Should voting power for Governance and DeFi mix?
Let me first address the OP’s (@Shaih) initial request:
We ask for your guidance on the fundamental questions of who should be considered an “active participant” for the purpose of governance, how to decide if a specific Algorand account belongs to that set and how many governance seats (and rewards) they should get.
Simply, all of the accounts that commit and vote in Governance are active participants. It should not matter if you participate in DeFi, own NFTs, just skim through the proposals (because you lack the time, knowledge or interest) but still vote, or you thoroughly read through them and engage in discussions before voting. Based on the above stated core values, all should be treated equally simply based on their committed ALGO stake alone, which defines individual’s voting power for Governance.
Now, this is not to say that we should not incentivize a more engaged participation in the Governance. I believe this can be done in a transparent and straightforward manner simply through how proposals are put up for vote (see the below section Update to Counterproposal to xGov).
Why I think allocating voting power for Governance to DeFi might be the wrong approach overall
DeFi does not include only ALGO but other assets as well. If those are counted towards the voting power for Governance, a foreign asset could overtake the Governance while having just a minimal stake in ALGO. Hence, I strongly believe the voting power for Governance should be based only on ALGO stake.
For the majority of DeFi solutions to be useful, the ALGO must be liquid which contradicts the current requirement of Governance to soft lock the stake for 3 months. If DeFi is granted an exception to this rule, which would be if they were to receive their voting power based on their average TVL (of only ALGO to not contradict point 1), the ordinary Governors would be treated unequally, contradicting the core value of equality between ALGO.
Even if we change the Governance system and remove the requirement of soft locking the stake for all, allowing DeFi’s ALGO stake voting power for Governance would still inherently clash with direct participation and equality principles. Consider the following two examples:
With lending protocols: Suppose you and I each put 1k ALGO in a lending protocol. Person C brings enough collateral to borrow 1k ALGO and uses it to participate in Governance. You and I want to vote differently on a measure. Each is supposed to own 1k ALGO to use in the vote, but there is only 1k ALGO in the pool because person C removed 1k ALGO from it. How to solve this? Of course you could simply say for the remainder of the pool, the vote is split amongst our contributions to it. But this is delegation of our voting powers. In this case it is simply a reduction of our own voting powers, where the difference is delegated to the borrower (i.e. another user of the DeFi platform).
With liquidity pools: Suppose you put 1k ALGO worth of liquidity into YLDY-ALGO pool on Tinyman. I do the same but on Pact. Because of the differences in usage, there will be a difference in the ALGO amount we own at any point of time. Although this difference might indeed be small in practice, we started at the same voting power point (i.e. were equal) but afterwards are inherently treated unequally. Here, we have essentially again delegated a part of our voting power (i.e. the difference in the voting power that you have simply because you are on Tinyman vs. me who am on Pact) to the DeFi platform/its users.
Note that these cases even assume that all DeFi protocols can implement the option for their participants to cast a direct vote without the intervention of the governing body of the protocol (which would otherwise be a point of centralization – e.g. if the governing body were to simply calculate each user’s voting power based on their contribution, and ask them to submit their votes, which would then be casted in an aggregated form through the governing body). The Foundation should not advocate unequal treatment, and definitely not points of centralization, especially since ALGO used in DeFi is expected to grow considerably, posing the danger of DeFi having the majority in Governance.
- Lastly, I would like to point out that DeFi does not offer the possibility to participate in consensus (with the exception of Algofi’s Vaults), and I have not yet come across Algonauts having issues with this. I think this just further highlights that Governance and participation in DeFi can be two completely separate topics.
However, if the majority is fine with the abovementioned altering of Algorand’s core values (even though they might be just slight), and considers the benefits to outweigh these values, then fine. But this should be made explicitly clear in such a vote that is submitted to the Governance. Anything else is just misleading.
End of part 4/5
Update to Counterproposal to xGov
The initial Counterproposal to xGov suggests a method on how proposals can be put up to a vote in Governance. Since the Governance is a direct vote by the electorate, the solution essentially mimics a referendum – a completely direct form of democracy. The counterproposal and the refinement described below simply adapts the procedure to the requirements of the Algorand Governance – having quarterly voting periods, a longer time commitment and higher stake for the proposers, aggregation of experience, and anti‑spamming. This is done while keeping the right to propose a measure with all ALGO holders, maintaining the equality among all ALGO.
The mechanism is modeled after this.
Anyone at any time can propose a new measure. Before that measure is put up to a vote, a predefined commitment limit (CL) must be reached. The commitment does not necessarily have to come solely from the entity that made the proposal. Any other entity can show support by adding to the commitment, thus contribute towards reaching the required CL. It is up to the proposer to find enough of this initial support. When CL is reached, the proposal is put up for a vote in the following Governance period (GP). The commitment consists of a time-weighted stake. I suggest the required CL to be 40 M ALGO in a quarter (for reasoning of this value see Protocol parameters). This means if a proposal gathers an initial support of 40 M right from the start and keeps it for a duration of one quarter, it would be put up to a vote in the next GP. If it gathered only 20 M, the commitment would have to be kept for two quarters before being put to the following GP. Adding additional stake is allowed, which speeds up the process of reaching the CL.
The vote itself has besides the options of “for” and “against” the proposal also an option “against with veto”. The last option is used to express strong disagreement with the proposal in case its quality is unacceptable or the content malicious. If this option wins, the commitment is slashed. I suggest to use a 2% penalty (for reasoning of this value see Protocol parameters). This mechanism prevents spamming of low‑quality or malicious proposals. For the slashing mechanism not to be exploited, only the already existing assumption that the majority of Governors act for the good of Algorand, i.e. without malicious intents to steal the commitments backing good proposals, is needed.
