Periodic confirmation - design

Thanks @wpb and @Lorimer for explaining your position. It sounds like we agree with the design except for how the rewards are computed.

I think we can split the points of open discussions / where we might currently disagree into the following ones.

  • If the full reward pot is always distributed among active voters, would this overall have a positive effects?

It was argued that it could be beneficial if the rewards per neuron are higher when there are fewer neurons as this could incentivize new neurons to join when this is more needed. This is an interesting input. A potential counter-effect of this could be that if rewards are distributed among active neurons, it is not in the (financial) interest of current active neurons to motivate others to join the NNS. It is therefore unclear if overall this has a positive or negative effect towards the decentralization goal.

  • Is it beneficial to have less rewards (and thus less inflation)? If so, should this be achieved by lowering the reward pot?

I don’t fully understand yet if we mainly disagree on the first point or if you would also be in favor of decreased rewards but just advocate for implementing it differently.

In general, this seems to be a matter of opinion: Different people (e.g., links above) think it is beneficial to decrease the rewards as they are a main contributor to tokens being minted while others might not agree. If rewards should be reduced, it is unclear whether this should be done as suggested in the design or rather by just decreasing the reward pot.

As was mentioned, one idea when introducing the current reward mechanism was that the portion of the sleepers that is now forfeit could be used for other purposes. This might be connected to an upcoming feature where we want to think about how the incentivisation of voting neurons can be built into governance. This will likely also require some form of rewards, that could be taken as a portion from the followers, from a new reward pool, or indeed from this currently unused portion of the reward pool.

All of these choices have different implications for the rewards that are minted overall (and hence inflation), for rewards for the non-sleeper neurons, and for the rewards for actively voting neurons.

We think it would be very natural to separate the discussion about rewards from the periodic confirmation design and rather include it in the larger discussion of how rewards can be designed:

  1. If we implement the currently proposed design, we would not change how rewards work and keep the status quo. The proposed design does not exclude any future direction for rewards (we could still decide on a next step to distribute rewards based on the adjusted voting power).
  2. As illustrated by the thoughts above, the reward topic is closely related to the topic of how we can further incentivise voting neurons.
  3. We think it would be beneficial to conclude this discussion with a decision that allows us to move forward with some implementation while we keep discussing other aspects of governance that can be separated.

For these reasons, we plan to submit a motion proposal that proposes to replace the original motion proposal with the refined design here. We could explicitly mention in the motion that it does not exclude any future direction for rewards, to ensure that we can separately have more in-depth discussion about this aspect.
I will not manage to draft the proposal this week, but plan to do so early next week.

2 Likes

Thanks for providing further clarification @lara. I agree to your proposed path forward. I appreciate that you are thinking about how governance rewards could be used in the future to incentivize additional active participation in governance. That makes sense.

It’s probably a good idea to offer this clarification in the proposal. I hope some day we arrive at a voting rewards distribution that is used solely for governance participation. At this point, I can see how that will take further discussion that is best saved for another day.

For what it’s worth, I think this is true regardless of how rewards are distributed. The more voting power that exists in the NNS, the less everyone earns. It has always been in everyone’s best interest to not promote staking in the NNS, yet so many of us proudly wear that #8YG badge. I haven’t really ever sensed that people hold back on promoting the need and value of having more participants in the NNS.

I personally don’t think our inflation is too high, so I don’t see a need to adopt policies that reduce it. However, I know there are many people who find this to be very important. I would have no issue at all if the NNS made an intentional decision to reduce inflation by reducing the Voting Rewards Function that defines our daily voting rewards.

Thanks again. I look forward to seeing your proposal come out soon.

2 Likes

I think it is excellent, to participate actively is to read the proposals and vote… it is not to follow a neuron and forget to vote forever ! I am glad for the change and I congratulate the team that participated !

2 Likes

Hi all,
the proposal is live now: https://dashboard.internetcomputer.org/proposal/132411
Please spread the word and vote!

Unfortunately we first made a mistake and submitted the proposal incorrectly here. While it has the same text, is hard to read. Therefore, we recommend to reject proposal 132410 and consider proposal 132411 for voting!

4 Likes

Active participation in voting is not necessarily useful participation, just as proof of work isn’t proof of useful work.

