Proposal: Update quorum and supermajority requirements

[tfns]

Proposal: Update quorum and supermajority requirements

This proposal is expected to appear on Snapshot for voting on 7th of June 2021 at 12:00 pm UTC.

TL,DR

• A thorough analysis of the voter demographics incentivised discussion to improve the current governance.
• The issues identified include, among others, low voter participation, difficulty in achieving quorum for certain types of proposals and inactive voters. Security and balancing concerns were expressed as well.
• The Introduction, Comments, and Conclusions sections present the initial analysis before the last rounds of high DAO vote participation. The Discussion section collects the thoughts of the DAO in the governance forum to make them available on snapshot for an informed vote [3]. An update of the proposal is shown in the discussion.
• The latest voting rounds with high participation were considered in the analysis that derived the final proposed requirements for quorum and supermajority.
• The proposed requirement changes are the following*:

1. Whitelisting proposals: 35% quorum, 66.7% supermajority.
2. Trading liquidity Limit < 1M BNT: 35% quorum, 66.7% supermajority.
3. Trading liquidity Limit > 1M BNT: 40% quorum, 66.7% supermajority.
4. Liquidity mining: 35% quorum, 66.7% supermajority.
5. Pool fees: 20%, 66.7% supermajority.
6. Governance and protocol changes: 35% quorum, 66.7% supermajority.

*These are shown again under the FOR section.

Introduction

The last proposals to whitelist tokens on snapshot have failed despite adequate engagement from voters. An improvement to the voting system is long overdue. Voting data on snapshot for all the wallets with vBNT currently staked in governance was collected [1]. With this data, an analysis of the current voter demographic was drawn to hopefully shed some light into the current voting issues to find adequate solutions for them.

Figure 1 - Active Voters and Voting participation

Comments

1. A vBNT staker in governance was considered an active voter if their corresponding wallet has cast at least one vote since the snapshot is active.
2. Voter participation was calculated as votes divided by total proposals.
3. Voter participation was taken from the moment a first vote was cast.
4. Some inactive voters might have staked vBNT without a chance to vote yet. This fact is further emphasized by the number of “Active but haven’t voted for 1 week” stakers.
5. “Active voters that haven’t staked in x weeks” stabilises around 3 weeks.
6. Average Active vote participation is 30%.
7. A substantial amount of vBNT stakers vote on less than 10% of the proposals.
8. There is a significant number of wallets with less than 10 vBNT staked in governance (Figure 1).

Conclusions

• A debate on the topic was already done and no conclusion came from it [2]. This proposal aims to bring two concrete ones and hopefully spark some debate.
• As a side note, in the future, a minimum vBNT stake in governance should be added to avoid wallets staking small amounts in order to get any community incentive that is given as a part of voting (88mph and ZCN are two examples of this) without contributing to quorum (Point 8).
• A solution to reduce the impact of whales on quorum is to change the vBNT voting power structure. A reduction to the voting power of larger vBNT stakers vs smaller ones could be considered, to level down the voting power of whales. The solution is non-trivial to implement and might incentivise bigger wallets to split their stake into smaller ones to acquire more voting power, especially if gas fees are lowered with coming L2s. Another solution is quadratic voting – however, it brings unnecessary complexity to snapshot. The idea is interesting and should be explored if easier solutions are not enough.
• A recursive vote on snapshot to check inactivity for addresses that have not voted for the past 3 weeks (point 5). Addresses that do not cast a vote on the inactive check are then removed from quorum until they cast a vote or a vBNT restake is made (whichever is simpler to implement really). The tactic can also help reduce the number of active voters with low participation (point 7). The number of inactive wallets holding the most vBNT staked in governance is sufficiently high to consider this as well.
• Quorum on whitelisting proposals can and should be lowered to 30% given the average active voter participation (point 6).
• I believe the last two solutions are simple enough. The simplicity of the solutions will reduce dev time while hopefully hindering the current voting issues. A more advanced solution was suggested by @tenzent [2] and should also be considered:

Supermajority Rule : Whitelisting proposals are still 40% Quorum but, having a supermajority of 80% (FOR) lowers it to 30% for whitelisting proposals. This will counter tactical voting since in order to beat the super majority they would have to vote against rather than abstain. 30% Seems like a good number because it is not incredibly easy to reach but in proposals like LPL the community, with the help of only a few mid-size whales, has been able to achieve it (The biggest one being xBNT which we can count on for consistent and reliable voting). This rule would essentially mean that overwhelming support for tokens will be treated differently than split decisions.

