Compensation Proposal: Distribute COMP to Affected Users in the DAI Liquidations

Has anyone looked into the list that is going to be (or not) refunded to see if there are addresses that are extra/added by the governance proposer?


This is certainly something important to consider. Please observe the following chart

Although you can’t see them, for the entire length of the x address there are addresses that were liquidated that don’t show up because their relative size is essentially zero.

There is a disparity in the magnitude of losses that is so large that it makes those who didn’t lose millions in this event seem insignificant. I would like to see an amendment made to this proposal that incorporates this function which will weight the percentage of compensation with the magnitude of the losses:
percent compensated = 1/(1000*(x+.0012)) +.0782

Obviously, the subsequent values would be multiplied to the pre-existing data on how much DAI was repaid. For an easy view of how this amendment would affect each affected account please view this modified google doc I made that does exactly what I described above:

The newly adjusted total COMPensation (get it?) would be 9,855,567.80 with a majority of users being compensated a majority of what was liquidated. Please see the document for a more detailed spread of what liquidated values get what percent compensation. The difference between the total in this proposal originally and the amendment is 3,037,935.29. That comes out to +20,951 comp which would result in a replenishing period of not <60 as calculated by @kybx86 but rather 81 days.

In my mind, this doesn’t come out to a significant amount of money relative to the size and growth rate of the treasury. Additionally, the result of this amendment will better accomplish goals 1 and 2 and will maintain the benefits of the original proposal: minimal market impact, short time to replenish, trust between the users of compound and the governance.


This second allocation proposal seems more fair to the smaller liquidated accounts who lost more than the liquidation incentive due to incorrect DAI price, but still bails out the big farming operations that had been keeping their accounts at unhealthy levels for months while dumping COMP rewards. I think that instead of compensating the smaller accounts by increasing total compensation, it should come from the compensation originally allocated to the big farming operations, and perhaps cap the total compensation for any account to something like 500K USD value or lower. This would 1) compensate borrowers unfairly liquidated 2) cap the bailout on big recursive farmers 3) reduce total paid out COMP


I have voted against cause of 2 reasons:

  1. If we compensate a bug, then the bug should be fixed first.
  2. I would like to see a more precise calculation instead of just going a flat 8%, using price of the assets when the liquidation happend.

Here is a proposal that keeps the suggested formula to increasingly compensate smaller accounts, but caps compensation to a single account at 500K USD value. Instead of increasing the total compensation, it saves the Compound treasury about 200k USD of value of what the OP @kybx86 proposed. Worth noting that this cap only applies to the 2 biggest accounts that were recursive COMP farmers (aka “looper whales”)


I’m mildly against the idea of reimbursement and strongly against this specific proposal. It’s very important we separate these two things because this proposal is terrible. However, I don’t think voting against this proposal should be interpreted as voting against some form of compensation. These are separate issues and there are more ways to do this!

Considering this proposal in theory
Protocol insurance funds are a great idea! If the Compound protocol had one, it should be used in this case. However, the Compound protocol does not have an insurance fund. So now we are in a situation where we have not promised insurance but we may want to reimburse some losses anyway because we see it as being in the greater good of the protocol long term.

The protocol does have “reserves” for each market which have slowly been growing over the last couple of years. There are have been people in the community that have considered these to act like an insurance fund for the protocol. They have never actually been used in that way but it is a reasonable use case. Currently the total reserve for the Dai market is $1.3 million Dai, this is the largest reserve of any market and makes up a large portion of total protocol reserves.

If there is going to be any compensation, the amount of reserves in the Dai market should be the ceiling for it. This would be incredibly generous as it amounts to giving away a large portion of all reserves that have been earned in the last couple of years.

Considering this proposal in practice

Practically this proposal amounts to a massive bailout for two yield farmers who have been vampiring value from the protocol.

  • Per @wario’s comment 54% of this payment goes to one single yield farming operation and 61% goes to the top two yield farmers. The top yield farm was simply selling all accrued COMP (generating $2.15 million in profit for themselves).
  • These yield farmers are NOT long term aligned with the protocol. If they were they would not be selling their COMP.
  • These yield farmers were intentionally and knowingly engaging in risky behavior in order to earn millions of dollars by dumping COMP. With that reward comes risk that anything at that scale should clearly understand.