Stakers who log in just to vote “yes” on all proposals (or “no”) aren’t necessarily helping the protocol, even though they are collecting a reward. They could be more helpful following someone who knows what they are doing, which is part of the reasoning behind representative government.

I think there are many ways to make active participation more useful to the protocol, and some have been mentioned here, but that probably belongs in a different topic.

1 Like

I think that everyone should receive rewards for voting based on whether they are an active participant in NNS governance. Therefore, I consider the idea of ​​taking away the vote from people who have been inactive for a very long time to be a pretty solid step forward towards decentralization and I completely agree with it. Initially, all neurons followed neuron 28 as the default followee for topic 0. I myself am also a proponent of a one-time reset in this topic, after which one could consider eliminating the “catch-all” exclusion in relation to topics 4 and 14. This would remove doubts about the centralization of the network from the very genesis of ICP. However, I do not think it is good to interfere with the settings of a neuron by automatically removing its followees after 7 months of inactivity. If we assume that this neuron will never wake up, what is the point of it in the case when after some time its voting power will be 0 anyway? If, however, we assume that the same neuron will wake up, for example, after 10 years and suddenly start voting based on the previous following neurons, is this really a problem? It’s good that we intend to promote activity in the management process, I wholeheartedly support this idea. However, I will not contribute to making it difficult for people who potentially return to the community after years to return. If I did not use any social media for a year, and then returned to contact old friends and did not find them there, I would certainly have a grudge against such a service and would not treat it as a long-term solution for lifelong communication when I plan to take longer breaks from using it. Usually if I set something manually and don’t delete it myself, it means it’s meant to stay that way.

That’s why I voted against this proposal.

we reject. don’t like the idea

My perspective is different. If a neuron remains inactive for an extended period, the individual behind it should research whom to follow rather than simply resuming their following, as neurons voting practices may have changed, or the account might have been sold etc… However, it would be beneficial to have a record of which neurons you followed before they were removed due to inactivity.

2 Likes

When implementing this please make it such that if you vote with one neuron and all your other neurons vote through following, then all your neurons are marked as active rather than only the one that cast the original vote.

This can be done by updating the code that cascades the vote to followers to make it mark neurons as active whenever the caller has voting permission on those neurons.

5 Likes

Just for curiosity,

Is it retroactive ?

Hi all, just an update that the motion proposal has been adopted with a clear majority. Thanks everyone for the discussion and inputs. We will share further updates as we start the implementation.

3 Likes

My previous messages in this topic got ignored but that wont stop me.

40 days later and this is not implemented yet ?

Is it the simply the periodic reset that takes time ? ( Honestly that doesnt look complicated )

Or the automatisation of periodic follow of Dfinity’s neurons / sub neurons that takes time ? Still wondering if the target here is to create case to screw/reduce retail reward and optimise dfinity’s.

1 Like

This has gone WAAAY off the rails, and not in a productive manner. I deleted posts with ad-hominem and baseless claims, and after it looked like random posts with no coherence, so I deleted the rest of borderline topic-related posts as well. Almost 50 posts gone :roll_eyes:

Please get back to periodic confirmation

4 Likes

Thanks @Severin to keep the conversation on topic.
As an update from our side: wrt the implementation, we are just picking up this work now. We plan to share the detailed APIs in the coming weeks so that everyone who needs to integrate with the NNS can start with their work as early as possible.

3 Likes

Now that the NNS dapp provides periodic reconfirmation warnings, can I ask if there is any means of establishing if periodic reconfirmation is due for neurons that are not controlled via the NNS dapp (a non-II controller), or community-controlled neurons that have no formal controller?

1 Like

Hi @lorimer,
we summarized a few use cases (including quill, HW) here.

Each neuron has a new field voting_power_refreshed_timestamp_seconds that should give you this information.
IIUC your known neuron (with name Lorimer) has the ID 16459595263909468577. You can then call the NNS governance’s endpoint list_neurons to list this neuron - for example on the dashboard using length 1 and entering the ID as the ID you want to be listed.
This will return the neuron info, incluing the line
voting_power_refreshed_timestamp_seconds = opt (1_739_350_820 : nat64); - which you can then convert to the date/time when your neuron was last refreshed using a converter (like this).

Someone could of course build a frontend to show this with less manual. Maybe this is actually something that would be useful to display for known neurons on pages where they are listed…

2 Likes

Thanks @lara, this is very helpful info

1 Like