Compensating vault holders that liquidated at 0 bid?!

Neils I want to let every person know who was liquidated here and lost all of their collateral all of us at Maker have the greatest sympathy. This was an unprecedented event and it is unfortunate that risks in total to the community may not have been made as clear as some wish. A lot of effort and attention was paid to doing just that and rest assured any inaccuraces in that will be corrected and addressed.

The market could have quickly gone to $50 and even if your collateral cleared it may have not been enough to cover your borrowed DAI. This is the real reality with collateralized loans where backing collateral value relative to the stablecoin being borrowed drops precipitously.

Even if the best market prices were paid virtually all borrowers who were liquidated AT BEST would have had no more than 21% of their collateral left in their vaults based on the current configuration of the system and I want all of you to know that Maker and MKR holders are basically eating the other 79% of the losses on all of the 0 auction bid losses.

I also want DAI holders to realize IF the system ever experiences a Emergency SHutdown the DAI PEG will theoretically follow a value curve related to the combined Collateral Oracle OSM price/LIVE collateral price. I skip doing the sum over BAT and ETH and simply consider ETH. IF the system is ES’d at $100 USD/ETH and the price falls to $50 USD/ETH the DAI peg value based on backing collateral in the system would be .5USD. Similarly if the price increased to $200 the DAI PEG value would approach $2 USD/DAI. These and many other risks are present.

I spent a long time before purchasing DAI and using the system to borrow before I entered these markets and hope others will do the same…

Again I am sorry for your loss and hope the value you hold of the loan at least provides some partial compensation for your collateral.!

5 Likes

Hello MakerMan,

Thanks for your answer. Can you please explain this paragraph?

1 Like

There is nothing left. I would be happy if had 21% colleteral back. I know it’s for a big part it’s my fault but the way it happened makes is difficult. Somoene had anticipated on this scenario to steal some colleteral. For me this is mentally difficult to cope with.

3 Likes

In my opinion, this was an ecosystem failure. We created a small Keeper monoculture. Since the health of the ecosystem and functioning of the peg are ultimately the responsibility of the MKR holders, I propose covering the excess vault holder losses through minting and auctioning MKR.

4 Likes

Let’s say mkr holders would agree.

As already mentioned by MakerMan:
“Even if the best market prices were paid virtually all borrowers who were liquidated AT BEST would have had no more than 21% of their collateral left”

I am unaware of any CDP owner suggesting what would be “fair” percentage for his (own) specific case.

1 Like

I think it would be very helpful for the discussion if we were all on the same page. Anyone cares to explain where this 21 number comes from?

I got it. At the time of liquidation 66 % of the cdp worth was already taken out as a dai loan. Add the 13 % liquidation penalty and you get 79 %.

2 Likes

Hello, community. I’m one of those users who were liquidated with nothing left.
I’d like to elaborate on point 1. I presume that:
a) Maker borrow app/site is built for users, not developers.
Hence:
b) Maker borrow app/site doesn’t warn users about potential loss over 13%.
c) It doesn’t ask if a user has read developer docs and fully understant all possible risks.
So I think it’s easy to amend Maker borrow app/site to put all responsibility on a user and avoid covering users’ unexpected losses in the future. As for now, you decide.

I want to add I use Maker for a long time since SCD and I get used to think that penalty is 13%. The app never didn’t challenge this my belief. Honestly when all the congestion has started and I had to quickly manage my CDPs I even hoped to get liquidated to avoid paying lots of money for many transactions (CDPSaver offered me to repay my dept on a single CDP for just $237). If I only knew all the risks I wouldn’t be so careless and don’t loose anything.

8 Likes

A one hour delay simply just isn’t sufficient, especially when you realize that the system is being used globally in multiple timezones. For me, I was 100% liquidated and didn’t find out about it until I woke up.

The entire crash happened within the span of just a few hours and a lot of people would argue that “a one hour delay and the network congestion wasn’t that bad for them to act.” is not a fair characterization.

2 Likes

I had a large (for me) CDP open, thankfully i managed to close it in time but I was very close to losing a lot.

Point 4 is important to me, if I am to consider re-opening a vault in the future.

The way it is phrased “could have been made less likely” downplays how likely this was with the old parameters, how much less likely it is now, and that MKR holders were previously made aware the old parameters were dangerous and failed to act (If I’m wrong about that please correct me).

