[Request for Comment] Quadratic Voting for Space Race 5

A special thanks to @Keegan, @124.eth, @VanArman, and @hex6c (among many others) for their input on this initiative, as well as @chikai, @Lauren-SuperRareLabs, and @EdSRL for all their work on the new Space Race tool.

Summary

I propose that the DAO adopt quadratic voting for the next space race to ensure that the election results more accurately reflect the will of the community while still allowing individual token holders to demonstrate strong support for individual applications.

Abstract

Coming off of Space Race 4 (SR4) we have received clear feedback, both from voters and applicants, indicating that large votes from individual wallets made the election feel less representative of the community’s interests. Ultimately, this is a symptom of the linear voting model (one token one vote) used to determine the winners of the election, which incentivizes voters to allocate all their resources to the application they care most about, creating a zero sum game where the loudest voices with the most resources win.

Whereas the power of each vote in a linear system is equal to the number of tokens allocated, the power of your vote in a quadratic system is equal to the square root of the number of tokens allocated. Under quadratic voting, users could still show how much they support a proposal, however it becomes much harder for the loudest voices (or those with the most resources) to unilaterally determine the results of the space race.

Based on the authority granted to them in the SuperRare DAO’s first governance proposal, the council has the power to iterate on the governance tooling and methods used to administer the space race. I propose that they use this power to implement quadratic voting in Space Race 5. The purpose of this post is to gather community input and gauge support for this initiative. If sufficiently supported (or in the absence of objection), I will submit a proposal to implement quadratic voting in Space Race 5 during the next Council meeting.

Motivation

Spaces play a critical role in defining which artists have access to the SuperRare platform. Because attention is a scarce resource, curation is necessary to maintain a high signal to noise ratio, positive brand image, and a healthy marketplace economy. This means that even though spaces act with individual autonomy, their actions impact the network as a whole. As such, it is critical that we continually work to increase the integrity, security, and accessibility of the Space Race so that the interests of the community are adequately represented. Quadratic voting represents just another step towards that goal, made possible by the amazing work done by Chikai on the new space race tool.

Specification

How does quadratic voting work?

The main difference between quadratic voting and linear voting is that while in linear voting, voting power is equal to allocated token quantity, in quadratic voting, voting power is equal to the square root of allocated token quantity, as demonstrated below:

For additional information on quadratic mechanisms, I recommend reading Quadratic Payments: A Primer, by Vitalik Buterin. For an interactive demo of quadratic voting, I recommend this fun page created by the Economist.

How would quadratic voting affect the Space Race?

Using a .csv export of the Space Race 4 results (available upon request), Keegan and I performed some analyses on how things might have gone differently if quadratic voting were used. Unsurprisingly, quadratic voting would have changed the order of the winners, though there would only have been one difference in who did / did not make the cut.

While the overall impact on which spaces would have been whitelisted is small, the election results are more representative of the community. This is especially visible when you plot the votes cast/voting power against the number of voters for each application. Under linear voting, there is a very weak correlation between votes cast and number of voters, whereas under quadratic voting, there is a strong correlation between vote power and number of voters. This indicates that the results, as determined by quadratic voting, would be far more representative of the overall voter base than they were under linear voting.

How would quadratic voting be implemented?

While the exact implementation will be determined by the product team, I can share some high level, anticipated changes that could be made to implement this functionality. Anywhere that we show the amount of $RARE allocated to each application, we could also show the vote power of that application. In addition, the list of space applications could be sorted by vote power rather than $RARE allocated. While voting, the user could enter the amount of $RARE that they would like to allocate and see how much vote power that allocation would generate. All that said, the product team has ultimate decision making power over how the user experience of quadratic voting is implemented, though they should take all reasonable efforts to ensure it is intuitive and that the conversion between $RARE allocated and vote power is clear to the user.

Benefits

As shown in Specification, implementing quadratic voting results in a more even playing field, increasing perceived fairness in the election, and as a result one could expect increased voter turnout. In other words, if voters know that their vote has a higher probability of influencing the election, they will be more likely to participate. In addition, by decreasing the efficiency of large amounts of $RARE allocated to individual applications, I would expect that we will see far more wallets voting in favor of multiple proposals. This will ultimately result in a more accurate representation of voter preferences.

Drawbacks

Quadratic voting may be difficult to understand for some users, and as such some effort should be dedicated to creating or, at the very least, linking to, educational resources which explain the functionality and benefits of the new system.

In addition, quadratic voting may be vulnerable to sybil attacks in which a bad actor could distribute their $RARE across many wallets in order to increase their vote power. While this is a significant issue in QV implementations such as Gitcoin grants where attackers can collect matching funds, I do not believe the incentive to attack the space race is anywhere near as strong. Even so, the product team is encouraged to investigate means of detecting and preventing sybil attacks through the use of advanced statistics, verifiable credentials, and captchas.

