[Agenda/Discussion] Scientific Governance and Risk #142 - Thursday, May 20 17:00 UTC

Agenda

The zoom waiting room will be on, and a password is set to: 748478, please ping us in chat if you aren’t let in from the waiting room. This Agenda will be updated over the following week leading up to the call as people make me aware they want segments.

2021-05-20T17:00:00Z

Introduction

Weekly Updates

I’d encourage those listed to try keep this to 5 minutes!

Procedural

Segments

  • Liquidation 2.0 performance analysis - @Primoz

Discussions and General Q&A

  • PSM Implementation
  • An anonymous Question Box is now available! Leave your question here and we’ll try to address it.
2 Likes

G&R #142 Snippet

Domain team updates, Collateral Onboarding, active discussions and more. All the updates from MakerDAO Governance and Risk Call #142 in this snippet

Team Updates

GovAlpha

  • Building documentation platform for broader audience.
  • Working on MIPs Portal adding views and menus.
  • Collaborating with SourceCred to make the distributions more trustless and automatic.
  • Eli has been tinkering with the forum, top level category changes and making it more pleasant.
  • Working on onboarding members into GovAlpha.
  • Forum at a Glance.

Protocol Engineering

  • Reviewed the delegated voting contract and the flash mint module.
    • Could be ready by mid to late June but the dependencies are front end integration and edge cases.
    • There’s a small edge case, the contract needs to account having time expiration in place.
  • Had an L2 strategy meeting.
  • Tabling the Keg for a little bit, might be utilizing something similar to DssVest for payments to teams in the interim.
  • Still need to pass last week’s Executive - necessary pre-req to getting the rest of collaterals into Liquidations 2.0.
  • Will be talking about Stablecoins and how we will be porting those over into Liq 2.0.
  • Safe to say Liq 2.0 was a wild success after yesterday’s stress test :tada:

Oracles

  • Exec to replace Oracle LP contracts is still waiting to pass…go vote.
  • We did considerably better with the heavy market volatility yesterday than we did on Black Thursday.
  • High volatility made for a lot of Oracle alerts, as they were updating every few minutes do to price movements outside the update band.
  • Yesterday’s gwei prices hit 3500, highlights difficulty in predicting Oracle costs for the DAO.
  • Working on a way to model Oracle Expenses, figure out expected range of costs for keeping a collateral type open compared to the fees they generate.

Real World Finance

  • 1.3M DAI coming from New Silver!
  • Shooting for an executive on June 11th with options for further RWAs.
  • Finishing review of 6s Legal Documents.
  • First RWA Committee Meeting occurred this week.

Risk

  • Proposed Parameters for Uni-LPs in Liquidations 2.0.
  • Looking at stable coins and possibility of adding PSMs.
  • Sushi assessments went live.
  • Risk Dashboard Road-map in the forum, check out the product here.
  • Liquidation 2.0 did well from a Risk prospective.
  • Might need to evaluate dust limit again after yesterday.
    • Incentives (like tip) do not cover the cost for small vaults, have to be careful of gaming tho.

Growth

  • Weekly Update.
  • Looking into DAI use on other chains.
  • Exploring “special use” vaults for bigger customers (like YFI).
  • Will be putting up post to explore promoting the ability to buy MKR on more exchanges.

Operations and Collateral Onboarding

MIPs

May Governance Bundle


What to expect next month for the June Governance Cycle:

New MIPs in RFC to look out for:

Other Presentations and Updates

Liquidations 2.0

3 Likes

Episode 142: May 20, 2021

Agenda

  • 00:00: Introduction
  • 01:40: Governance Update
  • 03:03: GovAlpha Core Unit update
  • 05:39: Forum at a Glance
  • 10:25: Protocol Engineering Team Update
  • 23:12: Oracles Team Update
  • 32:19: Risk Team Update
  • 36:31: Real World Finance Update
  • 39:14: Growth Core Unit Update
  • 44:40: Operational Support Team Update
  • 46:12: MIPs Update
  • 55:10: Submission Review
  • 56:13: Liquidations 2.0 Presentation
  • 01:18:00: Open Discussion

Video

https://youtu.be/9nwTPJo3d50

Introduction

Agenda and Preamble

LongForWisdom

00:00

  • Hello everyone, and welcome to the MakerDAO Scientific Governance and Risk meeting number 142 on Thursday, May 20th at 17:00 UTC. My name is LongForWisdom. I’m the Governance Facilitator at MakerDAO.
  • If you want to ask questions or have comments, please do speak up. Everyone here enjoys answering questions.

