Should the DAO Default to using Shielded Voting for Snapshot Votes?

Proposals
1

Edits July 25

  • Added explanations on the potential downsides of private voting based on delegate discussions observed on Telegram. Credit to @Immutablelawyer, @coinflip, @sid_areta, and @thedevanshmehta for these additional perspectives. Thank you as well as those who have commented thus far on this post @JoJo, @jameskbh, @EzR3aL, and @Bob-Rossi. This has led us to update the voting options to include using Shielded Voting for elections only.
  • Linked a 2-month shielded voting trial in Aave DAO in the Background section.
  • Updated the voting options to be: The DAO should default to using Shielded Voting For All Votes / For Elections Only / Against / Abstain

Abstract

This proposal aims to kickstart a conversation around the adoption of Shutter’s shielded voting as a default for Snapshot temperature checks. Shielded voting encrypts votes during the voting period and decrypts them only after the vote closes, mitigating bandwagon effects, voter apathy, and last-minute strategic voting. A Snapshot vote to gauge the DAO’s sentiment will be posted on Thursday, August 8th. Please note, this is an actual temperature check, just to gauge the DAO’s opinion on the subject.

Background

Shutter’s shielded voting differs slightly from a traditional secret ballot where voter’s identity and selection are kept anonymous. In this case, votes are encrypted during the voting period and are decrypted only after the vote closes. Put more simply, during the voting process, a voter’s position is private, and after the vote, positions are made public. You can see who voted, but not their choice or the total VP per option, until the poll has closed (to clarify any confusion please see the demo linked below). Shutter’s shielded voting on Snapshot has been live for over 2 years now and additional details about the implementation can be found on their blog. A live demo of shielded voting can be found on Snapshot.

Other DAOs have run experiments with shielded voting in the past. Results from a 2-month trail period in Aave DAO can be found here. It is important to note that the majority of Aave DAO’s proposals are related to protocols parameters, which require more specific knowledge sets on risk & economics. Therefore, it’s not immediately clear how directly the results from Aave DAO can be compared to Arbitrum DAO.

Rationale

In the 19th century, most western democracies transitioned from public to private voting methods in elections primarily to reduce voter coercion. While voter coercion is not a widespread issue in the onchain governance space, public votes are susceptible to a few other negative phenomenons:

  • Bandwagon Effect: Voters without a strong opinion tend to go with the current majority opinion. With shielded voting, no one can see how the vote is progressing, thus preventing delegates from being persuaded simply by popular choice.

  • Voter Apathy: On the flipside, when those in a minority opinion see that their vote no longer impacts the results of either an election or decisions, they may choose not to vote at all. This can be due to feelings that their vote doesn’t matter or that strategically it is not worth the social capital to publicly dissent against the majority.

  • 11th Hour Voting: With public voting there is an inherent information mismatch, as the last voter has much more available information to act on than the first voter. Currently, there is a strong incentive to wait until the last possible minute to vote. In some instances this is benevolent behavior as large delegates may not want to indirectly create a bandwagon effect, but other times it can be seen as borderline misbehavior when a late vote is used to strategically create an outcome.

To clarify, we are not arguing that Arbitrum DAO is currently suffering from these effects, but they are each possible with the current public system. There are also downsides to private voting that we’d like to highlight:

  • Overall participation may go down if smaller delegates feel like their positions are no longer impactful, as public voting gives those delegates a chance to strongly voice their opinions by voting early.
  • Politically savvy parties who have the ability or connections to coordinate now have an even bigger advantage.
  • Those who choose to publicly signal their positions gain an advantage over a silent opposition, as some delegate may hold a firm position on silence until the vote concludes.
  • Delegates that know how the votes are falling due to backdoor channels can potentially then use residual votes strategically. With public voting methods, this is something that most everyone can do with less effort and connections.

In general though, Shutter’s shielded voting allows Arbitrum DAO to remain consistent with practice in liberal democracies that votes taken by elected officials (delegates in this case) are publicized so that citizens can judge voting records. Since it addresses many of the aforementioned concerns with public voting methods, Entropy felt it was a worthwhile conversation to bring to the DAO.

Specifications

Given there are some advantages in certain scenarios to private votes for temperature checks, Entropy Advisors is kickstarting this conversation by posing the following question:

Should Arbitrum DAO default to using shielded voting for Snapshot votes?

There will be 4 Voting Choices:

  • For all Snapshot Votes
  • For Elections Only
  • Against
  • Abstain

It is our recommendation that proposal authors still be granted the ability to include reasoning for if a vote should be public instead of shielded. If properly justified, we see no reason to force all Snapshot votes to be shielded (at the DAO’s discretion).

Timeline

A Snapshot poll will be posted on August 8th to gauge the DAO’s overall sentiment and allow for ample conversation beforehand.

Additional proposals to follow will include temperature checks for COI policy, delegate code of conduct, and other ideas to improve the DAO’s operational processes. If the vote passes, shielded voting will be incorporated along with the other aforementioned ideas that pass a temp check into a larger proposal. This proposal will look to codify all of the DAO’s new processes into a socially enforceable proposal. After a trial period of the more grandiose unified proposal around DAO Operations, we hope to eventually implement the ideas that have worked best into the Constitution.

