[Constitutional] Increase resilience to outside attackers by updating DAO parameters to not count 'Abstain' votes in Quorum.

[cupojoseph]

Increase resilience to outside attackers by updating DAO parameters to not count ‘Abstain’ votes in Quorum.

Constitutional.

Abstract

Increase the resilience of the DAO to outside threats, for example bribe markets, by not counting ABSTAIN votes in quorum. This will make it harder for automated bribe systems to pass contentious proposals, and also easier to defeat spammed malicious proposals which is now a larger threat than at any time the DAO’s existence. We should ask ARDC to analyze the effects this change would have had on all previous votes, and if the number of votes required to reach quorum should also be updated with this potential change.

Motivation

For the first time, an automated bribe market (lobbyfi) has enough delegation to create proposals. This is a major change from only the top handful of delegates, all of which are generally trusted and currently being paid by the DAO, having enough delegation to create proposals.

Rationale

The recent contentious Op-Co proposal barely reached quorum, but was pushed over by abstain votes. If we only counted Yes votes towards quorum it would have passed by a <620k ARB margin (0.5% of quorum required.) For refence, there is ~35M ARB on aave available to borrow. Imagine another proposal created by LobbyFi or another market passing by this much or malicious proposals just barely

We do not want to wait until proposals can be made through automatic bribes from future protocols for the first time, or for contentious proposals which are generally disfavored to pass only because someone abstained at the last minute. It’s time to fight back and be proactive in securing our governance.

Key Terms

Quorum - the number of votes needed to participate in order for a proposal to the DAO to pass. Currently NO votes are not counted towards quorum, but ABSTAIN votes are counted.

Governor - Contract which counts votes and controls quorum.

Specifications

My understanding is this is a setting in the Governor which can not be updated by the DAO, and so would require a contract upgrade to the Governor. If there is enough interest in pursuing this strategy we will try to work with the OCL dev team who originally deployed the Governor to address the technical implementation before any onchain votes are made. It is heavily modified from the OZ Governor and may require their expertise.

Steps

1. Ask ARDC for research help and to publish an analysis on the effects of making this change.
2. Review their recommendation and analysis, and allow time for more discussions, including with the Lobbyfi team to see if they have input as well.
3. temp check
4. Come up with technical implementation of new Governor
5. vote on proposed parameter changes

Maybe 6 - 8 weeks?

Overall Cost

Primary cost here is engineering time and ARDC time, which should already be accounted for. I believe at least asking for this analysis and possibly some worst case situation simulation is worth pursuing. I will defer to ARDC on costs.

All feedback on this strategy and any other ideas to increase resilience are greatly welcome and appreciated. Thank you.

EDIT:
awesome dune dash here:
https://dune.com/entropy_advisors/arbitrum-dao-governance-proposals

[Chris_Areta]

Thanks for bringing this proposal, @cupojoseph, on this very pressing topic in the DAO.

Although I understand the intention here, I think removing the “Abstain” voting option disenfranchises legitimate, well-meaning voters who are genuinely undecided, are neither for nor against, or simply want to remain neutral on a proposal. So, not sure if it’s the best approach.

But it’s definitely true that we need to rethink the topic of quorum as a DAO, as it is looking more likely that the DAO may be susceptible to future governance attacks. To this point, we may want to consider:

• Implementing different (higher) quorum thresholds for proposals that request a certain % of the treasury.
• Moving away from simple majorities and requiring a specific % of YES votes for certain kinds of proposals to pass.

Perhaps the ARDC could also consider some of these options and their impact on DAO defense when conducting their analysis of this proposal and making recommendations.

Either way, we believe the topic of DAO defense mechanisms is worth spending time and resources on for Arbitrum DAO. There is a very large treasury at stake here, and as one of the largest DAOs, we need to be more proactive in safeguarding against governance attacks.

[EzR3aL]

What’s the point in having abstain then if you basically remove it as a valid voting option? There are reasonable times where abstain is signaling that one is not 100% for nor 100% against the proposal. So it might need modification if the majority is choosing to abstain from it.

Removing this doesn’t help “fighting” against these bribe markets.

[Vertex_Protocol]

@cupojoseph

We should ask ARDC to analyze the effects this change would have had on all previous votes, and if the number of votes required to reach quorum should also be updated with this potential change.