Weekly Updates

Governance

LongForWisdom

1:40

  • We had an executive on Friday, which had yet passed. Please, vote. It will bring SF changes. There are no major changes to the Uniswap LP Oracle’s new versions. Still, there are minor edits based on the audit results.

Polls

  • We had three polls regarding Uniswap liquidity pool tokens and moved those to the Liq 2.0 framework. The Risk Team proposed parameters for the Liq 2.0 framework for those Vaults passed and will be in this Friday’s executive. We had our Monthly MIPs’ Governance Poll for May, which also passed. This will be confirmed on Monday in an executive. If that passes, this will hopefully be the final executive we do on a Monday because we changed the Governance cycle.

GovAlpha Core Unit update

3:03

  • We have been working on building and setting up a more focused documentation platform for Governance. It’s still in the early stages. We are lifting documentation that we published elsewhere and organizing it to be more coherent.
  • We have been working on the MIPs portal and MIPs in general. Coordinating feedback with DSpot to improve the portal. We are also adding tags to the MIPs to support the new portal. You can now see a menu there, and we are going to add more stuff to it. So it will continue to improve.
  • I have been coordinating SourceCred around future technical work to make the SourceCred distribution process more trustless and automatic. We are looking into setting it up to not have manual distributions to ensure that the credit distribution is canonical if someone tries to do something rogue.
  • We have been figuring out how to improve the forum. Elihu has been working on this. I have made minor changes to the top-level categories earlier today. We are going to continue setting up the categories in a way that makes more sense for users.
  • Lastly, I have been working on figuring out how to onboard contributors into GovAlpha more efficiently. I want to figure out a good starter task that we can get people to do, things we want people to build up into, and generally get more organized on that front.

Forum at a Glance

Payton Rose

5:39

Forum at a Glance - Forum Thread

Three-Point Summary

  • Call it a stress test? Liquidations occurred with over 40M DAI worth of collateral sold at Maker on May 19th, 2021. Liquidations 2.0 showed serious strength :muscle:
  • Due to this wave of liquidation penalty revenue, over 1,100 MKR were burned in the 24 hours before this post :fire:
  • DAI is circulation has dropped to 4.4B as a further result.

Announcements

Discussions

Signal Requests

Protocol Engineering Team Update

Christopher Mooney

10:25

  • We did a security review and discussion of both the delegated voting contract and the flash mint module.
  • We had an L2 strategy meeting to discuss things such as expanding Maker to L2 networks. We also had a session regarding the keg, which helped us decide to move to a design that’s closer to the DssVest. So we will table the keg for a little bit because it has other useful functionalities that we could implement when needed. But it’s more complicated because we need it for sending funds to teams.
  • Last week’s executives are still out there and need more votes. This is a critical prerequisite for continuing the LP tokens to Liq 2.0, which everyone highly agrees is quite desirable. Liq 2.0 requires that it checks the OSM.
  • We’ll do the LP tokens this week and then finish up with stable coins next week. There is not much to do with stable coins besides cleaning up the technical debt and getting rid of Liq 1.2. Later in the call, we can discuss stable coins and the PSM, based on what Governance wants to do with the PSM for the rest of the stable coins. Finally, we can discuss how we would liquidate those stable coins into respective PSMs. If we’re not going to move something to a PSM, how do we deal with liquidating it?
  • Liquidations 2.0 was a huge success. Everyone who was involved did a great job of helping us accomplish this.
  • We’re also working on converting the clipper calls for Liq 2.0 over to the LP tokens and the auction demo keeper. It will be capable of liquidating LP tokens when it’s ready. Some MegaPoker or OmegaPoker work, but I am not sure what the exact details are.
    • Steven: Can you elaborate more on the thinking and timeframe behind the delegation.
    • Chris: We’re beelining it to the lightest touch or simplest implementation that we can. The irony is that most of the code was written back during Osaka; it’s been sitting around since Devcon. Sam updated it for modern times. We have one general concern with that implementation which is that it doesn’t expire delegate. There is concern that if you delegate MKR to somebody. That delegate passes away or something, then that person will have control of MKR up to infinity. We’re going to introduce expirations of 12 to 18 months because we do expect to have DssVest replacing the DssChief at some point. We have stuff for early-mid June, but I would bet that delegation could be ready by mid-late June.
    • Derek: The other dependency is the front-end work. Yesterday, we discussed with the JS team a bit of work involving dependency on completing the Liq 2.0. Mid-end June is a safe assumption, but their team has to size it. The code is mostly done, and we would like to perform an internal audit and some testing.
    • Chris: The good news from this approach is that it’s the modified version of the voting proxy, which means custodians can easily integrate with it.
    • Wouter: Are the delegates able and supposed to use vote proxy themselves?
    • Chris: The person who’s voting will be using the vote proxy interface to cast their votes
    • Kurt: Let me clarify a little bit. First of all, what’s the motivating question because there’s a separation between UI and the underlying contracts. There’s a delegation contract where people can delegate to it. The delegate is basically the owner or authority on that contract, and they interact directly with that contract to vote. There is no vote proxy per se, but it’s like the delegation contract itself is the voting proxy in some sense. So they could separately have their own MKR in the voting proxy. I am not sure whether this is a concern or a reasonable model?
    • Wouter: This is reasonable, but the practical concern is whether we need to support the voting proxy. At some point, we can completely replace it with the delegation system. My reason to keep the voting proxy is to secure the actual voting itself if you control a large MKR amount that you are voting with. However, the voting proxy can be deprecated after some time once the delegation system is in place.
    • Kurt: The utility of the voting proxy is that you have a separation between an address that can move the MKR and then another address that can only vote, and that’s the value of the voting proxy, right? Is it the separation of those two things?
    • Wouter: Yes, and the same separation is there right with the delegate contract.
    • Kurt: You have a very analogous thing where you can make your hot wallet or whatever as the owner of the delegation contract so that it can only vote. Then you delegate to yourself or the delegation contract from a cold wallet. Therefore, the address that is responsible for voting on proposals has no power to move the MKR.

Oracles Team Update

Nik Kunkel

23:12

  • The executive to put in the replacement Oracle LP contracts still hasn’t been executed. We’re currently using the older Oracle LP token contracts. Hopefully, we can get those soon because there are few edits that the audit caught, which are considerably gas efficient. Kurt and the Smart Contracts team did well in optimizing that.
  • Concerning events involving oracles, we did considerably better than Black Thursday. The lessons learned a year back helped us a lot in implementation. We do have oracle alarms going off left and right, but this is expected due to the extreme volatility.
  • We need to be cautious of costs. Many discussions in the #general channel concerning costs that the DAO is taking on. Oracle costs are a significant portion of them. For example, yesterday, the max gas price was 3500 Gwei compared to the normal day 100 Gwei, which means we’re spending 35 times the amount through Oracle costs in a single day. These costs are now handled by the foundation but are expected to be taken over by Protocol. Here are a couple of things we should consider:
    • How do you determine what the cost should be from an oracle coordinate perspective?
    • How much Gwei is too much to ask from the DAO?
    • What cadence do you want to ask for this money from the DAO?
  • We can ask this every month, but what happens if one of those executives were to not pass in time, right? We can end up sitting there thinking, ‘oh, shoot!’ If you start asking for a budget every three months and see a big spike in gas prices, this estimation would not work anymore. We’ve discussed this for the core units in the #general channel. We have an emergency multi-sig that we can dip into, part of the Protocol Engineering Core Unit proposal. It was also part of the Oracle Core Unit proposal where there is a reserve multi-state for times of protocol distress that we can dip into. We need a fund for oracle gas costs if we encounter crazy days like yesterday where we spend a month’s worth of oracle gas in a single day.
  • I’m going to be working on coming up with a way to conservatively model how much we need to spend. Part of that will include determining a collateral type and the expected costs or at least the expected annual range of costs to have a collateral type in the Maker protocol. This will give us a perspective of when we want to onboard something and how much revenue and stability a new collateral type will bring. If it’s less than the expected oracle costs, then it’s probably a no-go. This includes some of the existing collateral types. These ideas have been brought up before where we have a bunch of tokens that spend more on oracle costs on an annual basis than they make in stability fees.
  • There was a discussion yesterday involving ZRX, which has vaults that ended up being profitable for the protocol. Unfortunately, we also made a bunch of liquidation penalties. We should not account for revenue from liquidation penalties when considering if an onboard token is profitable. In addition, KNC, BAL, COMP are all unprofitable. Some of the LP tokens we added, such as DAI-USDT, haven’t had a single DAI minted against it, so that is definitely unprofitable. AAVE-ETH and LINK-ETH are unprofitable as well. It’s one of these things where we want to liberally onboard new collateral types. The demands for vaults have been all over the place. It’s difficult to predict what will happen. We do have tokens like YFI, which are runaway successes, and we have much higher market cap tokens such as COMP that have not been working. We can onboard liberally and test out the demand, but we can also off-board liberally as soon we realize that the demand is not there. There’s no point in continuing to take on the oracle costs if the demand is not being expressed. Oracle costs are just one of several costs associated with it; there are other costs with maintaining collateral, but I would argue that the oracle costs are the majority of the continuous costs of supporting a collateral type.
  • I’ll be updating this discussion in the forum, emphasizing oracle gas and how we can structure that from a DAO perspective.

Risk Team Update

Primoz Kordez

32:19

  • Last week, we proposed the Liq 2.0 parameters for Uni LPs. The remaining ones now are stablecoin vaults which also need new liquidation parameters implemented. There are a few ways on how to liquidate stablecoins. We’ll discuss this later in the call. Still, generally speaking, we have generic auction price curves and other parameters ready that can be used for all stablecoins even if you liquidate them into PSM, which was one of the ideas. We may discuss this today. Some of them are still over collateralized and are in the old state. We need to have this Liq 2.0 support nonetheless.
  • Sushi LP evaluations were done today @rema proposing similar parameters as for Uni LPs apart from stability fee, which is proposed to be higher due to yields on Sushi LPs leverage farming.
  • We released a new Risk dashboard for our site. You can read more about the roadmap on the forum. I saw many people use this dashboard yesterday when we had a lot of liquidation. I imagine the liquidations for the next OSM update were very popular. However, the site is still in beta. There might be a few errors, or some data may be updated late; don’t get scared when looking at those numbers right now.
  • We had liquidations yesterday—I published a presentation on the forums. Everything went well, and our metrics are great, but we’ll cover this later on. We also noticed that we need to reconfigure some of the parameters. Yesterday, we unwinded a $5000 DAI depth vault through high gas costs, which may have been unprofitable. We need to re-evaluate our dust limits concerning keepers kicking them and bidding for them.
  • Lastly, we discussed introducing a flat fee for kicking auctions, which is the tip parameter. This is different than the chip parameter, which is a variable incentive. The variable fee doesn’t offset the gas costs for kicking smaller vaults, and we knew about this. But there were always some edge case attack vectors. If the tip is too high, these flat fee keepers could be farming these kick incentives or reset incentives. We’ll re-evaluate this and propose a flat fee somewhere in the value of $300. it should be lower than the penalty fee on the dust size, which is $650.

Real-World Finance Team Update

Sébastien Derivaux

36:31

  • Good news! We now have more than 1 Million DAI coming from RWAs, and 1.3 Million DAI from NewSilver. In addition, these RWAs will be locked for 12 months, so there should be no issues even during market volatility.
  • We’re trying to publish an executive on June 11th with all the Centrifuge collaterals that you will want to onboard. The poll will go live either next week or the week after.
  • All the risk assessments are already out except for People’s Company, which should be done by tomorrow. Concerning the Trust Model, we were finishing the first review of the successor legal documents that Mathew shared in the forum last week.
  • We’re also working to see how we could incorporate a Maker representative of our company in the Cayman Islands. It’s similar to ‘Match Your ID.’ The collateral agents of the trusts are willing to do anything for money but don’t want to make any decision. We need someone to tell them what they need to do or trigger the launch of the script they need to follow. We need someone to give them a kick.
  • There is also the first RWA committee this week. This will help our progression a lot.

Growth Core Unit Update

Nadia Alvarez

39:14

Forum Thread: Growth CU Weekly-Update May 14-20

  • Thanks to Seth, we’re going to begin following DAI on other networks and see what’s happening outside the Ethereum network and Polygon. There is huge demand and lots of projects that are using DAI in Polygon.
  • We’re also having these initial conversations with new DeFi projects in Polygon that are going to launch. These include using DAI with various payment gateways and wallets.
  • We’ve also started conversations with YFI in building use cases about having custom-made vaults or industrial vaults for our largest vault users. It’s important to understand how users are using the protocol and help them. Please read the @ultraschuppi’s post about the YFI team looking to open even bigger vaults, and post your comments because we want to do this with other vault users like Nexo and Celsius. They are the biggest users of our protocol.
  • Lastly, we have the exchanges from different part of the globe which listed the MKR token. We didn’t know how to communicate with the community regarding how can they purchase MKR using their local currency. Please give us your ideas.
    • LongForWisdom: I think it would be great to create a forum post on the advantages and disadvantages related to the MKR token. The community can post their ideas and opinions there.
    • Nadia: Ok, I will do it by today or tomorrow.

Twitter Posts Adding MKR:
https://twitter.com/wirexapp/status/1btw394555898358603777

Operational Support Update

Juan Guillén

44:40

  • I posted the Operational Support Weekly Update in the forum.
  • @prose11 was presenting his candidacy yesterday. The recording is up on YouTube.
  • Next week, we have @monkey.irish presenting the Collateral Onboarding Core Unit, and on June 2nd, we have Nick presenting about Oracles.
  • We’re working on other workshops and MIPs, but nothing is on the agenda as of yet.

MIPs Update

David Utrobin

46:12

Submission Review

LongForWisdom

55:10

Liquidation 2.0 Performance

Primoz Kordez

56:13

  • Yesterday, we had the biggest liquidation event at Maker, like many other places in DeFi. For Ether, it was the second-biggest drop. The largest drawdown today was 46%, and Black Thursday last year was 48%. So it was a good stress test for Liq 2.0.

  • This is the cumulative debt being liquidated, including penalty fees, on one day, May 19th. It misses two other heat liquidations we had overnight. Roughly 50 million of debt was liquidated in one day if you include those. WBTC was liquidated initially, then ETH-A, and then some other Vaults like ZRX and ETH-B. Altogether, 50 million in one day.

  • Here are some performance stats more or less focused on penalty fees. I forgot to mention there were 177 auctions and that a 5,109,538 penalty fee was collected. Almost all penalty fee was collected successfully, apart from a few Vaults were it wasn’t. Three Vaults on ETH-A were unluckily liquidated at the worst point of time, considering the price curve. ETH-B has a lower collateralization ratio, so any discounts from auctions, from keepers’ profits, or just slippage related leads to penalty fees not being fully collected. And ZRX-A was a special case because it was liquidated in few batches. After all, we limited auction throughput because the launching liquidity is not that good. But all in all, 12,44% is the full number, so nothing to complain about.

  • I wanted to include the Loss Events to be fully transparent. This is a really small number of Vaults that got liquidated. When keepers liquidate a small Vault, it costs a lot in gas fees related to the debt being bought, so they bid lower prices. On ETH-A, you can see that the slippage towards OSM was 28%, but the loss was 81DAI. For ETH-B, it is not as small. Again, because of ETH-B, any slippage related to keepers’ profits and bad price trajectory versus the price curve leads to a loss. We had a 12 thousand DAI loss. The total loss for Maker was 12 thousand versus 5,1 million of revenues.

  • This is another metric that is also relevant. There are three parts to the whole thing. How much penalty fees Maker collects, the profits of keepers, and how much Vault gets back. This is especially important, and this metric might be confusing. It checks how much collateral a Vault owner got back versus his total deposited collateral. If the liquidation ratio is high, you normally get more back. For something like 175 liquidation ratio Vault, they should get back on some normal prices by about 30% to 35% for the remaining collateral. This is what we are mostly seeing. Whereas at some lower collateralized Vaults such as ETH-A or even ETH-B, those shares are smaller because the liquidation ratio is smaller and the penalty fee is higher related to the whole collateral. For ETH-B, the normal number is about 15%; it should be closer to 20% for ETH-A. This could be better, but it wasn’t mostly because of those two Vaults unluckily liquidated at the worst possible time.

