Way forward on spam - Proposal for tactical fix

Background

  • As mentioned in an earlier post, there is currently no capacity in the NNS engineering team to work on changes to the voting protocol, which would be required for a holistic solution for spam prevention. And it might still take some time until engineering capacity is freed up.
  • On the other hand, over the last few weeks there were growing concerns from the community and the DFINITY growth team regarding the current spam situation.
  • In particular with respect to introducing new developers and users to the Internet Computer, spam is considered quite confusing.

Proposal

  • As a temporary fix, DFINITY suggests to reconsider the (earlier rejected) proposal to temporarily revert voting rewards weights for governance proposals back to 1.
  • We are conscious of the fact that this has drawbacks with respect to decentralization, as it reduces the voting incentives for active participation on governance topics as described here. On the other hand, keeping the spam topic not addressed, hinders the onboarding of new dapp developers and users.
  • Given the current situation and engineering capacity, we believe that temporarily reverting voting rewards weights for governance proposals back to 1 is the most pragmatic solution for a short-term reduction of the incentives for spam.
  • Once the SNS functionality has been delivered, DFINITY will take on the task of proposing a more long term and permanent solution to the problem.

Analysis

  • The proposed tactical fix reduces but does not completely remove the incentive to submit spam proposals.
  • This is because approximately half of the total voting power is not voting on governance proposals after the reset of default following of all neurons in Q1’22.
  • Impact assessment: Assuming 4 governance (spam) proposals, 144 FX proposals, 10 other proposals per day
    • Currently: It takes approximately 110k staked ICP to offset the spam submission costs (4x10 ICP) with additionally gained voting rewards.
    • After the proposal: It takes approximately 920k staked ICP to offset the spam submission costs (4x10 ICP) with additionally gained voting rewards.

Next steps

  • Discussion in this forum.
  • Submission of governance update by end of this week.
  • If the spam is not sufficiently mitigated, we can consider increasing the proposal submission cost in addition.
5 Likes

Probably unpopular opinion but I’m against the proposed change:

  • I don’t believe spam proposals to be a hindrance to legit ones.
  • Recently we haven’t had many governance proposals anyway.
  • Spam gives early and active supporters a nice APY boost, which is much needed considered the price action.
  • Spam proposals combined with the current reward weights have helped increase participation in the NNS.

Considering it’s not hurting anyone exception made for those who haven’t had time to update their neuron settings in over 7 months, a group which wouldn’t have received maturity anyway if Internet Computer Network Status were already implemented, I say it’s better to keep everything as is until a TRUE fix is proposed and can be immediately worked on.

10 Likes

Also against the proposed changes
I see those spams as a temporary solution to reduce rewards to inactive voters, dead owner accounts and lost accounts in order to compensate the people who are involved in the governance. It does not create inflation.
When proposal 55651 get implemented, the spams will died by itself and will exit out those inactive accounts.
A lost account, or dead owner account voting and receiving revards is way worst then those spams.
Until 55651, I am fine with those spams beside the fact that it does not look good. It is only a cosmetic issue for now TMO.

5 Likes

Just an FYI, I almost missed this thread entirely. It might be worth renaming the title (or prefixing it with [PROPOSAL]).

2 Likes

Thank you and good point. I have included Proposal in the title.

2 Likes

Thank you @Zane and @coteclaude, I agree that for experienced users (like you) spam is less of an issue, as you understand the context.

The perceived hindrance is mainly for new users and developers who are interested to join or already joined and are then facing a voting tab consisting of confusing spam issues.

Concerns around that have been growing in the last few weeks and therefore it is suggested to bridge the time with a tactical fix until we have a strategic fix.

1 Like

@bjoernek does the strategic fix be proposal 55651?
Would it take that much work to program and implemement?
I think this would be the permanent fix and solve everything at same time.
I am annoyed thinking the possibility that a dead owner account can vote (having configured his following before) and make a difference for a proposal to pass or not.

