Dear ICP Community,
“Liquid staking” is one of the key innovations that has fueled the success of competing layer 1 DeFI ecosystems by freeing up huge amounts of liquidity that would otherwise be locked away and inaccessible for use as collateral/financing/swaps etc.
As I will argue, I believe ICP stands to benefit greatly by introducing a well designed liquid staking protocol. However, given the unique characteristics of ICP’s governance and staking mechanism, the kinds of approaches that have succeeded on Ethereum and other layer 1’s would not work on ICP. This post outlines a unique, conceptual approach to liquid staking that I have designed with the intention serving the unique needs and objectives of ICP protocol and community.
Arguments for Embracing Liquid Staking on ICP
Many have pushed back against prior proposals to “liquify” staked ICP (either by creating neuron marketplaces or by developing liquid staking protocols) on grounds that this would be disruptive to effective NNS governance.
While I am sympathetic to this perspective, I believe that the ICP community should support a well designed liquid staking protocol for the following reasons.
- As noted, a successful liquid staking protocol would bring a large amount of additional liquidity to ICP’s growing DeFi ecosystem.
- As a practical reality, it’s hard to prevent creative solutions that circumvent neuron illiquidity. For example, there is already an active marketplace where internet-ID’s (most containing ICP locked in Neurons) are traded.
- Implementing a well designed, community managed liquid staking protocol could ensure minimal risk/disruption to NNS governance, while also maximizing the benefits to holders and to the ICP ecosystem as a whole (i.e. minimal intermediary profit taking).
- A successful liquid staking protocol would support ICP’s price in two ways: (1) it would result in a higher % of circulating ICP being locked up, shrinking the unstaked token supply, and (2) it would attract more demand for ICP among crypto holders/traders who are enticed by yield, but are unwilling to lock away their funds in an illiquid neuron for 200+ days.
- The governance model I envision would help to illustrate the incredible “composability” of the ICP protocol, by illustrating the feasibility of what would essentially amount to a (first of its kind?) decentralized joint venture dapp.
Proposed Token Minting Mechanism
New tokens could be “minted” by exchanging native ICP for staked ICP (“SICP”) - initially at a 1:1 ratio. The native ICP would then be locked away in a community/SNS controlled 8-year neuron. Over time, owing to maturity from the neuron (and redemptions; see below), the minting conversion ratio would decline such that each newly minted SICP token continues to represent a claim on the value of 1 newly staked ICP token (similar to cbETH, for example). The ability to mint new tokens “at-value” would serve as an effective price ceiling, capping the maximum price of SICP at the value of the underlying staked ICP (unless a minting cap were implemented, as contemplated in the “additional considerations” below).
Proposed Token Redemption Mechanism
It would presumably be possible for SICP holders to swap/trade SICP for ICP on a DEX. However, to ensure long term price stability of the SICP token, I believe it would be important to implement a formal redemption mechanism such that SICP could be converted back to ICP. This would ensure that, during periods of reduced demand for SICP, the total supply of the token could gradually decline to restore supply/demand balance.
This could be accomplished by offering SICP holders the option to “dissolve” their SICP back into ICP by depositing SICP into an a redemption pool. New maturity from the 8 year neuron could be released and used to redeem (on a pro-rata basis) the SICP in this pool at the current minting rate. After making redemptions, any left over maturity could be staked backed into the 8 year neuron.
It’s worth acknowledging: since redemptions would be rate limited by the 8 year reward rate, during periods where the rate of new redemptions exceeds the 8 year neuron reward rate, the redemption pool would become congested and it’s likely that the value of SICP would dip temporarily below the minting rate (as those seeking to convert more immediately from SICP to ICP would be forced to swap at below the minting value). Occasional downward price volatility (relative to ICP) is a trade-off holders would need to accept in order to benefit from a liquid ICP token that generates yield.
Proposed Protocol Governance and Administration
A secure, transparent and decentralized governance system would, from my point of view, be fundamental to the success of the proposed protocol.
