to use a different voting mechanism such as a purely democratic poll. CTT voting is vulnerable to Sybil attacks, where an entity creates multiple identities for themselves in order to unfairly manipulate the outcome. Let’s take a look at why an actor has incentive to conduct a Sybil attack. Take the same example as from before, except this time, we’ll split Bob’s votes into two accounts, each with a stake of 15.
The final outcome ends up being the same, but since Bob’s votes are split, each individual account is less decisive. Each of Bob’s accounts is taxed 5, meaning that Bob pays out a total of 10, as opposed to 20. In closed organizations (like shareholding corporations) where each account can be internally verified, this isn’t a problem because duplicate accounts can be immediately identified. However, if it’s an organization like the DAO, people are free to create multiple accounts as they please. There are two potential workarounds:
- Polynomial verification. Every voter is assigned a point from an arbitrary polynomial. When they go to submit their preferences, they are required to submit their assigned point. At the end, if all the submitted points resolve to the original polynomial, then there are no issues. If there are any duplicate points, then there is a fake account.
- Proof-of-Individuality. In order to determine that everybody has only one account, groups of people get into video calls to ensure that nobody is in two places at the same time.
Once we have a way to ensure that everybody in the system has only a single account, then Athena isn’t just limited to shareholding organizations; it can expand to other, broader use cases.