Discussion

After some debate since it was posted, I believe most voters agree with @mbr 's thorough proposal.

1. All Whitelisting Proposals Include a Maximum 20k BNT Co-Investment, Subject to a 30% Quorum and 80% Supermajority Requirement.

There are good reasons for this. The lower quorum requirement makes it easier for an individual, or small group of individuals to pass a proposal. The system’s vulnerability is increased, and the DAO needs to maintain its ability to manage pool sizes. It is not the case that a deliberate attack is necessarily the most pertinent threat - highly enthusiastic communities, and even our own community, can achieve the same damage without any nefarious intent. A hard limit of 20k co-invested BNT immediately following a whitelist proposal forces these new additions to be managed more carefully.

2. Co-investment Increases are Subject to Dynamic Quorum and Supermajority Requirements.

At present, co-investment increases are held to a 20% quorum and 66.7% supermajority requirement, regardless of the co-investment being requested. The low quorum is partly a reflection of the difficulty in whitelisting. If whitelisting pools is made easier, it stands to reason that increasing co-investment limits ought to be harder; the reduced barrier to entry must be commensurate with strengthened security later in the pool’s lifecycle.

1. Co-investment increases up to and including 100k BNT are subject to 30% quorum and 80% supermajority.
2. Co-investment increases over 100k BNT are subject to 40% quorum and 66.7% supermajority.

3. Future Revisions to Whitelisting and Co-Investment Policy is Subject to a 30% Quorum and 80% Supermajority Requirement.

It is conspicuous that it is easier to change whitelisting policy, than passing a whitelist proposal. Strictly speaking, this is a type of governance hack. For example, consider that an antagonist who wishes to pass a toxic token has a reduced path to acceptance by first changing the whitelisting policy. This is not acceptable, and we have an opportunity here to close this apparent loophole.

I also want to address some of the user feedback, as well as the conversation that was hosted on our community call recently. We are absolutely committed to making Bancor the DAO to which all others are compared. This will require a comparable level of innovation as the product itself, and will be prioritized after the development pipeline is less congested. The main pain points are whale control and voter apathy - and I am working on a model to deal with both simultaneously. Rest assured that I am mindful of the challenges the DAO faces, and I am excited to co-author future proposals with the established community leaders here. To that end, the present proposal by @tfns, together with the suggestions provided by myself and others, does not need to be perfect. It just needs to be better than what we have, in terms of both usability and security.

We can get to work on the perfect solution soon.

I’d like to add the following to this proposal:

1. When put up for voting, a proposal shall include the quorum and supermajority requirements for it to pass.
2. The whitelisting, LM rewards activation/extension and co-investment proposal templates shall include a paragraph with the quorum and supermajority requirements for all types of proposal to streamline the proposal drafting process.

Update on the 6th of June

1. After the last 3 rounds of votes saw an increase in vBNT staked in governance due to highly controversial liquidity mining rewards extension proposals, this proposal was put on hold and a new analysis for the last rounds was made.
2. Some debate in the community resulted in broader quorum and supermajority requirements. The title of the proposal was adjusted accordingly.
3. An analysis of the voter demographics before and after the last 3 rounds of votes, that is, before and after the 23rd of May, is in Figures 2 and 3.

Figure 2 - Statistical Analysis of the vote rounds before the 23rd of May and after.

Figure 3 - Average voting % after first participation per wallet staked in governance.

@mbr proposed the following modified rule set after @eldude and @bias’s feedback and the latest voter demographic analysis:

• Whitelisting proposals: 35% quorum, 66.7% supermajority.
• Co-investments < 1M BNT: 35% quorum, 66.7% supermajority.
• Co-investments > 1M BNT: 40% quorum, 66.7% supermajority.
• Liquidity mining: 35% quorum, 66.7% supermajority.
• Pool fees: 20%, 66.7% supermajority.
• Governance and protocol changes: 35% quorum, 66.7% supermajority.