Outcomes

By implementing quadratic voting, we would expect to see increased voter turnout, satisfaction, and representation. These factors will be measured using data from the election, site analytics, and engagement with the community through surveys and regular communication channels. A retrospective report about the use of quadratic voting in Space Race 5 will be released publicly in a timely manner following the end of the election.

6 Likes

Thanks for developing this draft proposal. I think it is a very good step forward and very good method to engage more followers and increase inclusivity (with people with small numbers of rare tokens). I hope you guys will really implement this to reduce the role of whales.
It is indeed true that quadratic voting is not very well known yet and there is still some work to do to explain it well. Economist’s example is already a very good example in this regard.
There is also the risk of abuse, but this can be partly overcome by investigating when ambiguities become visible during the vote and if superrare then reserves the right to withdraw certain submitters.
For example, if a extremely large number of wallets with relatively low numbers of rare that have just been received vote on a specific proposal.
I would introduce the approach the next space race when possible and then evaluate what the disadvantages and advantages are.
You can then do the initial test with a relatively simple software solution and really only work it out nicely graphically at a final go.

2 Likes

I endorse your proposal and invite you to submit it for implementation.

I will list some minor observations below:

  1. Although the intersection of the two top-10 listings (linear vs quadratic voting) is large (80%), notice how the shifts in ranks are sometimes very significant when moving from linear to quadratic voting. In particular, this holds for Moxie – Miss O Cool Girls (it moves from rank 1 to rank 10) and CRITICIZE BY CREATING - Breezy Art (moves from rank 3, in fact very close to rank 2, to rank 8). Ironically (or maybe not), these are also the most controversial proposals, written very hastily, tained by conflict of interest, and securely backed by one or maybe two whale holders of RARE ‘governance’ tokens. On the other hand, Art Innovation, a sound proposal supported by a whole (not whale!) community, climbs up the ranks from 9th to 1st position. The correlation analysis among voting power and number of votes in linear and radical voting confirms these observations.

  2. It would be interesting to know the Kendall correlation among the two listings, and also the positions below 10. If you can share the CSV file of the voting results, I can compute it.

  3. Among drawbacks of quadratic voting (QV) you mention sybil attack. A simple example might be useful to explain this kind of attack. Suppose one voter has 100 tokens and allocates all of them on a single proposal. The corresponding voting power using QV is \sqrt{100} = 10. The voter has theoretically an advantage to impersonate (say) 4 different pseudonymous identities which each vote with 25 tokens of the same proposal, making a collective voting power contribution of 4 * \sqrt{25} = 4 * 5 = 20. The sybil attacker has doubled the voting voice with this trick. Vitalik Buterin describes this problem and proposes a solution: “QV relies heavily on the notion of verifiable, separate human identities, because a community member could multiply her effective influence dramatically by misrepresenting herself as multiple individuals.” In this case, Soulbound Tokens (SBTs) allow DAOs to identify unique users and prevent this issue. I insist that QV is only a partial solution, while QV + SBTs is a viable solution. Incidentally, this example also shows how an honest voter has a neat advantage to distribute their tokens on different proposals, instead of voting for a single one, since their voting voice is thereby magnified. This distribution effect is indeed mentioned as an advantage of QV in the proposal.

  4. I would add a more philosophical advantage of QV. QV creates a virtuous circle among individual and group as a whole. Indeed, if a voter has K governance tokens, it can cast \sqrt{K} votes for their own self-interest. The remaining voting power, that is K - \sqrt{K}, is, in some sense, ceded to the group. Notice that, the more powerful is the voter, the more of this power is ceded to the group, since \lim_{k \rightarrow \infty} K - \sqrt{K} = \infty (apologize for the LaTeX math notation, markdown is not rendering it here). As Steven P. Lalley and E. Glen Weyl put it:

It is a voting rule under which voters who intend to pursue only their own self-interest are guided, as if by an invisible hand, to promote the interests of society

2 Likes

I like the idea of Quadratic Voting, but as @hecx6c states, there are still ways around the voting mechanisms.

This is where Reputation Voting comes into the conversation, where users can establish their web3 reputation through the use of SBTs, enabling non-financialized governance if needed, or some sort of hybrid ($RARE + SBTs).

Through this, users are able to have their voting power adjusted according to their reputation. Reputation includes examples holding history, holding periods, SBTS traits, other credentials held, off-chain activities etc.

I believe QV is one-step closer to a more equitable, democratized governance structure, however it many disincentivize voting. The question is how do we incentivize it?