2 Likes

Hi @coteclaude, apart from the implementation effort, the proposal 55651 (periodic confirmation of followees) cannot be implemented as is (compare the assessment here), because of its impact on the reactivity of the system. A further enhancement is required.

2 Likes

I think this will be a positive change for the network. I hope this proposal passes.

8 Likes

I dont understand how 55651 reduces the spam? Surely the spam would just continue and those gaining from it would just gain more as neurons got unfollowed.

Is it just jumping to the conclusion that people will do X because they are gaining a greater reward even though 50% of people aren’t doing that right now.

Unlike most people in crypto, I am a big believer in the welfare state. Because opportunities clearly are not equal at birth IRL, and children should not be left behind for no fault of theirs.
Crypto is not the same at all. Everybody who bought ICP knew rewards were based on voting. They knew the whole point of the rewards tokenomics was to incentivise active governance. All they had to do was click a few buttons every few months to ensure they maximised their own rewards. Am I wrong in any of these statements? If not, why are we so concerned about people losing rewards that we try to tinker constantly with the system to equalise it? As I recall, some time ago, we were trying to find a fix for somebody’s grandmum’s tax bill. I just don’t get it. Maybe I’m missing something, could someone explain why Dfinity is acting like a crypto nanny state?

7 Likes

You seem to be using the word crypto nanny incorrectly. The point here is to Fix the defects and upgrade to make things better in the long run. Currently there is only one person creating spam. but if this is not resolved in some way to minimize spam. He alone generated 4x spam per day. So if there are 10 or 100 people doing the same thing, every day we participate in NNS administration will have 40-100 spam messages?? What do you think NNS will be like then?:person_facepalming::person_facepalming::person_facepalming:

So people don’t think I’m always against Dfinity ( :slightly_smiling_face:), I do support this proposal as a temporary measure.

We can even go further as to try to punish the spammers if they have locked tokens. It should be clear that the behaviour is not ok. (just create a proposal to scare them)

It’s not economically viable to make that many spam proposals, the difference between 4 and 8 daily proposals isn’t that big in terms of reward, with 100 daily spam proposals the only benefit we’d have is more ICP being burned.

4 Likes

Every day spam proposals burn 40 ICP, with no effect on inflation. Note how much ICP is burned in cycles. There are days when this number is even less than 40 ICP. Spam proposals - not a bad source of deflation:D
Last months quite a lot of new ICP were blocked in NNS, the number of votes for Governance proposals is steadily increasing, which is positive. We are moving in the right direction. NNS is the heart of IC, we need to educate people, get them involved in voting. If they don’t want to, why should we penalize active participants in favor of inactive ones?

10 Likes

Spam NNS proposals, as Zane points out, are subject to the law of diminishing returns. Also, it is easy to reject spam proposals based on knowing the spammer. The increase in governance rewards was an important step for decentralisation. Otherwise, Dfinity would always be in charge because most proposals are admin related. Why roll back an important advance, even temporarily, when it is causing no real harm?

2 Likes

Proposal 55651 is mathematically equivalent to submitting an infinite number of spam proposals every day without actually submitting any spam proposals. Any neuron owner that does not confirm their Followees would have their Followee selections removed, which means they are at higher risk of not voting. Any neuron that does not vote does not receive voting rewards. The total daily voting reward allocation is fixed, but the allocation to each voting neuron goes up when fewer neurons are voting.

Right now this is what is happening on the Governance topic. People who vote are getting a bigger piece of the daily reward pie because there are a lot of people who are not voting (51% of total voting power) on spam proposals. Proposal 55651 would cause this to happen for All Topics Except Gvoernance as well. However, in that case people who vote would be getting all the daily reward pie. There is nothing left to incentivize spam proposals.

The point is that ICP tokenomics incentivizes active participation in governance and there needs to be some minimum level of active participation in order to receive voting rewards. The only task that would be required is to re-confirm your Followees. It’s like voting periodically for representatives in local or national government. The drastic change in voting rewards for some neuron owners is intended to be a strong motivating factor for people who are not actively participating to start participating.

All that said, proposal 55651 will only eliminate the incentive to submit spam for financial gain. The low proposal reject fee still enables spam for advertisement and announcement.

3 Likes

@bjoernek

Edit: I have cast my vote to Reject this proposal. This comment below provides my explanation.

I’ll admit I don’t like this idea for all the reasons described by others so far. However, I’m inclined to support it. I think DFINITY is in the best position to know if the spam proposals are turning off developers and new users and I trust your judgement on how to best to grow the IC ecosystem. I think attracting these people is in the long term best interest of the IC and this bear market is prime time to build and educate. If the spam proposals are harming growth, then we need a temporary resolution.

I agree that the NNS team has higher priorities at this time, especially SNS. In fact, there are many features that I really hope to see implemented after SNS including proposals 38985 - manual voting, 48491 - neuron indexing, 48623 - compounding maturity, 55487 - sustainability policy, and 55651 - Followee confirmation. These are all past proposals that were approved that arguably should have higher priority than the long term spam prevention. When the NNS team finally has time work on the spam resolution, hopefully they will also focus on additional mechanisms to encourage active participation and a higher degree of decentralization, which we still need in my opinion.

If you are going to change the Governance topic proposal weight to 1 then I recommend you also change the Exchange Rate proposal weight to 1. That way you completely eliminate any incentive for spam for financial gain. There are days (usually weekends) when there are no subnet update or node provider proposals, which are prime times for submitting spam proposals if exchange rates continues to have a weight of 0.01. If seeing 4 spam proposals every day is a problem for growth, then I would think that seeing 10+ spam proposals on weekends is also a problem for growth. I’d rather see that incentive eliminated. Since there have been no quality governance proposals that have been submitted recently, there is no advantage for the governance proposal weight to be significantly higher than the exchange rate proposal weight.

Of course, taking away the higher rewards (35+% annualized return for 8YG) that people have received recently because of spam is likely to turn off a lot of folks as well. I wouldn’t be opposed to a significantly higher proposal reject fee at the first sign of spam for advertisement or announcement. Perhaps a higher proposal reject fee will motivate someone with a brilliant idea to crowdfund the fee and work hard to come up with something that will pass. After all, they will get to keep the fee if it passes since it must be staked in a neuron to submit the proposal. That could be a motivating force to cause people to develop quality proposals.

3 Likes

@bjoernek

Thank you for the prompt and thoughtful response. I 100% agree with and endorse this proposal.

To play devils advocate with my endorsement:

If this proposal passes and the financial incentives to funding NNS spam are decreased significantly, the current spammer could become outraged that his/her income stream has dried up and spam the NNS for some additional amount of time, potentially with much more harmful and/or abusive content.

On the bright side, eventually they will run out of funding - as even whales have limits on what type of content they’d support voting on in the NNS.

How would DFINITY and/or the growth team then view this situation? Would they be able to wait out the negative content due to engineering constraints, or would it become a greater priority?

I actually believe they are. I consider myself an “active” member of the community, but stopped paying attention to my proposal telegram bot b/c it was 4 spam messages every single day. (See my follow up question to @Kyle_Langham below).

This is in large part due to the increased rejection cost proposal.

The falling price action is unfortunate, but it should not be tied to NNS rewards policy decisions unless the security of the IC is at stake. We can have a discussion about ICP price in a separate thread.

I agree the rewards weights have increased voter participation - but is there any data that spam has increased voter participation?

@Kyle_Langham, is there any data at DFINITY that would back this up, or show the current state of active (manual) votes cast versus passive (default) votes cast through followees?

3 Likes

Since the spam has started, there has been an increase in governance participation from about 46% to 49%. As mentioned by @justmythoughts below, this correlation does not prove causation. The trend is visible on the dashboard.