Allow all neuron owners to transfer controller of their neuron

Governance
1

Goal

Allow all neurons, including NNS dapp and Quill/Ledger controlled neurons (not just those controlled by a canister) to be transferrable between parties.

This change will put NNS dapp and Quill Ledger created neurons on par with newly created, canister controlled neurons in terms of liquidity, and therefore overall value.

All neurons should have the same features (stake, dissolve, and transfer), regardless of how they were created. Otherwise due to their liquidity, 8 year canister controlled neurons will become more valuable than equivalent 8 year NNS/ledger neurons.

Background

Two days ago, proposal https://dashboard.internetcomputer.org/proposal/130732 was submitted to the NNS. This proposal allows canisters to easily control neurons without HTTP outcalls, making it easy to transfer newly created neurons between parties. The proposal is expected to pass in two days (~early 6/26).

Through platforms like IDGeek, and now WaterNeuron and StakeGeek, ICP holders and developers have shown a strong desire for neuron transferability - so much so that the DFINITY foundation has decided to remove restrictions on canister controlled neurons.

Since the controller of a canister can be changed, neurons that are created and controlled by a canister can be considered liquid neurons.

Liquid Neurons are now more valuable than your NNS dapp or Quill/Ledger Neurons

All else equal, liquid assets are more valuable than illiquid assets because liquid assets provide immediate access to cash, and a reduced risk of loss during volatile market conditions (as are cryptocurrency markets).

Therefore, by preventing NNS dapp & Qull/Ledger neurons from being transferred, over time staked ICP in canister neurons will become more valuable than staked ICP in neurons created through traditional methods (e.g. NNS dapp, Quill & Ledger-controlled).

People staked in the NNS via the NNS dapp or Quill/Ledger include seed investors, developers, and the majority of the ICP community. They deserve the same neuron transfer flexibility as newly created neuron owners.

Proposed Change

Note: The below is just a suggested implementation. The goal of this forum thread is to jumpstart ideas for achieving transfer parity between canister controlled and user controlled neurons.

Suggested implementation (not firm)

Extend the manage_neuron API on the NNS governance canister with an AddController and RemoveController option.

This would be similar to the AddHotKey and RemoveHotKey APIs which allow NNS users to link voting control of their neuron to another principal (i.e. voting in OpenChat via your user canister.

I believe this feature is a fairly simple engineering lift code wise (2-4 weeks), but would also require a tokenomics investigation.

I’ve tagged several members of DFINITY and members of community to jumpstart the conversation, but invite all from DFINITY and the community to participate!

@bjoernek @Manu @dfisher @wpb @WaterNeuron @GeekFactory

8 Likes
2

I would 100% support this if there is no huge security implications. It seems it would be fair to everyone and also give investors flexibility and accessibility. Making ICP more liquid is a good thing for the ecosystem overall. The whole locking for 8 years feel too much like HEX ( 16 years ) scam to me.

We should reevaluate the staking timeline to ensure it actually improves security more so than 21 days like you have on most Cosmos chains and Polkadot.

4 Likes
3

I agree. All neurons should be treated equally as much as possible. Hence, if new neurons can be easily liquid, then so should existing neurons. I’m not a big fan of restricting transfer of genesis and early neurons while allowing new neurons to be easily transferred. If neuron transfer is a security issue, then it needs to be addressed across the board. If it is not, then I’d prefer to see neuron transfer made easy for everyone.

It would be awesome if I could split my neuron(s) and transfer them to new principals that I can assign to my kids. Others may prefer to transfer from NNS dApp control to a Ledger or vice versa. All methods for managing neurons should be supported in the same way.

6 Likes
4

Yes this is the type of as flexibility that we need if it does not affect security. If there is no good reason to restrict something then it needs to be allowed. That is my default position.

3 Likes
5

I would support this in a vote on the NNS. Adding onto @Sormarler @wpb points, allowing the transfer of neurons to one central neuron or splitting them up should be a decision each individual can make. Also regarding security concers, recovery seeds and ledger should be looked into as if you lose your seedphrase or your computer is hacked, it’s over.

1 Like
6

Linking the post where we asked for improvements on SNS neuron transferability from the NNS dapp, so we can improve our rage quit mechanism.

7

Nobody is forcing you to stake 8 years non dissolve no? Everyone has a choice but it comes with a price…

3 Likes
8

If there are no additional responses from DFINITY team members, I plan to submit this proposal to the NNS in the coming weeks. It would be nice to have DFINITY input and perspective before submitting.

Tagging a few additional governance-involved parties that may be interested.

@lara @bjoernek @Manu @diegop

3 Likes
9

While I agree that illiquid neurons represent dead liquidity, the governance risk needs careful consideration before submitting a proposal. I’m not saying we shouldn’t proceed, but let’s take a moment to think it through, especially since having skin in the game is currently essential for securing governance.

1 Like
10

Is it fair that developer controlled and newly created canister-controlled neurons (e.g. Water Neuron DAO, or any technical developer) are currently able to stake ICP and transfer control of that staked ICP at any time, but retail/investor neurons are not able to utilize this functionality?

Especially those that have been staked in this network since the beginning?

1 Like
11

It’s definitely not fair, but I’m unsure how much risk this poses to the governance system.

What if there is a coordinated attack where ICP is gradually accumulated to pass a malicious proposal that includes code changes, and then the attackers dump liquid ICP on the market afterward?

Since the NNS has full control, this could be truly devastating. Unlike other blockchains, where DAOs are a ‘trust me bro’ situation and core developers can simply ignore any proposal, this threat is more significant for ICP.

Currently, with 50% locked (25% for 8 years), investors have skin in the game, and potential attackers can’t easily influence it. However, liquid staking changes all that. I am a huge proponent of liquid staking, but please, some thought needs to be put into it before proceeding.

I have several neurons with different dissolve times, including an 8-year non-dissolving one.

We definitely need to address this because, with liquid staking, it wouldn’t make sense to stake in the NNS.

It’s likely that most people, once their neurons are unlocked, would switch from NNS staking to liquid staking. Therefore, this needs careful consideration and a solution.

3 Likes
12

It’s not fair. I will support as long as it doesn’t cause security issues. If it does, then bsolutely no, even if it’s unfair.

Dfinity folks should chime in and provide some feed back.

1 Like
13

Thank you for sharing your suggestion, @justmythoughts !

As you noted, this thread ties directly into the recent discussion about canister-controlled neurons and the subsequent motion that was approved in May and is now implemented.

The approved motion states: “This proposal recommends lifting restrictions on canister neuron control in the NNS and monitoring their materiality through newly developed metrics. A threshold is set to initiate mitigation measures if canister-controlled neurons exceed 10% of total voting power. Additional measures to disincentive neuron transfers will be considered and implemented if this threshold is reached, balancing security enhancements with user complexity and implementation effort.”

This directly addresses your concern about the relative value of 8-year canister-controlled neurons compared to 8-year non-canister-controlled neurons due to liquidity differences. If canister-controlled neurons become material, steps will be taken to make holding them less attractive.

Conversely, allowing all neurons to become transferable seems contrary to the spirit of our recent discussion, which set explicit limits on the portion of transferable neurons. As also highlighted by others in this thread, general transferability raises security concerns that need careful consideration.

7 Likes
14

I share @justmythoughts sentiment that it is highly unfair to create a two tier neuron system.

I also would like to point out that no institution on earth is going to be happy managing a neuron using a seed phrase, on a ledger device. There’s a reason fire blocks is now one of the biggest companies in crypto. Its the same reason why Orbit is being created…

Eventually I do think real time and attention needs to be given to solve this issue. All neurons need to be able to be transferred to be canister controlled, which is not only the most fair, but also is actually a requirement for ICP to succeed.

At the same time real thought needs to be given as to how to pull this off in a way that doesn’t worsen the security of the entire system.

3 Likes
15

Thanks for the response @bjoernek

I am aware of the 10% total VP monitoring and mitigation measures. 10% of VP ~= 47 million VP.

I simply believe that a “wait and see” approach is missing the oncoming train. It’s already leaving the station, and trying to retroactively implement mitigation measures once 10% is reached will result in a PR train wreck for ICP. Mitigation measures will result in breaking changes, bringing apps to a halt and infuriating users that have gotten used to the new functionalities.

It’s worth mentioning that the current “mitigation limit” amount of VP (10%):

  • Is a significant amount of NNS VP to be transferrable
  • Will compound (continue to accumulate VP, especially if in a staked, 8-year neuron)
  • Should count:
    • Large neurons that are already controlled by canisters via HTTP outcalls and TECDSA. This includes the OpenChat DAO, which I believe has 600-700k total ICP staked in 8-year canister controlled neurons. Other DAOs have also taken this approach - I’d estimate that > 1 million ICP has already been staked in DAO owned canister controlled neurons (>2 million VP)
    • The Water Neuron DAO raise, which is putting the entirety of its VP (~300k ICP) into a canister controlled neuron (split between 6-month & 8-year neurons)
    • IDGeek has already sold over 190k ICP worth of assets associated with internet identities
    • StakeGeek has 25k ICP staked on it
    • Various projects (i.e. Jester Medallions) have already built a collection valued at ~5k ICP involving canister controller neurons
    • I can name 6 additional projects that are shifting to build apps involving canister controlled neurons right now.

So even with canister controlled neuron restrictions just being lifted this week, my estimate is that probably 1-2% of total VP could already be transferrable and controlled by canisters. And much of that VP is 8-year staked and compounding. And why wouldn’t it be staked the max, as those neurons are already transferrable, so there is inherently less risk than traditionally staked ICP.