A blockchain like ethereum has been and will again be congested for over 10 minutes during heavy price movement. There are plenty of historical examples for 10+ minutes of crippling congestion.

With the number of participants in the auctions, and 10 minute timeouts, $0 auctions weren’t some edge case risk that might impact one or two people individually but were highly likely to impact most auctions during inevitable network congestion.

The reason I think MKR holders should pay the 21% or whatever it is calculated to be based on data/debate is not “because of bad PR”, or because it could have been better documented (although it could be much clearer in the bit that users will actually read) or because it might be a large or small amount, but because mass $0 auctions were caused by an MKR vote.

I’d like to hear from MKR holders why they thought the old configuration was reasonable at the time, given what we all know about network congestion.

6 Likes

Hi, thanks for starting the discussion and i am happy to share my story and thoughts as a vault owner who got liquidated and lost my full collateral. I would like to reiterate and put up for discussion the 3 main points why i believe a compensation in MKR tokens (from foundation or inflation) should be considered:

1. Disclaimed Risk
I opened the CDP in the official app under the impression that the maximum penalty fee would be 13%, being liquidated at the Maker Oracle ETH price. In my opinion the UI should have said 13%+, link to the docs/disclaimers and explicitly explain full ETH price risks involved with liquidation auctions.

I do understand MakerDAO is in an earlier stage of it’s development and I will use Maker again if my initial understanding of the risk is fully met.

2. Responsibility & Risk Mitigation
After having my CDP open for a while, i have read through the docs briefly and understood, that the liquidation ETH price may vary according to Auction Bids. I trusted MakerDAO to have put in place a strong Keeper Infrastructure to mitigate the risks of the known issue that has been exploited a few days ago. Even i found it extremly hard to learn anything about the existing Keeper Infrastructure, I put my trust into MakerDao on the infrastructure side. I consider MakerDAO a legit project with an inspiring and well-funded development team. This team decided to promote opening CDPs via Coinbase Earn to less tech-savy users, which made me believe that the project and infrastructure is reasonably mature and may stand challenging market situations. If i had known the actual keeper situation, i may have under no circumstances risked a liquidation. In my opinion there should have been more transparency and risk mitigation. It should have been promoted and incentivized to act as a keeper, earn penalty fees and support the network.

I will use Maker again, if Maker admits responsibility to enforce a solid infrastructure. I will see the compensation of fully liquidated vaults as an indication of these.

3. PR and Trust Enforcement
I have used the CDPs as a tool to fund the on-going development of my company…so i convinced people around me to accept DAI and make themself familiar with MakerDAO…i have encouraged my peers to trust Maker and many have been using DAI and even opened CDPs as well…

If Maker decides to not compensate losses i may be discouraged to ever use it again and all my peers will feel that.

Furthermore it seems clear to me, that the value of rebuilding trust with early supporters outweights the value of the compensation massively. The compensation may look like:

  1. Figure out all liquidated Vaults on 12/13.03.2020
  2. Calculate ETH that should have been paid back to liquidated Vaults with the Maker Oracle ETH Price at the time of liquidation. The ETH Price may be discounted by ~5% which would represent actual conditions considering a rather solid Keeper Infrastructure.
  3. Mint/Allocate MKR that match the ETH value from point 2 at the time of minting those.
  4. Airdrop all liquidated Vaults from point 1 (transfer not more than 7 MKR per day to each vault address to avoid temporary selling pressure)

In my opinion a very fair deal. I would be happy to keep some MKR tokens and support the project growing up.

4 Likes

This is easier to say with the benefit of hindsight. In relation to how the old parameters were presented, I don’t think anyone present realised the extent of the risk at the time, if we had, we would have argued to change them back then. Here is the thread where they were discussed: Auction Parameter Derivations. Admittedly there wasn’t a lot of engagement at the time.

Personally, I didn’t realise it at the time either. If I were to try to justify it now, I might say: ‘Well keepers stand to gain significantly from auctions, so they’ll pay the gas prices even in the case of severe congestion,’ but in reality the issue of network congestion didn’t strike me as an issue at the time.

I’d like to hear from MKR holders why they thought the old configuration was reasonable at the time, given what we all know about network congestion.

I would also say that in my experience, this is an overestimation of what MKR governance as a collective knows about ethereum network congestion. There are a lot of aspects of both the Maker Protocol and Ethereum that we should know, and many of those involved have limited time with which to study up. Probably not something that is comforting to hear, but nonetheless true.

