Compound Delegate Analytics Dashboard

Compound Delegate Analytics Dashboard

This post highlights the results of the Governance Metrics grant program, an initiative created to help the Compound community better track and understand the current and historical state of the protocol’s governance.

The Governance Working Group (StableLab, Arana Digital, & PGov) awarded this grant to the WOOF team.

All dashboards and datapoints can be viewed on Dune here.

Dashboard Highlights

Ten total sections comprise the dashboard. Below are the outlined sections—

Common Governance Information

  • General trends regarding the amount of COMP delegated as well as the number of COMP holders over time are underscored.

It’s evident that the start of 2024 was a bit rocky in terms of delegation, with some large token holders pulling their delegations between Jan 2023 and Q2 2024—but trends reversed mid-year, especially due to the notable governance attack and the subsequent delegation from the COMP treasury via the Governance Working Group’s delegation initiative. COMP holders, however, have remained rather stagnant over the past two years.

Compound New Proposals Growth

The below illustrates a significant increase in the new collateral amounts and types across the various networks that Compound operates on.

Although total collateral has gradually fallen, there has been notable expansion in markets. Continual increase in the collateral numbers for these markets indicates positive sustenance and development for the protocol over the past couple of months, despite more laggard growth from mid 2022 - mid 2023.

  • Governance Impact: The TVL growth highlights the success of governance decisions in introducing and managing new collateral types.
  • Risk Management: The stability of TVL growth over time suggests effective risk management practices in collateral selection and parameterization.
  • Future Focus: Continued growth may depend on expanding to new networks, introducing additional collaterals, and maintaining competitive risk/reward profiles. This trend is important in the context of competing with emerging lending protocols.

Proposal Stats

  • This section outlines the details pertaining to each individual proposal, similar to the data accessible on platforms like Tally, including the outcome of each proposal and its respective onchain sponsor.
  • Cumulative unique proposers have also increased recently, indicating a larger diversity of eligible delegates with the ability to meet the proposal threshold. It also notes collaboration among governance stakeholders to help each other sponsor onchain votes.

COMP Token Stats

  • Each address’ voting power and COMP ownership amount is displayed and queryable in this section.
  • Although token ownership is quite concentrated, with a16z for instance holding a large quantity of COMP, there is healthy diversity in delegate amount—at least relative to many other DAOs. Future voter incentive and treasury delegation programs can help increase the breadth of delegates as well. This will be a focus for the year to come.

Delegates’ Effect on Protocol

Delegates’ contribution to the protocol’s TVL is a more experimental statistic that may be helpful in assigning a tangible numerical value to a particular voter’s impact on the protocol’s growth. It answers the question—how much in dollar terms did a delegate contribute to the success of Compound?

The above table divides the TVL added to Compound by proposal number. For example, the most impactful proposal in the given timeframe was number 268, which called to initialize a Compound v3 USDT market on Ethereum Mainnet. The current TVL for that market is portioned against the voting power of each delegate who voted For that proposal.

In this scenario, Wintermute held the largest amount of comparative voting power, assigning them the largest dollar “impact” as a result of this proposal.

The other table in this section shows each delegate’s individual TVL contribution to specific collateral type by market, again relative to the delegate’s proportional voting power.

Delegate Score

In order to best implement coming programs like delegate compensation, it’s important to establish a system for allotting scores for voting contribution. Below is a description of a potential implementation of a delegate score. We are keen to hear the community’s input on this present setup before we move forward with instituting a formal delegate reward program. This is subject to change over time.

  • Participation Score: This measures the frequency of the delegate’s participation in on-chain votes. It is calculated as the number of votes the delegate participated in divided by the total possible votes, expressed as a percentage.

  • Voting Impact Score: How impactful delegates’ votes were for each proposal. It is calculated by the delegate vote divided by total proposal votes, averaged for all the votes, and min/max normalized between 0 and 1. This means, if a delegate voted 50K votes among 500K votes for the proposal, the voting impact score for this proposal will be 0.1 (50/500). The scores are then averaged for each delegate and min/max normalized. Many DAOs have implemented delegate scores simply based on voting participation—we are trying to add a much needed level of sophistication to that metric by allowing voter impact to determine a portion of the overall delegate score.