It is important to note, that delegates are already free to post shielded votes. In the case that shielded voting is favored by the DAO, more shielded votes are encouraged while this larger proposal is being worked on.

Cost

There is no cost to the DAO as shielded voting is already available to all DAOs on Snapshot.

13 Likes
2

this is a great idea.

Would like to point out that a discussion on this topic, carried by @Bob-Rossi and to which i also contributed, was posted several months ago, and could be useful here as reference.

In general i would agree that shielded is needed, but I would limit this, for now, to only elections (in council, of service providers and others). I am personally not sure if we want to extend this to all voting, because technical snapshots for example do indeed benefit from a transparent vote.

4 Likes
3

This is an interesting topic and I would like to add a few things:

  • In modern democracies, voting for elections is usually private. But voting on the House of Representatives, for passing laws, budgeting and other matter related to how the nation should function/work, transparency is encouraged (with several being broadcasted live), as the representatives (delegates) must hold accountability. While the delegator-delegate relationship should be more close, the most easy way to check alignment is to see the cast vote while the vote is still running.
  • If we are going for this setup, I believe that we need to add a point about delegates not publishing their voting options before the end of the vote, otherwise the whole purpose is defeated.
2 Likes
4

Hey,
we have been doing a poc for this at the Aave DAO probably more than a year ago. Basically with the same reasons @Entropy has mentioned. Many DAO member wanted to give it a try for nearly a month if i remember correct.
The result was that nothing had changed. People still voted simply in favour of the protocol (do not create any harm) or based on their opinion which only the voter knows.

Being that said, votings are now again transparent and visible for everyone.
In the end nothing has changed.

It would still be interesting to see how the Arbitrum DAO would behave but in my opinion a true and honest delegate has to be transparent and stand by his vote. Even if the majority would vote different.

1 Like
5

Appreciate linking this, my thoughts are in there but the tl;dr is I think very strongly we should have shielded voting for at least any election. There is too much opportunity for gamification of voting without it. Below is what I wrote about it that I think still applies today.

Issue #2

Elections ran without the “Shutter” feature allow for voters to see election results before the election is over. Without hiding results until completion, delegates are incentivized to delay and / or change their votes up to the final minutes of voting. This unfairly benefits candidates who gain early support, as delegate voting behavior may change based on who is winning at that moment and this can materially winner / loser results. It will also not give true indication of support for lesser candidates who then may not feel it’s viable to run for future positions.

Proposed Solution #2

Mandate all elections must be run with the “Shutter” feature if on Snapshot (or equivalent if we ever more to another platform)

2 Likes
6

Updates to the proposal based on feedback and discussion from delegates.

1 Like
7

Thanks for the credit!

I am in favor of using it for elections only. It is useful for larger delegates to see how small token voters are reacting to a proposal.

There is also higher participation when it is unshielded since small voters feel like the beginning of a proposal period is the only time they have to actually be noticed, since later on they just get crowded out by the large delegates

2 Likes
8

Interesting proposal.
However, there are a couple of points:

  1. Indirectly, we already see what the main delegates will vote for - this is all openly discussed on the forum and in delegate threads. Therefore, this will not greatly affect the voting.
  2. There are votes for STIP or LTIPP, where we do not know at all how the delegates will vote (those who are not affiliated with the projects), so for these types of voting, secret voting may be relevant.
9

Thanks for kicking off the conversation. Rather than a requirement, I think of this as a change to the default setting.

Shutter unless otherwise explained why in the proposal.

1 Like
10

This is a very interesting conversation that has been discussed and/or implemented in several DAOs. I like the way you touched on both the positives and the negatives. That was great!

Right now, I see this being implemented only on an election level and not covering all Snapshots (at least for now). It should help gauge how well it works for us.

Just a suggestion/question, would it be possible to lock a Forum topic for the period it’s vote is live? This way, delegates don’t drop rationales behind their vote as comments but are forced to wait for the vote to end or post in their personal delegate thread. With the latter, voters could still get a sense of who’s voting what but it would require them to go through several threads.

11

Thanks @Entropy for the proposal and continuous efforts to make the DAO operations better.

We generally agree with the Shielded Voting at least for elections as Devansh pointed out; small voters are likely “encouraged” to vote on the current (at the time of their voting) top selections regardless of their true preferences.

This is very interesting but it would be overkill and a little distractions for some delegates’ workflows (e.g. voting done with a link to their rationales that is currently required for the delegate incentives program). At least, the sum of votings being not visible should be good enough to address the current main issue.

1 Like
12

Hey!

Luis here from Shutter, we’re working with Snapshot on this shielded voting solution using threshold encryption.

Awesome to see Arbitrum Foundation discuss shielded voting! Snapshot and us think real-time voting results should be private per default (just like they are in real world elections) to preserve as much information symmetry as possible during the voting period. I love @Entropy’s summary and rationale for this, nothing substantial to add from our side!

But happy to answer any questions or hop on a call to discuss the upsides, trade-offs and implementation!

2 Likes
13

We absolutely should be using shielded voting by default in every vote.