2 Likes

Thanks for sharing these points in a calm and rational manner. Hard to disagree with point 1.

Regarding point 2, there was some effort made to promote people to act as keepers, though this fell short of the mark. Part of what makes this difficult is that we can’t really make keepers sign up to a list so that we know how many there are (nor could we trust such a list even if we did attempt it.) Potentially the dev’s could build some sort of signal mechanism into the keeper bot, such that it called off to some API while it was active. This would be fairly trivial to spoof as a malicious attacker though. Needs some thought.

Point 3 is also hard to disagree with, it’s pretty obvious that people are feeling angry with the outcome, if we can smooth things over then we should.

  1. Figure out all liquidated Vaults on 12/13.03.2020
  2. Calculate ETH that should have been paid back to liquidated Vaults with the Maker Oracle ETH Price at the time of liquidation. The ETH Price may be discounted by ~5% which would represent actual conditions considering a rather solid Keeper Infrastructure.
  3. Mint/Allocate MKR that match the ETH value from point 2 at the time of minting those.
  4. Airdrop all liquidated Vaults from point 1 (transfer not more than 7 MKR per day to each vault address to avoid temporary selling pressure)

This seems mostly solid to me. A couple of comments:

  • On point 2: this will still result in some losing 100% of their collateral. Price moves were swift, and the Oracles only update once an hour.
  • On point 3/4: Mechanically the way this would work (if governance agreed to go ahead with this) is to mint unbacked Dai (not MKR) and distribute it to the affected parties. The bad debt created by this payout of unbacked Dai would then be dealt with via further MKR auctions (FLOPs).
3 Likes

Will it be possible for people who avoided 0 liq to get some compensation?

The 0 liqs going on resulted in mass panic and dumping of ETH - via closing CDPs through DefiSaver all at the same time.

If I got liquidated how I should have according to what was more in line with the documentation I would have been better off being liquidated.

But because I had to resort to defisaver and selling at massive slippage just to escape liq I’m still worse off than those who get compensated from the 0 bid action assuming they do get compensation

Is it possible ALL vault owners receive back the difference to what they should have gotten is my real question.

Not just those liquidated to 0.

Because those who also got absolutely decimated by slippage in trying to escape being liquidated to 0 should be no worse off than those who get compensated.

As you’ve said before Maker is EXTREMELY complicated. I am not sure if its actually possible to fully understand the entire system (risks), each part is constantly changing and all those changes interact with other parts, changing them. It is not reasonable to expect every user to understand those dynamics.

Maker is actively marketing the product. From the disclaimers I have seen on multiple UIs 13% is generally presented as the liquidation penalty. Presentation of the system in this manner is significant.

I also agree with characterizations others have made describing the keeper situation as a “failure”. MKR holders ratified foundation devs work on auctions and the mechanism did not work as intended.

Failure is expected in such a experimental system, complete collapse has not taken place yet so this is still an excellent learning opportunity. The desire to learn and grow yields deeper understanding. Here’s my thoughts on recognizing users that were misled:

First point:
Marketing needs to match the reality of the system. Maker is a really expensive experiment that basically no one understands completely. No one knows exactly how the maker specific software infrastructure, governance ( and all its messy social/economic realities), ethereum, and the wider crypto/global market will interface with each other. The robustness of MCD is unknown, it has only been live for ~5 months, and it was not released completely finished. Things like the GSM were to risky to implement immediately and the dark fix mechanism needed to be conceived. Migration from SAI hasn’t even finished…
Many other governance tools still need to be developed (both social and technical). For example on black thursday LFW and I were going to give the first presentation on a formal collateral on boarding process. Almost half a year into MCD and no clear path for onboarding collateral had been substantiated. Sorry for rant :slight_smile: . TLDR: This recognition does not compensate vault holders, but defines a responsibility for matching marketing (communications) to the presently understood reality of the system. This obligation aligns with “scientific governance” rather than business inspired marketing which aims to communicate simply and promote the system.

Second point: some sort of compensation:
Way trickier, its hard to judge how vaults “should” of been liquidated because its an auction, the market was going crazy, which would of influenced bidding regardless of how many keepers were participating. As Maker_Man stated at best 21% of the collateral would of been returned. Compensation should be less than 21% of loss collateral. Arbitrary, but how about returning 10% to vaults which lost more than 90% of their collateral? Calculated in terms of eth. I’m not knowledgeable enough on gathering eth data to do this, but I am pretty sure it would be possible to figure out.