  • This chart shows how you measure Auction Efficiency. Normally, it is related to the OSM price. We know OSM price is left by one hour. If keepers are a bit late, if they want more profit, you have these delays in bids if the price trajectory is worse than the auction price curve. The expected bid is about 40 minutes, but if the price keeps tanking faster than the auction price or if keepers want to make more profit, we have delays in bids. When this happens, the price is lower than the OSM. For instance, the average discount on all the ETH-A auctions was about 6% to 7%. If we wanted to have 100% efficient auctions, it would be somewhere around the 0,00% line or slightly higher depending on how the price trajectory goes. Normally, you want to have 100% efficiency, zero slippage in OSM, but why does this happen? One reason is just the auction price curve versus the market price trajectory. It is not perfect, and you cannot just determine it because you do not know how the prices will move. The other reason is related to keepers. Here, efficiency looks at settlement prices at auction versus the market prices. We wanted to check the market price at the time of the settlement, whereas we looked at the off-chain real-time prices. This is not necessarily 100% correct because many keepers just flip collateral on on-chain venues. It is not necessarily one-to-one synced with the auction prices. This is just an estimate, but it is about there. It shows that keepers who were bidding, for instance, on ETH-A, were actually bidding on average 6% or 7% below market prices. This was an estimated profit that is not 100% correct, but they need to take something for themselves. What does this basically tell us? This slippage that you see versus OSM is not necessarily related, just too bad pricing or bad price trajectory versus auction curve. Still, keepers are not 100% competitive. They are not working with zero profits, so we see part of the Vault performance because of this—still, good numbers in general. What may be worth adding here is that some smaller Vaults have higher slippage because gas costs are high. After all, keepers will bid lower because the liquidity tokens of other Vaults are not as good. When keepers offload, they need to assign the slippage, which they can already see on Uniswap, for instance, so they bid lower. That is why we have throughput limitations, as we did yesterday for ZRX.

  • Same thing but focused on ETH Vaults. Here you see the two Vaults that were liquidated at the worst possible time.
  • Yesterday we had luck, and suddenly we had not. The price was updated right before the drop, which means that the auction price started at a higher price, which is around 2440. Then, the price kept tanking, about 20% in 20 or 30 minutes. The auction price curve could not catch the market price, which was dropping, and that is why these two Vaults were liquidated at much lower prices, about a 25% discount to OSM. These are the Vaults that did not carry the penalty fee, and this is why we had a bit worse performance on ETH-A.
  • Otherwise, all the other Vaults were healthy. Everything up to 10% discount is good. Again, it is a random process.

  • This is the distribution of settlement times. The expected auction duration is about 40 minutes if keepers do take nothing for themselves, if prices are stable, and so on. On average, in all the Vaults types, we were between 30 and 50, so it worked as it should. But because of this random volatility, some Vaults got liquidated later. Here, you can see these two ETH Vaults again with 80 to 200 minutes. As soon as it goes above 100 minutes, it becomes for ETH-A. It’s important to mention that they would already carry a loss if those two Vaults were ETH-B.

  • This is Keeper Participation for KICKS events when they trigger an auction. Surprisingly 18 keepers were kicking auctions, especially knowing that retail did not kick any auctions. It is also good to notice that these are not necessarily 18 different entities because they may have many wallets. However, 18 different entities while it is kicking is always good. There is also the amount that they received as an incentive. The variable incentive is a very cheap parameter, and it was set at 0.1% of Vault. They get the incentive parameter back when they kick a Vault. This is the part that we need to re-evaluate because some smaller Vaults were not necessarily kicked immediately. This is why we need to introduce this flat fee that I was talking about earlier. There are mostly three keepers that kicked most of the auctions.

  • These are the entities that purchased collateral. There are 21 of them, and retail is included here as well. The amounts are great. Compared to Liq 1.2, where we had one to three people bidding, this is 10 or 20 times better. It looks good. There are again three performing entities, but that is probably Vaults’ related, so we want to see that as well.

Conclusions

  • The main conclusions are that we need to introduce the flat fee. There were some discussions to see if we should make the auction duration shorter. There are trade-offs to this. I have made a whole post on the forum about it. Definitely, for the two problematic Vaults, it looks that the duration should be shorter, and everything should be fine. However, it is tough to say if this would be the case because if you make the auction duration shorter, it means the auction price curve is more aggressive, is steeper. This means that if you have some delays, even of Vaults, you could get in a similar scenario with a 10% to 20% discount to OSM. This is possible because if you make a shorter auction, delays are more probable. You can have a more congested network, etc. For ETH-B, we should make it more aggressive because it is riskier to lead the losses. After all, any discount from 10% to 20% can lead to losses to Maker really soon.