The idea of giving 0.55% of the total governance token supply to people who’s proven on-chain behavior is detrimental to the protocol is INSANE! We are literally giving extra power to the most short sighted actors in the system.

COMP is not money, it’s a governance token. We should not be giving away governance tokens to somehow try to act as financial compensation and the particulars of this situation only make it a worse idea.

For these reasons I encourage everyone to vote against this proposal. I also encourage @kybx86 to put forth a new proposal using the Dai reserves for a pro-rata compensation plan.


These 2 paragraphs should be emphasized alot more :slight_smile:


This is not true for all the users that were affected by the liquidation , also compound uses the COMP token as a collateral by doing that it’s already being used as some form of money .

1 Like

I need to correct what I posted above about the second largest liquidated account (0x189c2c1834b1414a6aee9eba5dc4b4d547c9a44c), it is not as clear cut. It was doing recursive borrow/supply of DAI to almost exactly the same amount, which is why the supply line of DAI in this chart is hidden behind the borrow line. But it was not dumping its claims as frequently as the big liquidated looper whale. LLW had 1525 claims, versus 46 claims for this second one. And the receiving addresses are also not as obviously exchanges as is the case for LLW.


Also need to correct, the top 2 accounts would make up for 61% of the proposed comp distribution, not 75%, my mistake. Here’s the original sheet with an extra column for the proportion of repaid capital for each one


I want to make a point that’s been coming up. I’ve heard a lot of “fix the problem” then compensate users. While im not against the idea at all, my question to the community is: what IS the problem?

To an extent, we can agree the oracle needs work. But, I’ve heard a lot of let’s just cap the price of DAI at ± 5% bounds, and then that’s it. That fixes it; then we can move on. This couldn’t be further from the truth. In fact, proceeding in this manner would just mask the issues; if there comes a day when DAI does dislocate beyond these bounds, the protocol will suffer material losses, and people would be pointing the finger back to the oracle. So it’s not just about the oracle; it’s about re-designing from the ground up a self-sustaining risk framework that works across all kinds of environments, not only a subset of probability outcomes (because isn’t that what just happened?). This won’t be built in a day or week. It’s a multi-faceted problem requiring the best ideas to come forward and drive the protocol design.

UMA has one of the best approaches to the oracle challenge: It’s a priceless design and its liquidations go through a dispute process–if we were to replay the events of Thanksgiving, having this dispute process active on Compound would have resulted in most of the liquidations to be disputed and reversed since it was clear that the global spot price of DAI was not $1.3. But that’s just a minor point.

To that end, the compensation proposal I’ve put forward is not uni-faceted nor final. In fact, it’s the second proposal I’ve put forward (with the first one having been the adjustment of the reserve factors, which’s now live).

My intent here is to 1) acknowledge Compound needs work. 2) Compensate all users affected, regardless of whether one’s moral compass likes/dislikes small users, large borrowers, looper whales, or whatnot. 3) continue to work towards the security and betterment of Compound.

One final point: While I’ve recently become more vocal on the forums, I’d like to share that I’ve been a long-term and loyal Compound user since V2 launched back in March 2019 (~2years), and most recently, since COMP distribution began, I’ve also been a COMP holder. I seek an outcome that’s a win-win for everyone involved, including the users affected, the community, and its stakeholders.


This is a fairer compensation structure for smaller investors whose businesses is significantly damaged by the “DAI manipulation event”.
But also I agree with @blck first reason :

Although this event damaged my business, but I think that for the benefit of the protocol, it is important to fix the bug first.
I don’t know why are we waiting for the implementation of Chainlink oracles?
Coinbase open price feed as the only source is a mistake, this raises the question: why was this solution voted on (proposal 019)?
Is there a secret agreement with Coinbase?
This question was addressed to all those who voted for proposal 019.
Whose fault is it?
@blck what is your opinion about this situation and what kind of compensation structure would be fair in your opinion? I am asking you specifically because you are a more experienced protocol user.


Thanks for all your efforts but based on the feedback from the community i feel like the community is willing to support the compensation as long as the compensation % is revised . there’s some good proposal from @wario and @jacobdecatur also maybe it’s worth considering using DAI reserves or combination between $Comp and DAI reserves . in term of fixing the bug ,that’s a more complex issue the discussion to use chainlink has already started here chainlink price feed .
I think this two conversation can go in parallel and they are not mutually exclusive , the compensation proposal doesn’t depend on fixing the bug first , also increasing DAI reserves is a step in the right direction


I dont have any personal problem with chainlink however it had its own problem in the past (2 which is big deal for compound), while compound’s openoracle also has its advantages and disadvantages, those 2 issues would not happen with compound’s openoracle and thats already 2 “bug” that we avoided by not using chainlink.


@blck what’s your logic in linking the compensation effort to the bug fix as if the bug is not fixed the compensation can’t go ahead ?

1 Like

the compensation part i already said that each user “loss” should be calculated on spot price when the liquidation happend but with “normal” DAI price this would give us the exact “loss” of a certain user a range of 8% to 25%~ probably


No strong thoughts on if it should be 8% or any other number, overall, I am for compensation. Maybe it is this proposal or a different proposal, but I think we should compensate users. The main reason I’d want to compensate borrowers is because it is still in the early days of DeFi and anything that builds borrower trust in the long term is valuable for the protocol. If the extent of the oracle risks were not clear to the end users, then it is on us as a community to improve our communication around the risks.


Polychain has decided to vote “AGAINST” Compound Proposal #032.

While we sympathize with those who were affected by the dislocated DAI price and subsequent liquidations, we believe that CP #032 sets a potentially dangerous precedent for a few reasons:

  1. This proposal provides reimbursement before a clear path to fixing the issue has been established. We would like to see a little more clarity around potential solutions before finalizing any form of reimbursement.
  2. This proposal reimburses users even when the protocol worked as designed, and may encourage further systemic risk. We do not want to set the precedent that tail risk should be subsidized by the COMP governance process.
  3. This proposal suggests reimbursement denominated in COMP to a small group of users affected by the issue, despite displaying behaviors that are not aligned with the long-term interest of the protocol. We would prefer a reimbursement proposal denominated in the asset that was liquidated, but are understanding of the additional technical complexity.

We would like to see continued discussion and proposals that work towards improving oracle robustness, implementing insurance funds to provide a finite backstop for unexpected protocol behavior, and exploring alternative methods of reimbursement.

The community has come to a loose consensus that the current oracle system worked as designed, but not as expected. The actual market price of DAI on Coinbase was accurate at the time of liquidation. Regardless of the outcome for CP #032, it is very clear to us that the governing community must work to reduce market risk and improve oracle resilience.

We are voting “AGAINST” this particular proposal, but remain open-minded as it relates to potential reimbursement. We applaud the efforts of @kybx86 to solve these key issues, and we look forward to working with them and the rest of the community to reach a possible solution.


Does the protocol now use its own open oracle? It isn’t Coinbase “price discovery” solution?
The bug is in the Compound protocol or Coinbase open price feed?
Coinbase states that is a normal market activity (30% stablecoin spike in one hour on their exchange while globally was stable).
What is “bug” in this situation?

1 Like

Pantera Capital is voting AGAINST Compound Governance Proposal 32 to distribute COMP to users affected by DAI liquidations on Nov 25 2020 as compensation.

Financial markets are more prone to hidden tail risks when participants deploy instruments with high leverage or for uses far from their intended purpose. Such risks are difficult to measure, which leads to scenarios where returns are attractive and too low at the same time.

As a Compound user (note: not farmer or private investor), we’re responsible for knowing the potential risks involved in using the protocol, including faulty oracle feeds and illiquid markets. Compensating users encourages looser risk-taking when efforts are still underway to measure and mitigate such risks. It also undermines the ability to properly assess risk vs. reward in Compound markets by artificially deflating perceived risk.

Open protocols can aspire to be trustworthy, but they’re permissionless by design. Usage will inevitably go “out of bounds,” despite disclaimers or risk parameters. Compensating users when known risks materialize is unsustainable in the long term.

We’re supportive of solutions that are sustainable and healthy for the protocol, including:

  • protocol insurance funds, with explicitly-stated intent and sources (e.g., market reserves),
  • multi-oracle designs, and
  • removal of “safe max” language in Compound UI.