  1. Through Reputation Passports (SBTs)
  2. Empowering the 95%~ of the community that is not actively engaging in proposals, and/or non-token holders.

I’m working on a project that is solving this- and I’m more than happy to elaborate if this interests the community!

Hey Slimbokk, thanks for your input!

I am definitely interested in talking more about what you’re working on.

I agree around the importance of reputation as being a means of both protecting and incentivizing voting in the space race. I think that we should take a strong look at implementing systems similar to the Gitcoin Passport in future space races, which will provide a strong mechanism for sybil defense. It will also make it easier to reward voters with additional reputation mechanisms.

We are looking very closely at what reputation means within the SuperRare DAO, and are actively researching better tooling for on-chain documentation of network participation. This includes existing opportunities for network participation, such as being an artist or space operator, but also new ones as we build out more touchpoints for DAO contribution, for example through the grants program.

I would love to get to the point where we have a mosaic of solutions for documenting roles and responsibilities within the DAO, network participation, and community engagement which create a clear picture of how invested (in terms of time, care, and effort - not just money) someone is in the ecosystem which can help provide security around quadratic voting processes like the Space Race as well as better incentivization and reputation across the network.

Thanks for your reply Brennan!

I totally agree, we’re in the process of integrating some of the platforms that the Gitcoin Passport currently has. We’ve currently set up Discord + Twitter verification but the more we manage to integrate, the stronger the sybil defense as you’ve mentioned!

Here are a few examples for the implementation of a Badge/SBT protocol.

Tracking invested effort: In other words, gamifying the experience to promote valuable behaviour.

Building reputation: When discussing topics, such as curation, you can see just how much someone has been voting etc.

Permissioned Access: This goes beyond roles on discord to deciding committees based on behaviour and gating those discussions.

Augmenting Governance Power: Some have proposed hard working curators deserved more voting power for their continued dedication. These could enable that.

Permissionless composability: SBTs are on the blockchain, so other protocol could use them as a mechanism for recruitment and services.

Retroactive rewards: Easily see the different types of contributions and run automated accounting on different reward scenarios.

We’d love to explore these ideas further with the SuperRare community and we could aim to have some sort of system for a Reputation-based Space Race early next year?

I’m questioning this idea. Calling one coin = one vote a “zero sum game” seems disingenuous. Elections have winners and losers and that doesn’t change with quadratic voting.

How does diminishing the power of each vote lead to voters spreading out their votes to multiple projects? I would assert the opposite is true. Less voting power means concentrating that power to achieve the same results (my choice wins), leading to limiting choice not expanding it.

I’m also questioning how “the will of the community” and “the interests of the community” are being defined here. Many earned $RARE by supporting the platform long before there was any money on the table. Should the power of our input be diminished to that of people just coming in? Is there a difference between coins earned and coins purchased? Do SR artists earn still RARE coin from sales? Is that a mechanism worth exploring?

Is an “even playing field” the goal here? Is Democracy the goal? I’m not certain it should be. The original artists built something without immediate monetary goals. We are a unique subset in the community and I feel that should be valued. We could just as easily go with one artist = one vote and collectors be damned. But the DAO was set up to value our early contributions. Obviously buying able to buy RARE coin undermines that. But so does quadratic voting, imho. I’m not certain this is the solution, or that it should be immediately implemented.

Hey @Hidden_Forces

Quadratic voting has been discussed frequently in our bi-weekly DAO Townhalls and weekly Meta-governance calls since November. We have a meta-governance call today at 12pm ET, and a DAO townhall tomorrow at 2pm ET, if you are able to attend I would love to dive more into your concerns and make sure your voice is being heard and concerns addressed. If we’re unable to connect there, I’d be happy to DM here, via twitter (@bmulligan56), or on discord (tag me in #dao-chat and we can kick off DMs from there).

In the meantime, I would point you to @hex6c’s excellent post above outlining the main benefits of QV from a game theoretic perspective in promoting network curation as a public good serving a diverse range of stakeholders.

Looking forward to connecting!

A quick update:

I brought this before the DAO council during the December meeting, which took place on Wednesday January 4th. The council voted unanimously to approve the use of quadratic voting for Space Race 5, and work has now begun to implement it on https://spacerace.superrare.com

2 Likes

This is very exciting! let me know if you need any beta testers. QV + a reputation system is quite powerful to avoid plutocracy and sibyl attacks.
In terms of implementation, there are some open source tools developed either by Gitcoin or RadicalxChange: https://quadraticvote.radicalxchange.org/. But not linked to $Rare (of course).
If you need to be put in touch with RxC foundation, let me know @Brennan.
Go #plurality!

1 Like