While it’s easy to wait-and-see with mitigation measures in place, it’s extremely important to think of the PR and how all stakeholders, including the loyal ICP community and investors that support this ecosystem will be affected by mitigation measures and how they will respond.

If there is are serious security concerns about lifting all restrictions, then they should then be raised and mitigation work for transferable neurons should begin now. Otherwise, I see no reason why developers and apps shouldn’t race to be a part of the 10% VP that is transferrable. As far as I’ve read, PoK mitigation measures won’t retroactively apply to already created neurons, so these 10% of the VP transferrable neurons with no drawbacks will become “super neurons” compared to NNS dapp/Ledger neurons, or neurons created after mitigation measures get slapped on.

3 Likes
16

This is a good point, applying the 10% limit on VP for canister controlled neurons is putting a band aid on the problem, it is not solving the problem.

We’ll just need to revisit this entire thread when we hit the 10% limit. Seems very shortsighted to close our eyes and pretend it doesn’t exist.

2 Likes
17

I think the solution is fairly straight forward in principle. Liquid assets should absolutely not endow holders with governance power. That flies in the face of proof of stake.

It’s okay for liquid staked assets to be transferable as long as there was never any governance power associated with the ownership of those assets. This is how WaterNeuron does it. See here for some more commentrary on that.

I think talk of this sort of thing not being fair (whatever that really means :man_shrugging:) misses the point. You staked for governance power and rewards. If all you want is the rewards, then start dissolving, wait… then perhaps wait some more… then restake your capital in a liquid asset that has no direct governance power associated with it (such as nICP). Doing anything else would be short-cutting the staking and governance committment that you’ve already made and would undermine the whole principle that the NNS is built on.

We should frown on any project that seeks to facilitate the transferability of staked assets that endow their holders with governance power. This should be an absolute no no, for obvious reasons. I’d love it if there could be a way of enforcing this foundational principle (rather than waiting and seeing and making ownership of these assets less attractive if needed - that just sounds like a band aid)… @bjoernek

Note that WaterNeuron just gets this right, while StakeGeek just doesn’t (at least currently).

1 Like
18

Pfff tell me you are biased without tell me your are biased.

  1. If you stake your ICP on Stakegeek you give them your vote knowingly. Just like ppl in NNS could do
  2. If you have a sytem that supposedly waterneuron has that still allows user to vote. Then it destroy the incentive for long term commitment because you can sell instantly. Removing also all trust benefits from stacking.

If you think of it carefully the :

  1. would need stakegeek to be a bad actor to have a bad effect.
  2. Just needs you and other people to be the bad actor and destroy the network whilst keeping your stakes intact after a terrible move

Besides it’s a mistake to think that the swap xICP and ICP would leads to the same amount. This can be seen as a protection to avoir huge sales post world destroying votes.

Something waterneuron does not have

So no Waterneuron does not make it right* its at best different

Happy introspection !

1 Like
19

Do you understand the difference between nICP and WTN? The above makes it sound like you don’t.

If you stake your ICP on Stakegeek you give them your vote knowingly

In a democratic system, if I take a bribe (allowing someone to buy my vote), is that okay because I was bribed knowingly?

Just like ppl in NNS could do

How so? Liquid staking isn’t the same as liquid democracy. StakeGeek effectively pay you for your voting power, and they can do with that what they want. The stronger their voting power gets, the more dangerous this becomes. It completely reverses the decentralisation of voting power. They’re also providing no clarity on the long term plans for what to do with this voting power, but the suggestions they’ve put forward so far aren’t great.

would need stakegeek to be a bad actor to have a bad effect

There are no completely good and completely bad actors. There are only shades of grey.

Just needs you and other people to be the bad actor and destroy the network

Sure… that’s the point. :slight_smile:

whilst keeping your stakes intact after a terrible move

Again, sounds like you don’t understand the difference between nICP and WTN (you have to stake the latter for governance power, and it’s not liquid).

tell me you are biased without tell me your are biased

Of course I have biases. We all have our biases. Some are more justified than others.

2 Likes
20

Following our little chat in MP ( and thank you for your time )

It seems that there are misunderstandig of the use of nICP and WTN

For what I understand people holding both token could simply be completely different people, separating efficiently the value of the #ICP and its voting power. Since the WTN distribution is organised in a halving manner it also gives a quite centralised voting power for the early contributor.

Also I am not sure where the yield is from, in the beginning from those SNS ICP treasury in 6 months neurons and 8 years. But then, when they will mint more nICP there will be more ICP stacked to give better yield. So Basically all voting power in the iterations of future funds raise from the protocol will go from ICP stacker to EARLY WTN holders disproportionnaly. Not sure if thats improving any decentralisation.

1 Like