I believe it would be possible to leverage ICP’s unique capabilities to implement a novel, “two-party” governance structure that would:
- ensure adequate developer incentives to build, support and maintain the protocol on a perpetual basis
- protect the interests of SICP holders by ensuring they have an ability to veto protocol modifications that are detrimental to their interests
- avoid centralization of the ICP network by ensuring SICP holders remain in control over the voting power of the protocol neuron
I’ve diagrammed this proposed governance structure below:
SICP DAO (SNS A)
- Approves DAPP changes (via SNS proposals)
- Votes on NNS proposals on a “pass-through” basis (votes are forwarded to and ultimately executed by the canister that hosts the protocol dapp and neuron)
Protocol Dapp (hosted on its own, separate and jointly canister)
- Executes front end for dapp
- Executes the minting and redemption mechanisms
- Automatically executes code changes (subject to the affirmative vote of both SICP SNS and the Protocol Admin SNS)
- Executes votes on NNS proposals (on behalf of and subject to votes by the SICP DAO)
Protocol Admin DAO (SNS B)
- Develops and maintains the Protocol Dapp (subject to SICP SNS approval of any code changes to the protocol app)
- Markets the SICP token and develops the SICP ecosystem (ensures DEX liquidity etc)
- Earns an ongoing administration fee (which cannot be modified without an affirmative vote of both the SICP SNS and the Protocol Admin SNS), paid via a small % of neuron maturity.
SICP DAO Voting Rewards & Protocol Admin Fee
As with any liquid staking protocol, a large majority of the token rewards should accrue to the benefit of SICP token holders. Otherwise adoption will lag and holders will migrate over time to alternative, less costly, competing protocols that would inevitably be expected to emerge.
With that said, to ensure the sustainability of the protocol, a small portion of the protocol rewards could be diverted toward (a) paying protocol app canister fees, (b) incentivizing SICP DAO governance, and (b) contributing fees to the protocol admin for developing, growing and maintaining the service.
An option that particularly appeals to me would be to reserve the full baseline maturity (net of age bonus) for the benefit of SICP holders, and use the incremental rewards derived from age bonus to cover the other items. Age bonus rewards could, for instance, be divided as follows:
- Sufficient amount of ICP reserved for purchasing dapp cycles
- 50% of the remainder (after reserving for cycles) distributed as SICP DAO voting rewards
- 50% of remainder (after reserving for cycles) distributed as fees to the protocol admin DAO
Reasons I find it appealing to allocate age rewards to cover governance cost and fees are as follows:
- It means that SICP holders (or at least, those choosing not to lock up tokens and contribute to SICP DAO governance) can expect to earn a predictable intrinsic yield - the standard 8 year reward rate - without the added confusion of a fluctuating age bonus rate (which would constantly be changing based on minting activity).
- This approach better aligners incentives. If SICP holders benefited directly from age bonus, they would have reason to push back on incremental minting (as minting additional tokens after an initial period reduces the average age bonus for all holders). This is contrary to the preference/interest of the Protocol Admin, whose stakeholders benefit by supplying/minting as much SICP as market demand permits (thus increasing fee earnings).
- Age bonus parameters for the SICP SNS DAO could be set to match that of a native $ICP neurons, creating fairness among SICP holders in the sense that long term SICP staked holders can benefit directly from incremental age bonus their tenure as holders produces, without that age bonus being unfairly reduced by new minters who immediately stake.
Impact of Maturity Modulation
Due to the recently implemented ICP maturity modulation function, all voting rewards, admin fees and redemptions are modulated within a range of +/-5% when converted into $ICP.
You can find more information about maturity modulation at this link:
Since redemptions are paid out with maturity, this wrinkle is especially consequential with respect to the design of the redemption formula. There are - speaking generally - two ways the protocol could account for modulation when making redemptions:
Option 1: Redemptions could be assessed on a “pre-modulation” basis, with the value of pre-converted maturity treated as equal in value to $ICP. Under this scenario, the redeeming party would bear the full impact (positive or negative) of modulation. Which is to say - based on recent price action of $ICP - final redemptions could wind up being anywhere between 95% and 105% of the allocated, pre-converted maturity.