  • Voting Before/After Threshold: This measures timeliness in voting by assigning a score based on whether the delegate voted before or after the proposal reached the 400k cumulative vote threshold. 0 points for no vote, 0.5 for after threshold, 1 point for before threshold.

Weight:

  • Participation Score %: 60% weight
  • Voting Impact Score %: 20% weight
  • Voting Before/After Threshold %: 20% weight

Formula:

  • PR: Participation Score %
  • VIS: Voting Impact Score % (normalized between 0 and 1)
  • VT: Voting Threshold score

Delegate score = 100 × ((PR × 0.6) + (VIS × 0.2) + (VT​ x 0.2)​)

Participation Statistic

This section provides some descriptive statistics regarding voting trends and participation.

  • The overall number of voters per month has been on a general decline since inception, with intermittent increases in engagement, in particular during between Q2-Q3 of the past three years.
  • Participation rate has remained relatively stable throughout 2024, teetering around 20%. This number is determined by dividing the amount of COMP that was used to vote over the total COMP delegated.
  • There has been a significant upward trend in the number of proposals submitted over the past year, indicating the crucial role that governance continues to play for the sustenance of Compound.

DAO Costs

  • Since onchain voting and proposing cost gas, these dashboards outline each voter’s and proposer’s costs associated with conducting transactions.
  • In the event that the DAO would like to institute gas rebates, these data can be referenced. Such rebates may be baked into a generalized delegate compensation program, or they could stand alone. The important consideration here is deciding which delegates are eligible for voting rebates since the cost to vote is constant, regardless of voting power.

Governance Attack Affiliated Accounts & COMP Transfers and Delegations

  • These two dashboards will allow the DAO to monitor large delegations and COMP transfers, enabling delegates to check voting power flows.
  • Wallets associated with governance attacks are listed in the first dashboard. If these wallets become active, it may warrant further investigation by the DAO to ensure no malicious activity is taking place.
  • If these metrics were tracked leading up to the governance attack in Q3 2024, delegates may have been more keenly aware of socially contrary behavior by certain wallets, thereby voting down associated proposals.
5 Likes

I looked at the Dune

a few things stand out
Woof
Openzepplin
Gauntlet
DoDAO

all had low scores. If I look at Tally, they dont seem to be voting on most of their own proposals (unless Im blind)? I mean, that makes sense to me but if these four offer up the most proposals but dont vote on them, wont their govenance score always be extremely low? Doenst seem right if we are going to use these scores to drive delegate compensation and the top 4 proposals authors get little

1 Like

Thanks for your comment. The first aspect to address would be what entities should be compensated. Should this program sustain eligibility for all entities, exclude service providers, exclude risk assessors, etc? Multiple variations can exist. For example, for Aave, service providers are not allowed to be funded in their version of a delegate reward program. However, other DAOs like Uniswap and Arbitrum don’t make that clear distinction.

This concern was also noted during the treasury delegation program at the start of Q3 of last year. Delegates based on their participation history were allocated a handful of COMP voting power directly from the treasury. The criteria was simply voting rate and did not exclude any entities. We’ve highlighted voting participation rate since the start of the delegation program, stating that those who want to partake in future programs should begin actively voting, even if they lack sizable delegation. The weights between the three current criteria can also be altered to favor authors who may not have as much voting power—up for discussion.

Another way we can accommodate authors is by adding a fourth variable that counts the number of onchain proposals that a certain entity successfully passes. This can be like 10% of the total score for example to give authors a boost.

1 Like

all good points to discuss, I dont have the answers but maybe the community could come up with some

Another nuance, I looked at the dune dashboard list of delegates and compared to proposal author addresses, seemed like Gauntlet uses multiple multisigs? I mean Im not surprised but the data in the dune dashbaord didnt aggregate the wallets under “Gauntlet”

it took me a only a minute to find that one discrepancy, perhaps the data needs another look to make sure it is correct

That’s right, Gauntlet has two addresses. I believe the reason for this is so they can post more than one simultaneous onchain vote, which comes in handy for market updates.

cc @Gauntlet