AGP Discussion: ANV Reward Policy

+1000 to this. Also, congrats on your fisrt time participating in an Aragon community vote!

1 Like

“Although active participation was encouraged, attendance in the assembly was paid for in certain periods, which was a measure to encourage citizens who lived far away and could not afford the time off to attend. This money was only to cover expenses though, as any attempt to profit from public positions was severely punished.”

I have an idea to discourage random voting:

Next to the vote, there’s a text field where people have to write an explanation. The reward is delayed one year, during which time voters can flag accounts that didn’t give a good explanation. Before distributing the rewards, a special unrewarded voting session is held, in which ANT holders decide if the flagged votes are truly bad or not. If they are, the reward goes to the flagger instead of the voter. If they’re not, the flagger loses their reward.

A weak point of this idea is that there could be low turnout for the special voting session, as it is not rewarded. Another weak point is that if there are 100 flagged votes, people would have to vote 100 times and pay 100 tx fees.

But the mere fact of having to give a public explanation of your vote, I believe can make a huge difference.

Edit: in order to prevent bad flagger incentives, the extra reward for the flagger can’t be larger than their own reward.

Edit 2: in order to solve weak point #1, voters only get rewarded if they participate in the special voting session.

It’s not the large holders I’m most concerned about random voting, for reasons you explain well here. It’s the long tail, the “silent majority” of smaller holders who, individually, probably think “I’m a small holder, my vote won’t swing this either way” and would be more willing to vote randomly simply to collect the reward, despite the fact that in aggregate having too many small holders thinking this could be detrimental.

But sure, I will buy the argument that implementing this in a situation where the outcomes have a larger impact will tell us more than implementing it in a situation where outcomes matter less.

I still think it would be interesting to set up a smaller scale experiment in an org like the CFDAO first, if only to see if voters will make themselves aware of and play the meta-game which is: if they get rewards and random vote on a small scale, then it may lead some to the conclusion that they will also random vote if receiving rewards on a large scale, which could prevent rewards from being implemented in the main AGP process. Voters who want this outcome, or don’t care, will random vote, and voters who want to prove that rewards don’t lead to significant levels of random voting will take the reward and vote with integrity.

This meta-game would raise the stakes of voting in the CFDAO beyond simply whether the CFDAO proposals are approved or not, without costing as much or exposing the main AGP process to as much risk as implementing this proposal as-is.

It doesn’t surprise me either! Unfortunately the Aragon Network doesn’t currently have a similar “activating” mechanism like PoS networks do (namely tying up stake to make blocks and earn block rewards). In the future, maybe organizations that use ANT for collateral in their proposal agreements could be thought of as “active” and they will be more likely to vote and steer the direction of Network governance. Maybe regardless of whether we agree or not on the specific implementation, this proposal is pre-mature because there is not even yet a network of “active” participants besides a handful of enthusiasts and full-time contractors, and these rewards are kind of forcing an outcome (higher turnout) that we may see in the future when we actually have something more worth governing and a network of active ANT holders more willing to participate anyways.

I think if we measure “participation rate” as “the percentage of eligible voters participating in a vote” then I predict that the percentage of eligible voters participating in a vote would increase if we impose a locking requirement. If the whole point of taking the extra step and incurring the cost of locking ANT is to be eligible to vote, then it follows that a relatively high percentage of those who lock their tokens will make that cost worth it and will actually vote.

I agree, but I don’t think paying people to vote is the way to get there. I don’t think the incentives are aligned to produce good outcomes. While it wouldn’t be a surprise to me that paying voters would initially be a popular proposal among ANT holders I would urge them to consider the potential corrupting effects of these payments and, if they do decide to implement them, to agree ahead of time what “failure” looks like so later on if failure happens it is easier to agree on discontinuing such payments without it devolving into a big fight between the rational long-term voters and the irrational, greedy short-term voters.

I think futarchy is interesting but limited in its decision-making potential, for example if there are several proposals being considered in parallel the challenge is it is hard to know which proposal is actually impacting the metric being measured. So even though a bettor might bet correctly, there’s not a good way to determine that it was the proposal they were betting on that led to the outcome and not some other proposal (or some factor entirely outside of the governance process). And so for the purpose of _generally _ trying to steer voters to vote well I prefer a more generalize-able tool and have thus put forth Locked Voting for consideration.

(That’s not to say that futarchy couldn’t be useful in some scenarios, just that it is not imo as general a tool as Locked Voting for aligning long-term interests. I am interested in Locked Voting for this purpose for similar reasons that we use vesting rather than giving someone a bunch of shares or tokens up front: we want them incentivized to contribute and contribute well over a long period of time. But we can continue the Locked Voting discussion on the other thread.)

While I don’t think there is a huge risk to doing this at the AGP level for various reasons, I understand your concern. I still don’t think that doing this at a scaled down version with the CFDAO is a particularly useful experiment, but I think that one way to test this hypothesis without adopting it directly for ANT is to simply start a new project with a new token and new business model and apply the model there.

Aragon would support the experiment by making it possible to allocate an inflationary vested reward based on voting participation and ship voting delegation, but otherwise ANT holders would not directly participate or benefit from the experiment itself. Instead ANT holders could buy into the experiment using Aragon Fundraising when it launches. The goal of the project would be to maximize the value of this new token (not ANT), ensuring that the potential rewards and risks are realistically comparable to that of ANT.

I think something like that could work, but is much more of a wait and see what works approach rather than a pro-active approach to improve Aragon’s governance directly.

Delegation could be the minimum “activation” bar. It indicates a user has taken the time to pick a delegate who will consistently vote on their behalf.

This is exactly why I think such a proposal would be beneficial, because right now the process is largely driven by a small subset of ANT holders, most of which are on payroll participating in the process. These rewards would force inactive ANT holders off the bench.

I disagree strongly with the sentiment that we “dont [yet] have something worth governing”, I think the issue is that right now is that it is both difficult (expertise) and costly (attention/transaction fees). If we introduce delegation and rewards, we are addressing both of those issues. It becomes easy to participate by delegation, and the transaction costs and attention required to participate can be offset by rewards.

If we change how we measure participation rate, then I agree that it would likely increase. However, if the goal is to ensure that it is difficult for people who don’t have vested long-term interests in ANT from being able influence proposal outcomes, then we need to care about the absolute participation rate. Because this is what determines how much ANT must be acquired (and potentially written-off) in order to have a decisive vote.

I agree that it would be both useful and interesting to try and come up with pre-determined “failure” criteria… However, I think framing this as a tension between “rational long-term voters” and “irrational, greedy short-term voters” is a bit disingenuous.

I’m not sure what makes you think that this proposal would be primarily supported by “irrational, greedy short-term voters” especially if the rewards (as brought up previously in this thread) are subject to a vesting period?

I agree in general that this is an issue with Futarchy and its applicability to making decisions, but if it is used to augment a separate decision process then I don’t think this is nearly as significant a concern. Going back to the idea earlier in the thread about a bicameral approval process, you could have something like a futarchy decision market coupled with delegated + reward voting. The result is that only proposals that are predicted to have a positive outcome and have broad community support can pass. The two processes together can cover the weaknesses of the either mechanism individually.

I also tend to agree with you that in some ways the locked voting proposals is closer in spirit to that of Futarchy, though I would point out that it has similar flaws. Locking tokens is effectively a bet on the future value of ANT, in order to influence the decision making process. But a user must lock their tokens not just to influence a specific decision which they have a lot of information about, but across all decisions during the lock period. This is risky, and they may not have enough information, so it may be rational not to participate. Much like with futarchy, in situations where you don’t have enough participation/liquidity in the decision market, it becomes easier (and less costly) to manipulate the outcomes of specific decisions.

If the rewards that were given for voter participation where vested over time, wouldn’t this provide similar benefits as well?

That’s not how I would characterize supporters in general, I said if the experiment fails according to some pre-defined criteria then there could be a fight between these factions. If the experiment proves to be a failure, I would be comfortable calling those who push to continue it anyways “irrational, greedy short-term voters” because they show they only care about the rewards and not the actual quality of governance and long-term ANT value. And pushing a failed policy even when it doesn’t benefit one in the short term and could actually hurt them long-term due to rewards vesting would make them simply irrational, not even greedy.

I do believe it would be similar yes. I think vesting improves the proposal by at least incentivizing rewards recipients to stay engaged for a longer period of time.

A few more thoughts that come to mind about voting rewards:

  • Continuing the voting rewards thought experiment: what if rewards are only given to people who vote on the “winning” side? This could make it less profitable to random vote, and still raise the cost of Bad Voting attacks since voters who only want the rewards would wait until late in the vote for cast their votes for the side most likely to win.

  • Continuing my hesitation to implement this sooner rather than later: why not wait until after we have a few votes with delegation to see if delegation alone can help increase participation? If vote delegation gets us up to, say, 20% turnout I think that would be a significant enough increase in the cost of an attack to consider rewards unnecessary.

  • And on that last point, now that voting rewards are being discussed I worry that voters might begin playing another meta-game, which is, they hold out from participating until rewards are implemented. That is, they see we are discussing rewards, they know participation is low, they know that we are trying a few different things to try and increase participation, and they will decide “I won’t start participating until I get my rewards” and once rewards are implemented they will do everything they can to maximize rewards right up to the point that it threatens to actually harm the project. Whereas if rewards were not on the table at all they might at least delegate to someone else once we implement delegation, which would help increase participation and the cost of attack and hopefully the value of ANT in the long run.

This dose have some Futarchic flavour to but it but the incentives are not the same. with Futarchy, rewards are being tied to the metric being optimised, simply rewarding the “winning” side only optimises for less contentious votes.

Voters would be incentivised to vote as late as possible jumping on the winning side, regardless of the quality of the vote. If we take the ‘Bad Voter’ scenario again, this could make it worse. If this behaviour dose appear, the bad voter may end up paying less because other voters simply pile onto his side assuming he will “win”

This seems a little far fetched. This would imply non-active are engaged but actively not participating (in any forum, not just the ballots) thus far which makes very little sense to me. This would mean they have already paid most of the costs (attention) but haven’t already tried to maximise rewards?