Option 2: Redemptions could be assessed on a “post-modulation” basis. Under this scenario, any positive/negative impact of modulation would be born by the entire SICP holder base. To the extent that redemption activity tends to correlate positively with negative price action in $ICP, this design could produce a negative drag on the overall returns of non-redeeming $SICP holders.
Option 1 has the advantage of ensuring fair treatment for holders (who, under this scenario, wouldn’t stand to be negatively affected by another holder’s decision to redeem). It also creates incentive for SICP holders to game the timing of redemptions, which was actually part of the intended design of the modulation function (i.e. to smooth out upward and downward volatility in ICP’s price action). In this way, electing to implement option 1 might help to stabilize both the price of SICP (while indirectly helping to stabilize the price of ICP).
Option 2 has the advantage of keeping things simpler for SICP holders. It also avoids creating incentives for SICP holders to game the timing of redemptions, which could be viewed as a positive or negative depending on one’s perspective.
Of course, it is also possible to conceive of hybrid formulas that impose some, but not all, of the upward/downward modulation on redeemers (and which share the remaining impact of modulation with the population of SICP holders).
Token Distribution Cap
To avoid the possibility of oversupplying the market with SICP during periods of strong demand, a limit could be placed on the maximum number of SICP tokens available for minting (or, alternatively, a maximum minting rate could be designated). Changing these limits could require joint approval of the SICP DAO and the Protocol Admin.
Ability of SICP DAO to revoke protocol admin DAO
A rule could be established at the outset that enables the SICP DAO to revoke the Protocol Admin’s joint control over the Protocol DAPP after some period of time (enough for the admin DAO to generate a fair return for developing and popularizing the protocol) or in the event of failure to meet certain performance conditions (i.e. if the Dapp went down for a certain % of time, or if market swap rate of ICP/SICP fell below a certain level for a certain duration of time, indicating possible supply mismanagement by the protocol admin). In such an event, the SICP DAO could take over admin responsibilities directly, or could assign protocol admin responsibilities to a new DAO, individual or centralized organization.
Proposal that SNEED (or another well respected SNS DAO with an ecosystem oriented mission) consider taking on the role of Protocol Admin DAO
I believe the Protocol Admin role for this envisioned liquid staking protocol would complement Sneed’s “blank canvas” vision (Proposal to create an SNS DAO for SNEED) extremely well, as it would create a perpetual cash flow that Sneed could, in turn, allocate as grants/investments in the development of community projects and ICP infrastructure. Essentially, with a regular flow of admin fees from the SICP protocol, SNEED could essentially act as a second, community-voted, prong to DFINITY’s grant program. In turn, this would serve to accelerate ICP’s development and the growth of its developer base in a positive, community oriented fashion. Finally, such an arrangement would bring confidence to SICP holders that fees paid for access to the protocol will be reinvested in the success of the ICP ecosystem, and not simply cashed out as profit.
Disclosure: I personally hold a small amount of SNEED (I hold just over 3 tokens), but have no other personal incentive for making this suggestion. I have not, as of the time of posting, discussed this idea with SNEED’s developers/founders/community. I simply think this is a neat concept; I hope the Sneed community reads proposal and considers the idea.
The protocol admin function could theoretically be shared by more than one SNS DAO. For example, a project like SNEED could theoretically collaborate with one of ICP’s seasoned DeFi oriented DAOs (such as Sonic) and share admin fees.
I am not an experienced developer and I do not have the means to implement the envisioned protocol. I’m also not looking to benefit personally from the idea. Simply put, it’s a conceptual design that I’ve enjoyed thinking through, and I wanted to share my ideas with the ICP community for consideration.
Beyond my enthusiasm for the underlying liquid staking use case, I am broadly intrigued by the concept of harmonizing/harnessing the otherwise discordant incentives of different groups of actors (represented by distinct SNS DAOs in this case) toward a common purpose under a fully transparent “code as law” framework represented by a jointly controlled canister.
For this reason, I am excited by the prospect for a successful implementation of the proposed liquid staking protocol to help establish a novel(?) paradigm for innovative, cross-DAO/cross-canister collaboration - further bringing to the forefront an expansive vision of what is possible on the Internet Computer.