TVL - The reason why only developer canisters, and not investors can control neurons

Governance
1

For the purposes of this post, “DFINITY” or “the foundation” refers to the DFINITY foundation’s interests, and not necessarily those of the ICP protocol or community.

TLDR (Opinion)

I believe the current reason that the DFINITY foundation has chosen to not pursue transferrable NNS neurons is TVL, and not a 51% attack, as has been stated in the past.

TVL, or Total Value of ICP locked in the NNS has a significant impact on the price of ICP. DFINITY is the single largest holder of liquid ICP.

If existing investors are allowed to transfer their neurons, this could potentially put downward pressure on TVL, locking up less ICP and making it harder for the foundation to sell its ICP at a good price.

What is TVL, and how is it affected by canister controlled neurons?

TVL, or Total Value of ICP locked up in the NNS, is a key goal for the DFINITY Foundation. A higher TVL means less liquid ICP.

Less liquid ICP == increased price of ICP == good for the foundation (DFINITY moves an average of ~720k ICP per month, or 5 million USD equivalent).

Removing canister controlled neuron restrictions has, in the short-term boosted TVL by locking significant liquid SNS treasury ICP in either SNS canister-controlled ICP neurons, or in the WTN DAO. At this point in time, the WTN SNS and nICP liquidity bootstrapping event has locked up close to 1 million ICP.

If DFINITY allowed existing investors to control and trade their NNS neurons, this would provide an alternate avenue for people to buy ICP. While benefiting NNS investors, transferring NNS neurons would counteracting the TVL goals of the foundation - locking up less ICP and making it harder for the foundation to sell its ICP at a high price.

Background

In late June, DFINITY introduced a proposal to relax canister controlled neuron restrictions, voting later to pass it (with 92% of NNS VP) right before Waterneuron’s SNS Swap (swap started 6/27, ended 7/3).

At that same time, I published a proposal to allow NNS neuron owners to transfer controller of their neuron. I strongly believe that solely providing neuron transferability to canister developers, and not to existing NNS investors, creates a two-tiered system wherein canister controlled neurons become more valuable than “legacy” neurons controlled through the NNS.

Impact of allowing canisters to control neurons: Since the relaxation of canister controlled neurons and the launch of Waterneuron, almost 1 million ICP has been locked up in two canister controlled neurons, both controlled by the Waterneuron DAO over the past month. Many in the ICP community now exclusively choose to invest stake ICP to nICP instead of the through NNS.

The 51% Attack

At Genesis DFINITY mentioned a 51% attack as the primary reason why canisters were originally prohibited from controlling neurons.

DFINITY Does Not Believe a 51% is Likely

DFINITY has made several changes to neuron transferability and voting power over the past year, suggesting they believe that a 51% attack is not immediately likely or a high risk outcome in either scenario. These changes include:

  1. Relaxing restrictions on canister controlled neurons, with a suggested threshold that could trigger mitigation measures if canister controlled neurons exceed 10% of VP. These mitigation measures are “difficult to implement”, and so they aren’t be considered immediately important (until the threshold is hit or an issue arises)

  2. Introducing a Periodic Confirmation proposal that reduces the VP of inactive neurons. If implemented, this proposal is expected to reduce the passive VP following DFINITY by a larger amount than other neurons that vote on all topics. DFINITY currently holds 92-99% of VP on all topics other than governance and SNS, and some estimate that the upcoming Periodic Confirmation proposal will reduce that to ~80%.

Needless to say, we are 3 years past genesis and the only body holding enough effective voting power to make a 51% attack is DFINITY - by far. Technically, by holding and voting with over 90% of effective voting power, one could state that DFINITY has been making “51% attacks” on the network several times a week since genesis.

DFINITY Does Not Believe a 51% is Likely…from a non-DFINITY entity.



So then why does DFINITY choose implement certain neuron and VP transfer proposals, but deflect NNS neuron transferrability, when the only entity with enough internal NNS voting data to make informed decisions on the subject is…DFINITY?

TVL

If you talk to anyone within DFINITY, TVL, or total value of ICP locked in the NNS, is one of the foundation’s top growth goals.

From a distance it makes sense - locking up value makes the protocol securing that value more valuable.

In the specific case of canister controlled neurons and the Waterneuron project, by offering liquid staking to new ICP investors and not original or legacy neurons, this presents a favorable opportunity for investors to choose to stake in “transferrable” ICP. Many that were holding liquid ICP have locked their ICP, and/or are dissolving with the intention of moving that ICP to canister-controlled ICP.

More ICP locked up == increase in TVL.

While it would be easy for DFINITY to update the NNS to allow existing “legacy” investors to transfer neurons, allowing this would provide an alternate avenue for people to purchase ICP via NNS neurons.

If DFINITY were to provide more flexibility for original long-term NNS investors, this would counteract the TVL goals of the foundation, locking up less ICP overall.

As DFINITY is the largest holder of liquid ICP, less TVL makes DFINITY’s ICP holdings less valuable.

Less ICP locked up == downward pressure on TVL == harder for the foundation to sell its ICP at a high price.

New Investors don’t need “Skin in the Game”

Source

If anything, legacy investors, or those that locked their ICP in the NNS and supported DFINITY since genesis have the most skin in the game. Those who convert their ICP into nICP via Waterneuron receive rewards while retaining tokenized transferrability. And those who hold WTN receive both NNS ICP rewards, and amplified NNS voting power through fractionally shared control of how the 8 and 6 year WTN neurons vote.

Questions for DFINITY

What blockers exist in the way of supporting transferrable NNS neurons?

Given that DFINITY controls 92% of effective VP, is there any data showing that transferrable NNS neurons pose a risk to the NNS? Is this risk greater than the combined implementation of canister controlled neurons + periodic followee confirmation?

Just like with canister controlled neurons, can DFINITY implement the feature but with mitigation measures? For example, put in place monitoring to see how much VP is transferred by NNS neurons. If this exceeds a certain desired threshold, turn off the feature.

1 Like
2

There seem to be numerous misunderstandings in this post. Perhapse the biggest one →

Where did you get this information? It’s a proof-of-stake system. It would be rediculous if it wasn’t. You have to stake WTN for it to provide you with any voting power. WTN is also becoming very expensive, and this will likely grow with VP.

If your stake is transferable, it’s not really a stake.

If you’re not a fan of proof-of-stake, what are you suggesting should replace it? Proof-of-?

1 Like
3

Appreciate the correction. I removed the this part about transferrable WTN voting power stake, since in fact you do need to stake WTN.

However, from what I can see on the ICP Dashboard the longest amount of time people stake WTN for is 3 years. This is much shorter than an 8 year ICP neuron.

So WTN stakers get an amplified VP share, for a lower staking time commitment. Less skin in the game for rewards with nICP, and less skin in the game for greater governance impact with WTN.

nICP is a wrapped transferrable stake. Technically, those could each be individual canisters that control a stake and are transferrable, but that solution would not allow for tokenized subdivision of staked ICP.

I’d prefer for the topic of conversation in this post to stay on why only canisters can control and transfer neurons (via controller), and why this staked transferability is not available to genesis investors who have staked for years. I appreciate and welcome corrections like the above, but please DM me if you want to discuss WTN price/tokenomics.

1 Like
4

Not all stake endows the staker with voting power (such as nICP). It’s important to draw this distiction in this discussion about transferability, and when it is and isn’t okay.

Are you able to point to an example of a transferable canister controlled neuron? Could you clarify what you mean - what would the transfer look like / how would it take place?

5

I’m glad this question has been raised, as it warrants a detailed discussion. With 46% of the supply currently locked and unavailable for DeFi, finding a secure method to utilize it for liquidity would significantly enhance DeFi’s.

From a financial perspective, staking through WTN is definitely more advantageous than using NNS.

2 Likes
6

When a developer deploys a canister to ICP, that canister is controlled by the developer’s principal. Any principal listed as a controller of that canister can add new controllers, or even remove themselves as a controller.

Now that canisters controlled neurons are explicitly allowed by the protocol, anyone can:

  1. Deploy a canister
  2. Send ICP to that canister’s ledger address
  3. Use a library like Neuro to create a neuron controlled by that canister
  4. Stake ICP in that neuron, and manage its stake

Note that before this restriction was lifted, a few teams such as OpenChat were able to get around this with tECSDA + HTTP outcalls. OpenChat has a significant portion of its ICP staked in a canister controlled neuron controlled by the root canister of their SNS DAO. Additionally, the Jester Medallions project involved transferrable, canister controlled neurons.

Developers have been able to create canister controlled neurons for some time now. It works by simply transferring the neuron by changing the controller of the canister. There are other teams that have set this up (silently) already.

A NNS/DAO controlled/blackhole canister contract can act as the intermediary to remove trust assumptions, but it’s quite simple to create a transferrable neuron marketplace by facilitating the exchange a canister’s controller(s) for ICP value in return.

7

Edit: Just added this part. Perhaps there’s an incremental approach that can work for investors while implementing monitoring as a measure of security. From talking to others this past week, I’m convinced transferrable NNS neurons could unlock additional liquidity for DeFi, and open up new opportunities such using neurons as collateral for loans.

8

From what I remember of past discussions, canister controlled neurons that are ultimately controlled by a DAO are viewed much more favourably (in terms of the security they offer), and are therefore less likely to need restrictions placed upon them in the future (if at all). For example, if some members of WaterNeuron wanted to transfer the 8yr neuron and 6mo neuron to some very high bidder, how easy do you think that would be?

When it comes to canister controlled neurons (with significant VP) that are not controlled by a sufficiently decentralised DAO, I suspect it’s very likely that restrictions will be put in place to mitigate potential for governance attacks. My understanding is that this is being considered, and will happen if the need becomes apparent.

Instead of talking about removing restrictions for everybody, the discussion (in my opinion) needs to continue to be about establishing a means of detecting and limiting transferable governance power held by individuals. I think the individual (vs group/community) element here is key.

Unless I’m mistaken, community controlled neurons have been transferable for years (requiring community consensus first). Is that not correct?

1 Like
9

@justmythoughts : We have previously discussed the topic of neuron transferability, including the pros and cons of various options. I appreciate your differing viewpoint, based on which you derive at a different solution design.

However, I would like to point out that the foundational assumption of this forum thread is incorrect. DFINITY has consistently emphasized, both in discussions about neuron control restrictions and in earlier posts on governance, that NNS staking and neuron non-transferablity is intended to incentivize long-term thinking in voting behavior. It is also clear from the design of the proposal on the revaluation of neuron control restrictions that the materiality of transferable canister-controlled neurons has been a key aspect — hence, the implementation of corresponding monitoring metrics. Finally, I would like to highlight that the proposal on neuron control restrictions was decisively approved in May 2024, with 56.8% in favour and only 2.4% against. Given this clear mandate, I am disinclined to reopen the discussion on the same topic.

2 Likes