Maker still isn’t in the clear, its an active situation, so in my mind the actual compensation for vault holders has to come in the future. The short term stability of the system is very much in question currently, adding a larger deficit right now might be dangerous. Also the dev team has little to no extra bandwidth.

Thanks everyone for their stories and rationales.

2 Likes

I agree with a lot of the sentiment for the CDP holders. The product was aimed at enticing users to come create Collateralized Loans on the platform. The users accepted and actively managed the risk of their CDPs and the product that was advertised to them. Any average user would be fine with any clearly stated slippage. (IE, Uniswap) There was no way for the average user to prepare for the critical failures of the bidding system/liquidity or to manage the risks associated with the inner workings of the platform, by definition that is the MKRs.

The documentation, UI, and even general chat did very very little to allude the risks to users. I’m a very active user in the crypto space, probably as savvy as a non-coding user gets and this still blind sided me. I spent time researching and trying to read medium articles prior to the dip trying to figure out what was my best option in general for that day. There was nothing that alluded to how complicated this system was and I had to go to chat just to get a somewhat passing answer and still got fooled with the 0$ bids when I decided to just pay a fee instead of taxing myself beyond a 13% effort.

At the very least setting everything about the users aside. 0$ bids in this market is a critical design flaw. There has never been 0$ ETH. The system was obviously not tested for a “Black Swan” event or w.e despite all the hub bub about the emergency shut down etc. Congestion is not a good excuse as we are all crypto veterans and understand congestion intimately. This was a critical failure of the DAO to provide liquidity and use the available markets/tools/resources available pro-actively.

I would be very very hesistant to ever use MakerDAO again until I know that the code has been reviewed in all the places where it’s just “assumed” that there will be enough good actors/liquidity. There should be fail-safes/plans for every outcome and a strong steady general belief that the system always needs review. Anything less and why bother. If it’s my responsibility as a CDP holder to audit your code and assume operational risk as well I will simply use the better market options provided and save myself the headache.

3 Likes

I think the DAO would greatly improve long term confidence if it covered liquidation at even a minimum fair market price minus the 13 percent liquidation fee. For example if someone had 100 ETH covering a loan of 5,000 Dai, even if they where liquidated at the bottom of the market price for ETH around 95 dollars they would have 40.52633 ETH remaining = 100 ETH - 5000/95 ( ETH sold at lowest market price to cover debt.) - ( .13 * 5000 )/95 ( 13 percent liquidation penilty). I think this would be a first step in repairing the damage done by running a monopoly auction or mono culture auction with liquidating assets.

Because of the 0 bid liquidations happening I had no other choice but to accept the price quoted via DeFi Saver in order to avoid liquidation. Because of the flaw in MakerDao I ended up being forced to sell Eth to payback dai and close my CDP at such a loss I was left with over 750 ETH less than I should have been.

Had I been liquidated I would have lost everything and had I been liquidated correctly I would have more ETH left than I do now.

But if compensation is on the table I feel like those like me should also be compensated the difference so everyone effected by what happened is compensated fairly and equally.

I hope this gets heard because the entire history of these events can be seen on my vault number 2288 and by that vault number owner address

Needless to say this caused a much bigger loss than necessary and I would have been much better being leverage long on any major centralized exchange. I don’t see how we as CDP owners can feel confident enough in this system feel confident enough to create dai again.

I believe there needs to be a show of good faith in doing right by all vaults both liquidated and force closed be compensated according to how things should have been under the correct liquidation conditions so that confidence in opening CDPS can be restored.

The issue is the “correct” liquidation conditions are determined by keeper liquidity which was highly strained on Thursday. AFAIK we dont have any other analogous historical situations with MCD, so there isn’t even precedent on how vaults are liquidated in these extreme downturns. Otherwise I agree with you.

Is this correct CDP? https://defiexplore.com/cdp/2288

I think it would be fair that the Maker multisig wallet (0x8ee7d9235e01e6b42345120b5d270bdb763624c7; 117,993 MKR) gives some of its MKR to affected users. We can use the MKR price before the drop as a fair value ($400-$500)?

Additionaly, how about taking some MKR from (large?) wallets that never voted?

It was a big mistake by the MakerDAO not to add a simple (*) on the Vault page by the number “13%” with a brief explanation of the total liquidation risk.

1 Like

Yes that’s correct - here is a damage report I put together detailing it: