Periodic confirmation of following is released!

Governance NNS Governance
, ,
22

Yes, and that is the problem - you’re taking away voting power from the security threshold, and make it less expensive to execute an attack on the network.

Let’s “assume” the 10% voting power reduction was 8 year neurons, 20,000,000 ICP.

You just made it 100m$+ cheaper for someone to attack the network.

It’s strange that the foundation argues against Neuron Markets for this reason (feasibility to execute an attack), yet argues it’s a non-issue here.

2 Likes
23

no, that’s just me not understanding something. Thanks :slightly_smiling_face:

24

Fair enough, sorry for interpreting this response as the foundations response to this issue :handshake:

1 Like
25

Indeed this is not a small number, which is expected from what we have seen in the past:

  • We saw that there are a lot of “sleeper neurons” when following was reset for the governance topic. Shortly after the reset the participation for governance was around 40% and only after some time it went up to roughly 60%. You can see an indication of this in the “Voting Participation” graph here
  • This indicates that a large portion of the neurons are “sleepers”
  • those are likely strongly correlated with governance participants that don’t regularly check their neurons and now have adjusted voting rewards as they haven’t taken any of the required actions
3 Likes
26

Since the implementation of this system, voting power has continued to trend downwards.

My initial post shows 45M voting power had been removed at the time of writing.

IMG_7525
IMG_7525828×446 87.7 KB

8 Days later, we find there has been 66M voting power removed at the time of writing.

27

For the other questions, I think it is useful to distinguish the following questions

  1. How much voting power do neurons have (did the change decrease or increase this)?
  2. What would it take to “buy up” 51% of the voting power? (This is related but can also include new players who don’t have a neuron now)

I’ll have to spend more time on the numbers / analysis to provide a more detailed answer, but here some initial thoughts.

  1. I think this was already pointed out by others, but to recap:
  • it might be useful to distinguish the voting power one neuron has and the voting power induced by one neuron (including its followers). A neuron can have more deciding voting power by itself, but have less weight in the overall voting because its followers’ voting power falls away. In this case one can still argue that this is a gain for security.
  • it might also be useful to look at different proposal topics. For example, since following has already been reset for the governance topic, the “sleeper neurons” already fell away for this proposal and following was already more diverse for this topic. Therefore, I expect that the numbers look slightly different here compared to other topics, like protocol and canister upgrades. But of course we’d have to confirm.
  1. Indeed, if the overall voting power considered for proposals is smaller, then a smaller number of ICP is needed to get 51%. In analysis before the feature it was suggested that this number was still large enough to make such an attack unrealistic. Especially also because a large portion of the ICP is still locked in (sleeper) neurons, even if they don’t contribute to the voting power.
1 Like
28

Indeed I assume that a large portion of the neurons that haven’t refreshed until now will not refresh in the next 2 weeks and this number would further be increased until the end of the month.

1 Like
29

Neurons now execute 13% more of the vote than they did previously, so while the voting power did not directly increase, it does increase the proportional vote of each active neuron.

I have done research on this in the past (at different prices & staked quantities), as this is a concern that has been pushed by the foundation in relation to canister controlled neurons <> neuron markets, however, I’m curious why this is worth implementing “barriers” in one instance, and a non-issue in another?

Using DFINITY as an example;

I agree, randomly selecting the topics “Application Canister Management”, “IC OS Version Deployment” & “Node Admin”, we find;

Application Canister Management:
Past Vote Power: 94.8%
Present Vote Power: 87.8%
(This difference is due to the 30M Voting Power that unfollowed the DFINITY foundation & votes individually, had this 6.8% continued to follow DFINITY, they would execute 94.6% of the vote)

IC OS Version Deployment:
Past Vote Power: 94.5%
Present Vote Power: 97.1%

Node Admin:
Past Vote Power: 86%
Present Vote Power: 81.7%

Yet not unrealistic for those who already have voting power - this reinforces the proportional voting power of ‘active’ neurons and reduces the quantity of ICP required to pass a proposal.

If you expect voting power to continue to decline for another 2 weeks, that means we’re half way at 66M - we can extrapolate this to ~132M Voting power being removed from quorum, resulting in a total voting power of 362M.

DFINITYs 99.3M Voting power will then execute 27.4% of the vote, but I could have sworn the foundation sold on us for years to achieve “greater decentralization”.

31

Would you please perform this analysis for all the topics that are included in the Grants for Voting Neurons program? They are listed below. I would argue that IC-OC Version Deployment isn’t really a topic to worry about too much. That topic only serves to slowly roll out to each subnet the IC-OS Version Election proposal that has already been blessed by the NNS, which is where the review really needs to occur and where the voting power matters more. In all cases for the topics below, you should find that the DFINITY voting power has gone down by a reasonably substantial amount. I would be surprised if the data shows otherwise.

IC-OS Version Election
Protocol Canister Management
Subnet Management
Node Admin
Participant Management

1 Like
32

DFINITY was used as an example, but for the sake of staying on topic, is the purpose of your comment to say that it doesn’t matter that we’re removing voting power from the quorum because other entities aside from DFINITY benefit?

Even if I spent my time doing this analysis for you, we still find a situation where active neurons are benefitting through artificially inflated voting power.

The root of my concern is that we should not be removing ICP from quorum, regardless as to who holds what % in each category.

34

I just wanted to see the data. No ulterior motive. I can do it myself later, so don’t worry about it if you are not interested.

35

yes you are right, the data will show it