Vote delegation is not something we should be striving for

Hmm, perhaps there are two ‘options’ for aligning incentives. Anonymity with a matching MKR stake or a publicly known identity that has reputational value that would be lost by voting badly with the delegated MKR. I think there needs to be some incentive alignment though, otherwise we’ll be open to a world of hurt.

I disagree that this is the purpose to delegation. The purpose isn’t to centralise votes into specific hands (trustworthy or no), to me that’s the negative side effect of increasing the vote turnout which is the purpose of introducing delegation.

100% Agree.

Could you develop this point? My feeling is that it could be worse than not voting, but not that it definitely is. My assumption would be:

Fully Decentralized Voting > Delegated Voting > Not Voting

A lock-up period does help with this, it also helps resolve potential problems around lending/borrowing MKR. It’s something that I think we may want to implement at some point.

1 Like

I agree, both make sense! The overall goal should be (imo) to develop a strong MKR holder culture that if you’re not going to vote, research your proxy well and make sure that they have a stake, one way or another.

I spoke too fast, you’re right that it’s not the purpose of delegation – concentration itself is a negative side-effect. But I still think there is a natural specialization where some people, not interested in exposing themselves to MKR risk, would still make very good voters because of their skills.

Can I ask your opinion on liquid democracy? The general idea that you’d delegate your medicine-related vote to your doctor friend and your technology-related vote to your programmer friend (and they in turn would either vote or delegate to someone they trust more) has always appealed to me. That’s 100% where my current stance on vote delegation comes from. If the 2 are different in important ways I’m very interested!

Can you expand on this? I don’t see how it helps.

Absolutely agree, the difficulty is ensuring that incentives are aligned.Prediction markets could maybe help solve this? A delegate could bet that MKR would not be printed in the next year, this way they would be separated from direct MKR volatility, but still be on the hook for decisions leading to inflation… There are probably more insidious ways to harm the system though… Unfortunately being exposed to MKR risk is the foundation of a governance body that is directly responsible for their choices, if you allow these to be separated, it harms the system.

I am pro-liquid democracy. There are lots of sensible ideas in there. I particularly like the idea of being able to undelegate and redelegate as quickly and as easily as you want to. Everyone is unique, and will have their own priorities, these will not always match 1:1 with their delegate or representative. In most legacy representative systems there is limited scope for recalling an elected representative, in liquid democracy representatives (delegates) can be more effectively held to account or replaced.

The recognition that different people have different skillsets and levels of expertise in different areas is also very valuable.

A lock up period prevents evasion of responsibility by voters. I have no doubt that there will be circumstances where MKR Holders make a decision and then we only find out a year later that we absolutely shouldn’t have done it. This can be used to attack the system, a token holder could help to setup a situation that would lead to delayed consequences and sell before the consequences ocurr. This is also the reason that MKR Holders need to understand the whole of the system really, really well: so they can see through attempts like this.

With respect to delegation, I think the idea was that the delegates MKR (in a system where a delegate was required to have their own MKR) would be locked for a time. This would counter their oversized influence with the negative consequence of having their personal MKR ‘on the hook’ to a greater degree.

Why? That is somewhat like saying investing is good… but you shouldnt be able to hire an investment advisor.

In this case, having MKR could be a decent investment. What is wrong with delegating the voting to a known party and keeping the upside?

In an ideal world, everyone would participate and know what is going on. That said, folks get busy.

Very much agreed. I guess my expectation is that most of these problems are solved at the base level by having actual MKR holders collectively careful about who they proxy their vote to, who they lend to, etc… so any separation of concerns is originated by someone who has a stake in not giving power to a bad actor. That may be a naive take that gets thoroughly falsified in the future though!

I see, but that doesn’t help with a bad actor borrowing MKR? What I mean is that the initial concern is “bad actor gets 100 proxy votes, has no stake”. Then we say : must have 1 MKR to receive 1 additional vote power, and the owned MKR is locked. So the bad actor borrows 50 MKR & receives 50 proxy votes. Same voting power, no exposure to MKR.

Thinking more about this, the structure is the same for borrowing and delegation: with borrowing, an actor uses funds to obtain the pure “vote” part of the MKR token; with delegation they use persuasion to the same effect. We can force delegation-based attacks to use a mixed borrowing-delegation strategy, but that doesn’t seem very helpful.

1 Like

It definitely gets confusing when you add borrowing and delegation together. I was thinking about this more, and I think there are some parallels to draw with going margin long on MKR.