Entropy might be able to at least partially help with this instead of the ARDC. Not all of the queries are public right now but they’re already tracking quorum and voting results over time.

https://dune.com/entropy_advisors/arbitrum-dao-governance-proposals

Interesting to see that even though these are arguably some of the most important ones, of the 10 proposals that have passed with a relatively low quorum 8 of them are related to network changes or the security council, even BoLD barely made it. We’re not the first to say this but regardless of whether we decide to keep abstain votes as part of quorum or not, this is probably something we should find a way to get more energy around.

[paulofonseca]

@cupojoseph do you mean to propose to change this just on the Arbitrum Treasury governor (the one for non-constitutional level proposals, that has 3% quorum) or on the Arbitrum Core governor (the one for constitutional level proposals, that has 5% quorum)? or to change it in both governors?

In your proposal, you are linking the constitutional, 5% quorum governor, in here:

@cupojoseph

Governor - Contract which counts votes and controls quorum.

but then you are using the example of the OpCo proposal that used the non-constitutional, 3% quorum governor, in here:

@cupojoseph

The recent contentious Op-Co proposal barely reached quorum

[JoJo]

I don’t honestly see how removing the abstain preserves from gov attacks. Matter of the fact, it allows us to reach quorum in an easier way will still keep participating (and for what it matters if I recall step 1 proposal would have not passed for example without the abstains).

Also, right now, the against are not counted in the quorum. In your example of opco that had a 140M votes vs a 120M quorum, if you would have instead counted the against, you would have had instead a 196M valid votes.

So, really, I am not understanding the logic here.

[lobbyfi]

@cupojoseph

For the first time, an automated bribe market (lobbyfi) has enough delegation to create proposals.

Just to give a little more context on this: LobbyFi is not able to propose anything to the governor by design and will not be able to do this in the future. You are welcome to take a glance at the DelegationsHoster smart contract.

@cupojoseph

The recent contentious Op-Co proposal barely reached quorum, but was pushed over by abstain votes.

Just as a matter of coincidence, LobbyFi voted against on the proposal you chose as an example.

Our comment on the proposal overall, in short points:

• while the problem you described is real, it is not realistic at this point of time (at least given the current delegates scene and LobbyFi willing to cooperate and hence changing our setup to make sure we are helpful with the quorum).
• changing the way quorum is counted will likely make it even harder to reach. Given the current tendency of rising supply and less participation in governance, this cannot be proposed as-is, since it would be counter effective for getting the most of the proposals through that are not malicious.
• we agree with the feedback @Chris_Areta, @EzR3aL and @paulofonseca provided here.

@cupojoseph

created by LobbyFi or another bribe market

And the last point: LobbyFi is not a bribe market and we insist on not calling it so. Just like in a tweet from a project that we suspect you are working on, we define what we do differently than bribing - and you perfectly know it from numerous conversations.

[WinVerse]

Thanks for putting this up, cupojoseph!

This is a very good point.

@Chris_Areta

Although I understand the intention here, I think removing the “Abstain” voting option disenfranchises legitimate, well-meaning voters who are genuinely undecided, are neither for nor against, or simply want to remain neutral on a proposal.

Aside this, what happens to voters who abstain because they benefit directly from a proposal? Should they now be forced to vote in favor or against proposals?

On this,

@cupojoseph

or for contentious proposals which are generally disfavored to pass only because someone abstained at the last minute.

Please do you have statistics backing this up? We don’t see how excluding “Abstain” would help solve this.

The OpCo proposal you linked was indeed contentious but we cannot explicitly say it was generally disfavored as the ratio of for to against vote stood at ~3:1 as we can see here

Finally, building up on Jojo’s comment, “Abstain” does indeed help us reach quorum. Take that out and we might need to reduce quorum generally which is not a good signal for the DAO.

[cupojoseph]

Its not to remove abstain, its to remove its counting abstain towards quorum. Thanks, agree its definitely worth asking ARDC to re-analyze quorum levels - maybe every 2 years or something

[cupojoseph]

@lobbyfi

Just to give a little more context on this: LobbyFi is not able to propose anything to the governor by design and will not be able to do this in the future

awesome feature, this gives me a lot of long term confidence in Lobbyfi as a viable protocol that can work WITH the dao.

