[Proposal] Reducing minimum staking time for expanding governance participation

@kvic utilities nft is the future/ just look at the success of uniswap V3

Sure, it’s not me to decide that, people have many talents and sometimes surprising hobbies.
And I’m sure that a NFT from @dominicwilliams or @Jan would be very popular!

The proposal will get enacted only after taking adequate feedback from community and changes as per discussion.

1 Like

That’s Rejected for me

Any comment on possibly limiting this to nom-governance proposals?

@nikhil.ranjan @diegop

Thank you for starting this discussion on how a decrease in the minimum dissolve delay to receive voting rewards can attract new participants to NNS governance. I agree with that hypothesis, but needed time to study the proposal and model the changes before I wanted to comment. During that review, I concluded that there may be a better way to make the change that I’d appreciate if you will consider. I have documented it in the medium article linked below because it is too much information for a forum post. The article contains a link to a Google Sheets document that I shared that enables you to test the full model, check it for accuracy, and challenge assumptions.

I can definitely support tokenomics changes that lower the minimum dissolve delay to receive voting rewards. However, I want to do it in a way that also makes it more attractive to stake long term and mitigates an increased security risk from enabling voting participation from stakers that are not committed for the long term. I believe the modeling data presented in this article shows that a more holistic approach to these tokenomics changes can be achieve by two simple changes to our current tokenomics: decrease the minimum dissolve delay to receive voting rewards from 6 months to 1 month AND increase the maximum dissolve delay bonus to from 2x to 3.5x. These parameters seem to make it more attractive for very short term staking and for very long term staking.

I look forward to learning what you think about this alternate suggestion.

10 Likes

Wenzel, how would you feel about the reduction in APY / VP that arose from providing a steep curve to short term stakers between 0-6 months to NOT come pro rata across the curve but rather just reduce the starting point of 6 month stakers?

I believe that would solve your issue of unfairly penalizing long term stakers for the benefit of short term stakers. It would add in a steep curve between 0 and 6 months and marginally steepen the curve between 6 months and 8 years

Drawbacks will be added in the next version of the post.

1 Like

The current proposal requires developer time as also the shape of the curve changes.

1 Like

Thanks for this proposal @wpb, especially appreciate your sheets with all the numbers.

I would prefer to keep proposals as small as can be. Could you not have two separate proposals, one for reducing the min voting dissolve delay, and one for changing the max dissolve delay bonus? I think keeping the dissolve delay bonus linear while lowering the min voting dissolve delay is appealing, but I don’t think we should increase the max dissolve delay bonus and incentive 8y neurons even more than we already do today.

The part wrt 51% attacks in your blog post i believe is not exactly right. If you want to do a 51% attack on governance, you would use 8 year neurons, because if you can fully take over control, you could unlock your tokens anyway via a proposal. By boosting the max dissolve delay bonus, you increase the power for 8y neurons, and therefore make it cheaper to do a 51% attack on governance (but i still don’t see it as a realistic threat).

5 Likes

Changing tokenomics (reward vs dissolve delay curve) destroys the assumptions we used to choose a particular dissolve delay. So If we change it, maybe we should also allow changing the dissolve delays of our neurons?

1 Like

I wouldn’t want to implement this change as two separate proposals. In order to add staking options at the low end of the dissolve delay spectrum, I think we need to incentivize the high end of the dissolve delay spectrum. While they could be separate proposals, I wouldn’t want one getting implemented and not the other because that would not meet the objective of the OP. I believe it is a very simple change to constants in a few lines of code and requires no new code, so it seems it could be a quick and easy proposal to implement.

I agree with your point about the 51% attack vector. I only felt the need to say something about it because it is sometimes used as an argument against lowering the min dissolve delay. I just wanted to preemptively put data behind a response that lower DD is not a security risk with either of these proposals.

1 Like

I’m not opposed to someone using the models to test different hypothesis and come up with additional suggestions. The AltProp that I documented in the article is what I came up with as the best solution to meet the objectives of the OP during my analysis, but there could be other good solutions. This forum topic is a proposal by @nikhil.ranjan, so any actionable ideas that are suggested need to convince him to change specific details of his proposal.