This is exactly why we should be using shielded voting. The votes of others should not influence my vote, my vote should be based on my opinion about the proposal.

I can see it in my own votes (and I’m sure you can see it too if you notice). We are naturally biased by seeing the results. It’s scary/hard to vote against the herd, and lazy/easy to vote with it.

Exactly.

2 Likes
14

I found out about shutter voting about 8 months ago, actually when participating in some gov workshop at eth istanbul. It sounded pretty solid to me and I was wondering about adoption etc, built small dashboard around that here: https://dune.com/kwizzles/snapshotshutter

I then had some discussions with several people, and most were against it, mostly because of potential coalitions and the inability to spot them and maybe react? maybe that changed now?

Personally I think it could be beneficial to use it with a certain set of proposals. Definetly agree that an open current state of voting influences voters, and also saw that some bigger delegates vote late to avoid that.

2 Likes
15

A small update, this proposal will now be moving forward to a vote next week on August 15th in order to line up with the other conversation on COI and Self-Voting.

1 Like
16

I’ve voted for this proposal. Shielded voting enhances voter privacy, protecting members from potential coercion or vote-buying, which can lead to more genuine and independent voting outcomes. While it may reduce transparency to some extent, the benefits of safeguarding the integrity of the voting process outweigh this drawback. By prioritizing privacy, the DAO can ensure that decisions truly reflect the community’s will, fostering a healthier and more resilient governance system.

1 Like
17

hey @Entropy I’m assuming that you were the ones publishing this snapshop poll today with the 0xb4c064f466931B8d0F637654c916E3F203c46f13 address.

I understand the goal of making this snapshot poll already a shutter vote, kinda like a demo, but that’s actually a little bit too cheeky for my liking…

this violates the current governance practices in Arbitrum DAO and even worse, it kinda jumps the gun on the outcome of the vote. which is not demure at all.

also, you didn’t stated beforehand, that you were going to enable shutter voting for this specific proposal. I think that if you were intending to make this snapshot poll a shutter vote, you should have obviously made it part of the proposal itself.

Also, I’m glad you shared the link to the Aave experiment with Shutter voting in their DAO last year, but it would have been more intellectually honest to actually share the analysis of the impact of enabling shutter voting in Aave. As you can see, enabling shutter voting in Aave for 2 months, reduced governance participation to ~15% of what it was without shutter voting from 13.5k votes on average per proposal, to 2k votes on average per proposal.

IMG_6400
IMG_64001284Ă—1195 164 KB

These numbers are a bit hard to ignore.

and I’m sure you don’t want to have less governance participation in Arbitrum DAO as well,
but proposing enabling shutter voting on snapshot polls for Arbitrum DAO is A, unenforceable (because delegates can always choose to post their votes in the forum, as they are currently doing even in this very proposal above, and nobody will be able to police that), and B, leading to less casted votes overall, and therefor, making it harder to have legitimacy for the decisions we’re making in here.

4 Likes
18

As stated here, voting in favour for elections only.

It would definitely help allievating the “lobbying” problem, which means also newcomers in the dao might have better chances as byproduct of this.
But i am not convinced about other type of voting. As we have seen, most of the discussions on certain topic, for the good or the bad, happen not really in the preliminary weeks/months but only when the vote is live. While I think this is not good, it’s still better than no discussion/engagement at all, and I think shutter vote on non election snapshots would just kill the general engagement for delegates.

1 Like
19

We are leaning towards for elections only. We’re not the biggest fans of shutter and feel it’s caused more confusion and annoyance in DAOs than the benefits it has offered. Overall, this could be useful, but we think only in highly politicized events elections, which we are in favor of testing out.

One key thing here is, before we’re able if we want shutter to be used, we think it doesn’t make sense to already use shutter on this vote, it seems as if it defeats the purpose a little. Why vote if we want to use something if that’s the way we decide whether we want to use it or not. i.e. this vote to decide if we want to use shutter, should be like a normal vote. Nonetheless, not a big deal, but something that’s probably not good to start.

20

Can you please clarify which “current governance practices” you believe we are violating? This isn’t the first time Shutter has been utilized in the Arbitrum DAO [example 1], and there are no explicit policies or discussions of policies across any of the DAO’s channels that prohibit its use. In fact, offering a live opportunity to see how it functions can be a valuable tool for making informed decisions regarding the proposal. We would appreciate it if you could explain in more detail how making the decision to use Shutter might be harmful.

Also, I’m glad you shared the link to the Aave experiment with Shutter voting in their DAO last year, but it would have been more intellectually honest to actually share the analysis

Thank you for sharing the link to the Aave experiment with Shutter voting. However, it’s worth noting that the same link is already hyperlinked within the proposal. Sharing it again doesn’t seem to add anything new to the discussion. We have also previously noted why it can’t be compared 1:1 between Aave and Arbitrum governance.

Surrounding enforceability, this is discussed in great detail both in the proposal itself and in the comments above. Additionally, it is worth noting that we were asked to run this poll by delegates and don’t have strong opinions in either direction. This vote is exclusively to gauge the DAO’s sentiment on the subject.