[cupojoseph]

@WinVerse

The OpCo proposal you linked was indeed contentious but we cannot explicitly say it was generally disfavored as the ratio of for to against vote stood at ~3:1 as we can see here

the number of voters is irrelevant, the only thing that matters is number of votes. I would cost me about $20 in gas to create 5,000 voters with 0.001 ARB each. We go by total number of ARB voting, which was much closer.

[cupojoseph]

you can still vote abstain to signal. But now we could remove the possibility that an abstain vote makes a vote pass, which is currently how it works.

[cupojoseph]

I never suggested counted No votes towards quorum. Not sure where that came from

The 140M votes counting towards quorum counts 18.5M Abstain votes. The Yes votes only passed quorum on that example by 620k votes. This small amount would cost only a few thousand dollars to borrow for a week from aave. That’s 29X times as many Abstain votes counting towards quorum as the margin that the Yes votes had over the quorum. I dont think thats as optimized as it could be and is definitely a good justification to spending some energy looking into updating params.

[cupojoseph]

We should do some retroactive analysis on both, and they both currently count Abstain votes toward quorum if I understand the code correctly.

[JoJo]

there is a misunderstanding, maybe i was not able to properly explain myself.

What I meant is that, currently, the against votes are not counted toward the quorum while the abstain are.

If you want to exclude the abstain from the quorum (there is a logic in this) you would also likely need to reconsider the against toward being counted; in that case, what I was pointing out is that the proposal you highlighted, opco, would have passed with an even higher quorum.

[paulofonseca]

Yes, both Arbitrum DAO governors count the Abstain vote choice towards quorum.

And for example, in the OpCo onchain vote, if Abstain wouldn’t count towards quorum the proposal would have just very very barely passed, with 122.4M ARB voting For and the 3% Quorum threshold being 121.8M ARB.

[0xTALVO.ETH_MTY]

Hi @cupojoseph, I don’t think it’s a good idea to remove the “abstain” option, as many delegates (myself included) sometimes lack sufficient knowledge on a topic and, while still wanting to participate in the voting process, choose to abstain from voting on certain proposals. This option should remain to allow us to stay neutral and avoid being perceived as negative toward other users’ ideas. Perhaps the solution is to assign a different value to the abstain vote, such as a minimal percentage that still benefits the listed proposal.

Throughout my time as a delegate, I haven’t encountered controversial or questionable proposals. In that regard, I believe those of us who are active and attentive on TG and Calls ensure the security and integrity of the process, making sure that projects ultimately benefit Arbitrum.

[cupojoseph]

This would not remove abstain as an option.

[0xTALVO.ETH_MTY]

As I mentioned before, perhaps reducing the value of the “abstain” vote would be better than not counting it.

[Curia]

Hi, thanks for the proposal!

After reviewing it and considering input from other delegates, we’d like to share our thoughts.
We appreciate the focus on strengthening our governance and protecting against potential attacks. However, we have some concerns about completely excluding abstain votes from quorum calculations. Abstain votes allow members to express neutrality or indecision; if we exclude them, voters might be forced to take a side even when they’re not fully convinced if they want to contribute toward the quorum. Additionally, as @paulofonseca pointed out with the OpCo example, the data doesn’t clearly show that abstain votes are the main issue—especially given the overall favorable vote ratio.

We also analyzed historical data by comparing two scenarios: one where we counted only “For” votes, and another where we counted both “For” and “Abstain” votes. The quorum outcomes for past majority proposals remained unchanged, except for two cases highlighted in the sheet. In one case, a proposal(Arbitrum Hackathon Builder Continuation Program) had more “For” than “Against” votes, but the quorum wasn’t reached, causing the proposal to fail. In another, a proposal (ArbitrumDAO strategic “Off-site” (online) updated proposal) had more “Against” than “For” votes, so regardless of reaching quorum, it still failed. This suggests that excluding abstain votes wouldn’t significantly affect outcomes, but it could make it harder for a good proposal to reach quorum if some top delegates choose to abstain.

Here is the historical data of Only FOR vs FOR + Abstain

We also think exploring dynamic quorum adjustments could offer a more flexible approach. Future governance upgrades might look into dynamic quorum models that adjust based on proposal type or delegate participation rates. such as those requesting significant treasury allocations—could require higher quorum thresholds, while minor treasury allocations might operate with lower thresholds. This added flexibility would build on Arbitrum DAO’s existing governance-security framework.

Thanks again for sparking this important discussion.