FOR

Update Quorum and Supermajority requirements:

1. Whitelisting proposals: 35% quorum, 66.7% supermajority.
2. Trading liquidity Limit < 1M BNT: 35% quorum, 66.7% supermajority.
3. Trading liquidity Limit > 1M BNT: 40% quorum, 66.7% supermajority.
4. Liquidity mining: 35% quorum, 66.7% supermajority.
5. Pool fees: 20%, 66.7% supermajority.
6. Governance and protocol changes: 35% quorum, 66.7% supermajority.

AGAINST

Don’t update Quorum and Supermajority requirements:

1. Whitelisting proposals: 40% quorum, 66.7% supermajority.
2. Trading liquidity Limit: 20% quorum, 66.7% supermajority.
3. Liquidity Mining: 20% quorum, 66.7% supermajority.
4. Pool fees: 20%, 66.7% supermajority.
5. Governance and protocol changes: 20% Quorum, 66.7% supermajority.

[1] https://duneanalytics.com/queries/18671/37836
[2] Proposal: Modify WL critiera for voting - #14 by tenzent
[3] Proposal: Update quorum and supermajority requirements

[TheOneJM]

Yes!

20% of vBNT has never voted?!? Horrible. This needs to be fixed. It’s currently a bad look for Bancor and we’re alienating crypto communities.

[CryptoR]

Great proposal!
Why lower the quorum to 30%? I suggest lower it to 20%. If i look at the past proposals its already hard to reach 30%. See for example the UBT proposal (28%)

[CryptoR]

Also check the proposal for $DIP (21%). I rather lower the quorom to 20%

[TheOneJM]

20% WL / 15% everything else

[furs]

This proposal has my support. I’m open to lowering it further if things don’t improve but I urge a balance of caution and protocol growth.

[bias]

@CryptoR

Also check the proposal for $DIP (21%). I rather lower the quorom to 20%

Since we are trying to change two variables (quorum % and discounting inactive voters), it’s important that we don’t make too drastic of changes at once. We have a quorum threshold for a reason. We can always lower the % from 30 → 20 later if we decide we need to.

Also when making process changes like this, it’s important to do them because it improves the process, not because it would have captured some arbitrary result (like passing DIP). DIP is a good example because I think there are real concerns over whether it should be WL’ed and the 21% reflects that. You don’t want to lower the net just to capture everything, otherwise why have the net?

@TheOneJM

20% WL / 15% everything else

I haven’t seen a reason to lower the quorum threshold for other types of votes. In general we have no trouble passing them.

Since this “inactive voter removed from quorum” idea is new/needs to be coded, the proposal should include the language that says the 30% threshold gets activated ASAP and the inactive rule gets activated when it’s ready. We’ve needed some kind of change for awhile, no sense in waiting for anything. Overall, definitely pass this.

[tfns]

Definitely agree with @bias. We don’t want too drastic changes at once. Lowering to 30% quorum is fine, let’s push this forward asap and see what happens.

The stats show that active voter average participation is around 30%. Inactive voter % is positively skewed (point 4) so might not be as much. Quorum should NOT be decreased just so that proposals pass. If there isn’t enough interest to vote on a proposal, that also means said proposal is not that important to Bancor as far as the community is concerned.

As for the removal of inactive members, the feature should be quite simple. A clean-up proposal is put up on snapshot affecting wallets that haven’t voted in the past x weeks (currently suggesting 3 based on most recent data). If the qualified wallets don’t vote on it, then their vBNT stops counting towards quorum. Once any of the inactive wallets cast a vote, their status is reset and their vBNT starts counting again towards quorum.

I believe setting up a routine clean-up proposal instead of an automatic booting system gives a chance for temporarily inactive members and abstainers to avoid being ignored in the snapshot quorum calculations. Finally, this feature will also allow us to routinely remove members that have lost access to their wallets for whatever reason.

[tfns]

Here’s a complete picture of the current pool of voters vs votes cast. Click on the image twice to zoom in.

[BasedAnon]

Good work Tiago. I believe a community poll in proposals at Level 1 and/or Level 2 could be used as a temperature check from the community. This would encourage a rudimental level of participation, especially from users who do not participate actively in discussions and the poll results would give some transparency in rest of community’s initial interest prior to a BIP.

Do we currently have a system for delegating voting power as well? If not, this could also encourage indirect voter participation if said voters place more trust in the Proposer’s judgement and would rather grant their voting power to said user at time BIP were to go live.

[pilor]

Can you update the proposal to be more clear about what the immediate actions you are proposing are? It describes a number of conclusions based on the data but is not explicit enough about what actions it is proposing to take based on the results of the vote on this proposal.

If you haven’t nailed down exactly what the proposed actions are yet then disregard.

[eldude]

Does snapshot offer these abilities? Or would this be instead of current dev work?

[tfns]

Was still discussing solutions. I’ve now updated the proposal with two items proposed.

@eldude

Does snapshot offer these abilities? Or would this be instead of current dev work?

Lowering quorum is simple. Disabling vBNT to vote should be easy as well, it’s just a matter of changing how quorum is calculated, all the data needed is on-chain.

[mbr]

@tfns has prepared an excellent discussion. The insight afforded into the general lack of voter engagement is clear; the facts are what they are, and the measures proposed here demonstrate a level of pragmatism that, quite frankly, puts most governments to shame. My faith in this community’s ability to govern itself is renewed every time I check-in on these boards.

It is important to remember that it should be challenging to pass a DAO decision. We are not interested in discussing a policy that effectively passes every single proposal. To my mind, this discussion is about bringing agency back into the hands of those who are consistently participating, and diluting the influence of those who cherry-pick what proposals they vote on, or those who are not voting at all.

This is a reasonable ask in a paradigm where voting is 100% free from the burden of gas fees.

First, I want to affirm the accuracy of the data presented above. You don’t have to take our word for it; If you wish to audit these results, the addresses with vBNT staked on governance are available via the Bancor Dune Analytics page, and active voters can be ascertained directly from Snapshot. The only source of discrepancy is the block height from which the analysis is performed, but ultimately does not affect the conclusions drawn by @tfns.

So we are in an awkward spot. The dead weight of inactive vBNT stakes is frustrating the DAO’s ability to perform routine governance. However, dropping the quorum requirement also comes at the cost of further exacerbating whale control over decisions, and weakens the DAO process overall. @tenzent has raised a potential solution - the drop in quorum is concomitant with an increase in the supermajority rule. I have previously voiced my support of this idea, and I support it still. Therefore, I would like to suggest this be added to the proposal before escalating it.

However, this might not be sufficient. There is an opportunity here to strengthen the governance system overall by introducing some new standards to whitelisting to reflect the lower quorum requirement.

1. All Whitelisting Proposals Include a Maximum 20k BNT Co-Investment, Subject to a 30% Quorum and 80% Supermajority Requirement.

There are good reasons for this. The lower quorum requirement makes it easier for an individual, or small group of individuals to pass a proposal. The system’s vulnerability is increased, and the DAO needs to maintain its ability to manage pool sizes. It is not the case that a deliberate attack is necessarily the most pertinent threat - highly enthusiastic communities, and even our own community, can achieve the same damage without any nefarious intent. A hard limit of 20k co-invested BNT immediately following a whitelist proposal forces these new additions to be managed more carefully.

2. Co-investment Increases are Subject to Dynamic Quorum and Supermajority Requirements.

At present, co-investment increases are held to a 20% quorum and 66.7% supermajority requirement, regardless of the co-investment being requested. The low quorum is partly a reflection of the difficulty in whitelisting. If whitelisting pools is made easier, it stands to reason that increasing co-investment limits ought to be harder; the reduced barrier to entry must be commensurate with strengthened security later in the pool’s lifecycle.

1. Co-investment increases up to and including 100k BNT are subject to 30% quorum and 80% supermajority.
2. Co-investment increases over 100k BNT are subject to 40% quorum and 66.7% supermajority.

3. Future Revisions to Whitelisting and Co-Investment Policy is Subject to a 30% Quorum and 80% Supermajority Requirement.

It is conspicuous that it is easier to change whitelisting policy, than passing a whitelist proposal. Strictly speaking, this is a type of governance hack. For example, consider that an antagonist who wishes to pass a toxic token has a reduced path to acceptance by first changing the whitelisting policy. This is not acceptable, and we have an opportunity here to close this apparent loophole.

I also want to address some of the user feedback, as well as the conversation that was hosted on our community call recently. We are absolutely committed to making Bancor the DAO to which all others are compared. This will require a comparable level of innovation as the product itself, and will be prioritized after the development pipeline is less congested. The main pain points are whale control and voter apathy - and I am working on a model to deal with both simultaneously. Rest assured that I am mindful of the challenges the DAO faces, and I am excited to co-author future proposals with the established community leaders here. To that end, the present proposal by @tfns, together with the suggestions provided by myself and others, does not need to be perfect. It just needs to be better than what we have, in terms of both usability and security.

We can get to work on the perfect solution soon.

[PlacingBetsDefTruth]

Isnt removing inactive on its own enough? Why merge two items at once?

[mbr]

There is no feasible method to remove inactive addresses from governance.

[glenn]

Just wanted to share that there was a good discussion on governance in the last community call. It starts at 11:10 if you are interested in listening in on some of the topics being raised here (@bias participated). Specifically, @mbr went into some details about removing inactive members and why that’s impractical (might sound simple from a surface understanding, see 23:30) but there are some issues (summarize below):

1. It might raised eyebrows if we conduct our DAO in this manner.
2. This will require a new governance contract (development pipeline is quite congested at the moment) and migration of vBNT. The migration portion would require users to migrate their own vBNT and there is a high probability that the inactive users on the old contract won’t actually migrate over and continue to be inactive.

Since we are discussing Whitelisting and Co-Investment policy, I think it is also important for us to consider Liquidity Mining proposals as part of this discussion. On a similar note:

@mbr

The low quorum is partly a reflection of the difficulty in whitelisting. If whitelisting pools is made easier, it stands to reason that increasing co-investment limits ought to be harder; the reduced barrier to entry must be commensurate with strengthened security later in the pool’s lifecycle.

1. Co-investment increases up to and including 100k BNT are subject to 30% quorum and 80% supermajority.
2. Co-investment increases over 100k BNT are subject to 40% quorum and 66.7% supermajority

LM campaigns emit 120K BNT at a minimum over the course of 12 weeks for medium cap rewards. This can go up to 2x that amount (240K BNT) if LPs stay in the pool for the full campaign and get 2x rewards. Since this causes inflationary pressure on the token (inflation can be good or bad and its complex topic for a separate discussion) by emitting 120K-240K per campaign, then I am proposing that LM campaigns should also require to have their quorum raised to 30% with 80% majority.

Some quick numbers on BNT rewards:

BNT Rewards Restaked: 20,474,526 BNT
BNT Rewards Claimed: 7,316,436 BNT
Total Rewards: 27,790,962 BNT
Percent of Rewards Claimed: ~26.32%
Percent of Rewards Restaked: ~73.68%

[eldude]

This seems reasonable - Lowering the Quorum for small coinvestments but maintaining a high on for larger ones.

[tenzent]

This all sounds great, at 20% quorum right now for co-investments numbering in the millions seems way too low. At the same time Whitelisting seems incredibly high and 30 seems to be a good balancing point judging by the size of the biggest wallets currently in the governance contract. I believe overall this should also have a positive effect on turn out.

[Jefe-fintechv1]

Excellent research @tfns, thank you! After reviewing the discussion, I would vote FOR the following combined proposal:

@tfns

• Lower whitelisting quorum limit to 30%.

@mbr

• Co-investment increases up to and including 100k BNT are subject to 30% quorum and 80% supermajority.
• Co-investment increases over 100k BNT are subject to 40% quorum and 66.7% supermajority

This would resolve the DAO’s biggest issue: to get more TKNs whiltelisted; therefore, there is no need to purge inactive voters from quorum at this time.

Instead, the inactive voter issue should be addressed as part of the broader How to increase voter turnout discussion.