When you go margin long, a decrease in price eats into your deposit before the borrowed funds, this could map to voting in such a way that some of the delegate’s MKR is transferred to the delegators if the price decreases while it’s locked in the voting system. This would mean voting with delegated MKR would be a much larger risk.

If the price rises, perhaps the increase in value is split between the delegators and the delegate (In practice this looks like MKR going from delegator to delegate.)

Basically I’m imagining some weird combination of the two problems where delegators become lenders and delegates become leveraged long traders.

I need to think more about it, but I have an inkling there’s a solution in there somewhere.

1 Like

Just wanted to add a short technical perspective.

First, I think these threads are a great way to work through ideas around delegation, its benefits and it’s disadvantages. Perhaps by the end of this we can at least say, “If you’re going to delegate your vote, this delegation model is the gold standard.” To that end, keep the discussion coming.

However, to focus the discussion a little, I want to stress this fact: there is no technical way to stop vote delegation with our current system. That is, even if we are repulsed by the idea, or think it would make our ecosystem unhealthy, we cannot stop it from happening. With this condition in mind, I believe we are best to inform MKR holders that want to delegate what they should be looking for in a delegation system.

Some ideas (some solve the same problem):

  • Only allow a delegate to be in control of some multiplier of their stake. E.g. if a delegate has 10 MKR, only allow and additional 1x, 2x, or Nx stake to be delegated. This would result in total values of 10 + 10 = 20, 10 + 20 = 30, or 10 + 10*x respectively.
  • Lock the delegate’s stake up for some timeframe to ensure their MKR is accountable to the results of their vote.
  • Allow a delegating user to withdraw their delegation at any time.
  • Burn or slash the delegate’s MKR if they verifiably vote against a known model.

If you reply to this list with ideas, I will try to keep it up-to-date. We may want to spin a thread off to discuss the merits of each option at some point.

My current interest in this is centered around the idea of risk teams (or funds) insofar as we may decide to delegate our MKR with a risk team based on their models, voting record, and results. I was particularly captivated by the idea of using a PID controller to drill down on the peg, and I would love to delegate some of my MKR to a PID controller that successfully accomplished this goal.

I share the centralization of power concerns that some of you have expressed in the same way that I am concerned about the effect of gravity on me when I fall: both are of concern, but are immutable laws of the system we’re operating within. To this end, let’s focus on best practices for such a delegate system and hope that MKR holders will not delegate unless such a system possesses those ideal properties. If we find a good mix of healthy properties for delegation we may be able to make using a system with more relaxed constraints as appalling as sending one’s credit card over HTTP or logging into a remote system with telnet.

7 Likes

Are you aware of any MKR delegation contracts already deployed to mainnet?

I have no idea how high on the priority list vote delegation is. But this makes me think that it should probably be pretty high in priority to get a proper vote delegation contract deployed before someone deploys a shitty one and people start using it.

Should there really not be some ERC20 governance vote delegation standard? That the “w3c” of Ethereum created? Assuming this doesn’t exist :slight_smile:

I don’t think I understand this bullet. What would qualify as a known model?

My Solidity/smart contract development skills are so rusty that they are very close to non existing :slight_smile: so please bare with me. But would it not be possible to only allow delegated votes from a specific vote delegation contract/address? Or would that require that we don’t use an ERC20 token to vote but instead maybe create a ticketing system similar to Decred’s where you receive a (non ERC20) vote token/ticket when you lock/stake your governance token? Or why is it that MakerDAO cannot prevent vote delegation is probably what I am trying to ask?

MakerDAO need to allow smart contracts to vote for example for purpose of multisign wallets to vote and at that point there is nothing you can do to prevent votes delegation.

1 Like

@LongForWisdom
Why vote delegation is worse than not voting?
Because it promotes lazyness.
And worse - once somebody gets their hands on MKR votes that do not belong to them the whole idea of 1 MKR = 1 vote starts to deteriorate.

1 Like

Ok - there is most likely nothing we can do to stop vote delegation. That being said I just feel there is something that does not feel right with this. We are not in an ideal world - we are in a situation where just a fraction of MKR is being used for voting.

Like I said in my initial post, I expect delegation to be developed so this wasn’t a post to try stop it. I just want to point out that vote delegation might/will have consequences for the integrity of the voting system and that it may not be something we should prioritise, even-tough people are asking for it. The people asking for it are basically telling us they want to be less involved and I don’t see why we should prioritise facilitating that. Certainly not in this very early stage of the system. It needs more engaged people now, not less.

But again, I might be wrong …

Since it will be built anyway, @cmooney brought up some very good suggestions on how the keep the delegatee honest!