The committed stake is soft locked until the CL is reached. If an entity changes its mind about the support of a proposal, it can remove its contribution (in full or just partially). This would also clear correspondingly that entity’s contribution towards the CL. However, once the CL is reached, the stake is hard locked until the vote passes.
If the proposal is vetoed, the slashing percentage is sent to Algorand Ecosystem Resources Pool (AERP), and the rest returned to its supporters. If the proposal is rejected, the whole amount is returned. If the proposal is accepted, the commitment is locked for three more quarters. This ensures at least a one year commitment of the proposers towards the good of Algorand.
However, while the ALGO is either soft locked for support of a proposal, or hard locked after the proposal has passed, it can still be used to vote in Governance. While the stake is hard locked, it can still be used as initial support for another proposal.
The initial supporters for a proposal take on additional risks – possibility of slashing, longer time commitment and being associated with a potentially damaging proposal. Therefore, they will have to inform themselves well about the proposal. Moreover, in order to ensure the proposal passes, they will try to inform other Governors about it. This additional effort and taking on the risks should be rewarded.
Therefore, I suggest modifying the Governance rewards as follows:
- Reduce the rewards for voting to 120 M/year or 30 M/quarter. Assuming the current commitment of ~4B to Governance remains, this would equal a 3% APR. This would be a compromise, where DeFi could more easily compete with Governance, while the voters would still be decently rewarded.
- The remainder of the Governance rewards, i.e. 162 M/year (in 2022) should be used to reward the initial supporters of proposals that pass. Since not all proposals that will pass have the same difficulty associated with making them, I suggest to have three reward levels: minimal, middle and maximal. The suggested values for these are: 0.25 M, 1 M and 4 M, which correspond to 2.5% APR, 10% APR and 40% APR respectively with the CL set at 40 M/quarter. When a governor casts its vote, another vote should be submitted that indicates the amount of rewards the proposal deserves based on how thought-out the proposal is. When a proposal is initially set, it should be required to also submit its own suggestion how much rewards it deserves. The rewards for an accepted proposal are split between its initial supporters based on their contributions to reaching the CL.
Example 1 – Proposal to improve Algorand’s marketing:
- A proposal to change the shade of green in Algorand’s marketing just because the current one seems a bad choice to the proposer and likes another shade better. There is not much work for the initial supporters besides to consider if they agree or not. The majority of Governors in the end indeed find the suggested shade better and the proposal passes. This should receive a minimum reward since not much effort went into the proposal.
- A proposal to change the shade of green in Algorand’s marketing, where the proposer did a thorough research and studies that back the change of the shade of green (e.g. something like Facebook did for their blue). The initial supporters have to analyze the studies themselves and present the reasoning to other Governors. This required more effort, and should receive a middle reward.
- A complete and detailed plan how to improve Algorand’s marketing strategy is made. This requires the most effort both form the proposer as well as from the initial supporters to study and advocate it, thus should receive the maximal reward.
Example 2 – Proposal to change transaction fees:
- Simply because the proposer thinks the fees are too high, and suggests a new value that is agreed by the Governors. Minimal reward.
- A study was conducted among many users that the fees seem high and effects of the high fees studied on chain, based on which a new value is suggested. Middle reward.
- Besides the work of previous point, a new value is suggested based on simulation of the effect of the fee change. Maximal reward.
With the suggested reward structure for initial proposal supporters, we could incentivize 10 high-quality proposals per quarter. If more high-quality proposals pass, the rewards would be split among them accordingly. If not all rewards are spent in a GP (e.g. due to not enough proposals submitted or approved), the rest is sent to AERP (which can be used in the next period to e.g. incentivize DeFi or development).
The slashing percentage of 2% is based on the following simplified reasoning. High quality proposals are likely to get enough support quicker, e.g. to reach the CL in a single quarter. In this case, in oversimplified terms of estimating reward/risk ratio as reward percentage vs. slashing percentage is 10%/2% = 5. Low quality proposals will likely need a longer time commitment to reach the CL, e.g. in a year. In this case, again in oversimplified terms, the reward percentage vs. slashing percentage is 2.5%/2% = 1.25, where the potential reward is still higher than the potential slashing penalty.
The value of CL set at 40 M/quarter corresponds to:
- 0.4% of total ALGO supply (which is in a range commonly used in referendums to reach before a vote is called)
- 1% of initially committed stake in G3. If 10% of Governors are engaged in the Governance, we can put 10 proposal up to the vote each GP.
- Retail (here defined as accounts with a stake < 1 M) have 360 M committed in G3. Hence, 11% of retail alone is enough to put a proposal up for vote in the next GP.
- Just around 30 governors of G3 have enough stake to put a proposal up to vote singlehandedly.
Last thoughts, I think such a Governance structure would be a good compromise between incentivizing more engaged participation in Governance, thus improving its quality, while still keeping the existing holders satisfied. It would also help to skew ALGO dilution into the hands of active participants – where retail can easily participate, while exchanges would be discouraged to participate due to the risk of slashing. Further, it would reduce the barrier for DeFi to compete with Governance on the reward/risk ratio.
Possible additional considerations:
Because multiple proposals are likely to come in a single GP, I would consider adding an “abstain” option for each vote to encourage good decision making, e.g. if one feels uneducated enough to make a good decision, a vote should not be forced. To get rewards during a GP, the governor should not abstain on at least a half of the measures put up to vote.
End of part 5/5