I think a shorter term staking option is a reasonable idea to incentivize more people to participate in IC governance, but the returns need to be attractive enough for people to participate. I’m not sure 1.7% for 1 month accomplishes that goal, but 6.4% has a better chance of success.

1 Like

I thought the entire idea behind the NR proposal was expressly not to provide meaningful rewards to short term stakers (less than 6 months).

Rather, the intent was to provide a “taster” of governance such that they learn how it works and stake for longer. It’s buy before you try. One risk with your proposal is that we may have a bunch of folks who actually stake for 1 month.

You’ve mentioned that low rewards won’t be very good (1.7% for a month) but that’s actually the point. It’s not an unintended consequence. It’s just a way for folks to tinker with governance.

1 Like

I invested into the NNS stake protocol because I read a about this article: Internet Computer price today, ICP to USD live, marketcap and chart | CoinMarketCap

May 1, 2022 - Staked, so I am a newbie and lots to understand.

While I believe other parts of ICP seem to be moving forward and making contributions to further this project I have been very disappointed with Governance.

Governance has given followers a bad name that they should be penalised for not actually voting but the only proposal being put up are system team updates that only need to be followed anyway.

Other than Governance calling me a name and wanting to take away my rewards for themself I also see a pattern for lowering other standards that were put in place and it’s more about how to further reward themselves while claiming to make more newbies find staking their investment in the NNS more attractive.

Reality is that no one is investing in anything at the moment, check out the stock market, war, disease and the main thing is the colossal debt racked up by most governments around the world.

This is the perfect time to be staked in this project.

I propose that the followers of the NNS invest further funds to their stake for the future of this project, as I will, wait out this period of fear and the desperation that comes from it, vote NO.

We need an easy to use wallet and easy to use staking to attract people that don’t want to learn how to use nns. at the moment binnance use fund only for shorting ICP.

Hi @wpb
Many thanks for your detailed analysis again! This spreadsheet is very useful for bringing the discussion forward. I would like to highlight one subtle but important point, which appeared higher up in this thread in a discussion between @Kyle_Langham and me:

  • For the impact on voting power and APY we need also to consider whether the neurons with dissolve delay < 6 months will actually vote or not.
  • I made the assumptions that approx only 50% of these neurons will vote, which leads to a minimal impact of 0.25% on overall APY.
  • If I include this assumption (i.e. half the voting power of the 0-6 month bucket) in your spreadsheet it seems to suggest that the impact on APY of the 8 year gang between the current set-up and original proposal is less than 0.10.4%.
1 Like

I would need more information to understand how you arrived at those numbers. However, the model in it’s current form may not lend itself to performing this analysis. It’s built with the same assumption as the dashboard APY reporting, which is an assumption that all neurons will always vote. I think this is a reasonable assumption for this type of analysis since the design intent of the NNS is that all neurons always vote and voting rewards are allocated proportional to that participation.

There isn’t enough information in the API to easily build a model that accounts for voting participation rates on Governance topics. It requires information like how many governance and routine business (e.g. subnet update) proposals are submitted each day and what percentage of each dissolve delay bucket are voting on Governance proposals. I do plan to work on this more complicated model to see if it is possible using information from other API sources, but I don’t think it is necessary for the purpose of this discussion.

For every neuron that does not vote on Governance proposals, their voting rewards go to neurons that do vote on Governance proposals. Hence, if we were modeling the 8 year gang based on actual APY today, the range would be something like 12-30% APY as opposed to 20-23% APY as shown in the Current model. The same goes at the low end of the dissolve delay range. For every neuron less than 6 months that does not vote on Governance proposals, there will be neurons in that same region who get an increase in voting rewards relative to the average because they do vote on Governance proposals. Of course, it’s not that simple because you really need to model the entire data set to understand how lack of voting on Governance in one bucket affects APY in all the other buckets.

So to make a long story short, I don’t think lack of voting on Governance proposals results in a change in the average APY. My hypothesis is that it just spreads out the range of APY for each bucket.

1 Like

To bring my overall point across, I see the value in your proposal:
Instead of introducing a kink in the dissolve delay function you suggest to keep the linear shape and slightly rebalance it in order to not negatively impact the long stakers. This has the additional advantage of simplifying the implementation.
@lomesh and @nikhil.ranjan: What is your view?

In terms of estimating the impact on the long-stakers, I am making the point that we can refine the estimate on how much additional (active) voting power we will add, because the majority of rewards is paid out to active voters (and thus the additional inactive voting power which does not vote on governance topics will not matter so much).

As derived in the calculation above, I estimate that we will add 7mn additional voting power assuming 50% actively voting neurons (which is still a conservative estimate as these neurons probably vote less actively than the global average of around 50%). If I enter this estimate in your sheet in the tab NRProp (cell I65) I derive a much lower impact of less than 40bps on the APY of the long stakers. Can you follow this line of argument (and did I interpret your spreadsheet correctly) ?

2 Likes

It is definitely true that when there are fewer votes cast the voting reward goes up for everyone who votes. The daily reward pie gets split between fewer votes. It actually doesn’t matter if you take away those votes from the new <6moDD bucket, the 8YG bucket, the dissolving bucket, or the non dissolving bucket. The result is the same.

According to the IC-API, there are currently 28mn ICP in nuerons that are dissolving that are less than 6 month dissolve delay (cell H15). They would all vote on all Exchange Rate and all vote on other routine business proposals (e.g. Subnet Updates) since they are following Dfinity by default as configured at genesis. The APY for all of those non Governance proposals would be equal to the results displayed in the models because the models assume all neurons always vote, which is true even for this new <6moDD bucket. It is only for Governance proposals that 53% of total voting power in the NNS is not voting. If you were to subtract this 53% of total voting power from any combination of cells I65, I54, and I25, then you would end up with APY that match all neurons that vote for those Governance proposals. If you then apply proposals weights to the APY for all Governance proposals and the APY for all non Governance proposals for any given day (based on the number of proposals of each type that are submitted), then you end up with the daily rewards that are reported to us each day in the form of maturity.

So to make a long story short again, yes you can subtract away votes from any bucket and increase the rewards for everyone else that does vote. However, I don’t think that represents an apples to apples comparison of what would happen to the baseline case (aka worst case scenario) from our current participation rates if we were to add this new <6moDD bucket to the reward distribution.

All that said, it would be just as valid to think about how many new votes will be added that don’t exist today if NRProp (or AltProp) were successful at attracting new governance participants to the NNS. We can pick any definition of success and add 10mn, 20mn, 40mn, 80mn votes to to the model and see how much the APY prediction goes down for all buckets. If any of these scenarios are accurate, then NRProp in the current form will end up putting more and more voting power in the hands of people who own neurons that don’t have the long term best interest of the IC in mind, which runs counter to the objective of the tokenomics incentives. Hence, this change needs to also increase the max dissolve delay to counter this effect.

This just brings me back to the point that, in my opinion, we shouldn’t lower the dissolve delay that pays voting rewards without also increasing the max dissolve delay bonus enough to 1) ensure that the voting power increase for the longer stakers at least matches the increased voting power of the shortest stakers after recognizing some definition of success and 2) give the longer term stakers a tokenomics incentive to vote yes for this proposal. I also happen to think this proposal has a higher probability of success if the new <6moDD bucket has reasonable rewards, which is why I’m hesitant about the steep dissolve delay bonus slope for the new bucket. However, if we do move forward with a steep slope for the new bucket, then I would recommend a max dissolve delay bonus of 3x (which is 1.125x at 6 months instead of 1.0625x) to help offset the new rewards (this is also consistent with the preference that @dfisher expressed above). I would support this proposal under those conditions, but I would still be hesitant about the probability of success of attracting new NNS participants due to the very low APY for a 1 month dissolve delay.

2 Likes