  • Only allow a delegate to be in control of some multiplier of their stake. E.g. if a delegate has 10 MKR, only allow and additional 1x, 2x, or Nx stake to be delegated. This would result in total values of 10 + 10 = 20, 10 + 20 = 30, or 10 + 10*x respectively.
  • Lock the delegate’s stake up for some timeframe to ensure their MKR is accountable to the results of their vote.
  • Allow a delegating user to withdraw their delegation at any time.
  • Burn or slash the delegate’s MKR if they verifiably vote against a known model.

Another question you could start thinking about, let’s say you had voting delegation today, who would you delegate your vote too? You don’t need to name names, I’m not trying to make this into a popularity contest but are there even neutral/reliable options available yet?

I don’t think I understand this bullet. What would qualify as a known model?

I was trying to imagine who or what I would be willing to delegate votes to. For me, this would happen if I was so busy with something else in my life – work, tragedy, etc. – that I could not spend time doing the research required for governance. In this case, I would likely delegate to someone or something I trusted.

I am far less likely to trust a human than a mathematical algorithm, but I would be willing to trust a human or a human informed by an algorithm, if there was some balancing of incentives. To this end, I imagined a delegate informed by a verifiable algorithm. As others and myself have suggested, we may ask a delegate to stake their MKR so that it’s at risk if they misbehave. One option for this would be to have a delegate lock their stake for a period of time where they must suffer the fallout from their actions, another option would be to lock it for a period of time so that those delegating could burn/slash the delegate’s stake if they misbehaved. If the delegate deviated from that algorithm by some unacceptable margin, then those delegating could slash or burn the delegate’s stake. This is what I meant by a known model.

Probably the easiest model as an example would be: If we are breaking the peg by a percent or more high or low, then a delegate should never vote for an SF change that would increase or decrease the SF respectively, as this would further break the peg in the same direction.

In this world, not only do we have a gold standard for a delegate contract, but perhaps a list of rules that would expose the delegate to harm if violated. This obviously works much better for something like single collateral DAI’s SF change than it would for MCD’s collateral approval process. In the former we could even delegate to an algorithm, where in the later case we may want to delegate to a risk team (perhaps even a risk team that could disagree with the one making the collateral proposal).

I hope this makes the thinking behind that one statement more clear.

1 Like

Another question you could start thinking about, let’s say you had voting delegation today, who would you delegate your vote too? You don’t need to name names, I’m not trying to make this into a popularity contest but are there even neutral/reliable options available yet?

I gave a little bit of my answer to this question in the response to @ginworkflow. I would be interested in knowing how others would answer.

I just wanted to say in general that I think technical perspectives are extremely valuable, and that currently we don’t have enough of them. So thanks for taking the time to write this up.

I’d like to comment on this specifically, while there is no technical way to prevent someone from doing this, ‘social’ incentives do prevent it. Those that want vote delegation are specifically those that are least likely to take an active role in developing vote delegation. It requires time and effort to write up.

If all the currently engaged people decide that it’s a bad idea and don’t feel that it’s worth writing the contract, the contract is unlikely to get written. Obviously this is not a hard ‘this can’t be done’ it is more of a soft: ‘this is an uphill battle to get this done’

Thanks for clarifying Planet, it’s true that it does allow people to be lazy, but I’d argue those people would be lazy regardless. It think it’s more about having lazy people’s MKR under the control of actively engaged people.

There’s some truth to this, and I would hate for us to develop delegation and for those people to disappear. As you say, we need more engagement, not less.

Personally if I wanted to delegate I would be looking for someone with the following qualities:

  • A larger MKR stake than myself (that would be locked in a delegation contract)
  • Past votes make sense to me.
  • Active in the community.

This is an interesting idea that I hadn’t considered. Is there a risk of this being done maliciously though? Would you need to enforce a ‘burn everything down’ system, where the delegator can burn a smaller amount of his own stake to burn a larger amount of the delegates stake? Or perhaps an ‘x of y’ voting system within the delegation contract to allow multiple delegators to band together to burn the delegates stake. Is that vulnerable to sybil attacks? Could it be MKR weighted?

It did thanks. It would really be ideal imo if one had a broad choice of possibilities when delegating votes, algorithms, people, risk teams etc.

It would be nice if one could easily see an overview of the voting track record of a specific address/entity. I think I would maybe delegate to an entity with a track record I agreed with.

1 Like

As I said in my original post, I expect delegation, there is indeed no way to stop it but we could create an incentive for people to do their own voting. Something along the lines of Idea: Additional financial incentive to get people to participate in governance voting

1 Like