Questions

  • David Utrobin: (About the slide “Keeper Participation (KICKS)”) Your guess is that of 18 unique keepers, some of them might be run by the same groups?
    • Primoz Kordez: Yes, it is possible. Someone could make an auction analysis and confirm this.
    • Christopher Mooney: Is it possible that these could be from the Web3 UIs as well?
    • Primoz Kordez: I have been thinking about it, and it could be.
    • Nik: I don’t know if anybody kicked through the UI, but I know of at least one person who bid through the UI and said the experience was great and profitable. We are still in a regime where it is not all Vaults winning these, which I think is interesting.
    • Primoz Kordez: Yes, it is good no to rely on Vaults solely, because if it is one or two of them and then you have this tragedy when Vaults will take all the volume, and then they stop bidding, but one Vault goes offline, then you have an issue. It is good to see this kind of bidding.
  • Frank Cruz: Thank you for that presentation. What is your thinking behind raising the dust parameter? Doubling it, tripling it?
    • Primoz Kordez: It is hard to say. Monet-supply made a table where you input gas cost parameters, and you simulate how much it costs. He focused more on Liq, but this was back on 1.2. Here we need to make a different simulation because it works differently now. I believe it is less gas-intensive, but this is only one part of the story. The other part is related to the user perspective. They want to unwind, but if it costs 2000 dollars to unwind and his debt size is 5000, he will not do it. We need to make those calculations, and I cannot tell you a number right now.
  • Kurt Barry: I think the Liq system performed really well yesterday despite a bad market event. This is the worst that we have seen since Black Thursday by far. But I also want to not celebrate too much because this is the warm-up for whatever happens at the end of the current Vault run. Today everything is green and going back up. Still, there will be another one of these in the future and probably worse, so this is a dress rehearsal. We need to ensure that we iron out any issues around profitability, dust levels, keeper incentivization, etc. This was a great data point, but we should not get complacent here. We should take it forward and figure out how to make this bulletproof.
    • Christopher Mooney: I will second that. There was a moment when we were staring down the barrel of a one billion dollar Liq wave. That made me feel despair. Overall our box is set at like 100 million right now, but we need to think about what the system can push through it carefully. Obviously, the DAI supply dropped by half a billion or 400 million or so. That is, in a way, soft Liq, but it is worth thinking if this is a dress rehearsal for a bigger event, what can we actually do.
    • LongForWisdom: These are great points. I am sure that you guys and Primoz will continue to worry about it for the foreseeable future. When you are ready to present proposals, everyone will be happy to see them.

Open Discussion

01:18:00

PSMs

  • Christopher Mooney: We have to figure out what we are going to do with stablecoins. The mandated actors were starting to discuss what we might want to do with stablecoins going forward. It feels like most of the decisions turn around whether or not Governance and the Community sees the need for a PSM for various types of stablecoins. There is one possibility for low liquidity where stablecoins could be put in a PSM with a zero DC, so you cannot add any more stablecoin types, but you can provide DAI and get the stablecoin out. In theory, it would be one-way. You cannot throw the stablecoin back in and then get DAI out of it. Maybe we will do that with TUSD. Maybe we do that with PAX. We wanted to open to see what people were thinking. Maybe create a signal thread about exactly what Governance wants to move into a PSM style format. Once we know that, we can set up a custom liquidator that would liquidate Vaults that are undercollateralized into that PSM-style setup on genuine collateral types of stablecoins we have left. We can think more critically about how we want to handle those and Liq for those.
    • LongForWisdom: It feels like the issue is to figure out which stablecoins we want PSMs for.
    • Christopher Mooney: That is the first branching decision, and that is what I was hoping everybody would think about.
    • Primoz Kordez: I am going to add here that there are six stablecoins. USDC-A is going to be liquidated for PSM, which exists for USDC. We have USDC-B, which is totally unutilized, DC is to zero. Doing anything to it will cost on-chain more. The same goes with TUSD: the DC is set to zero. It feels like with USDC-B and TUSD is not reasonable to spend any time. Then we have PAX USD, which is unwinding now because we introduced 1% SF. This token is pretty liquid lately. Last time we checked, if you sell 5 million of it, there is already 20% to 30% slippage on-chain. I am not sure about PSM here, but the solution where you only allow a certain amount of DC makes sense to me, so I propose this. It goes similarly to TUSD. The launch liquidity is better, but this Governance voted to limit DC to zero because of debt acquisition. We can do the same with PAX USD. We can introduce these two PSMs limited. The only one left is TUSD. It is probably one of the most prospective ones currently looking from the on-chain activity and different farming types of activity. It is not under collateralized, so we cannot put it into PSM. The question is just to introduce PSM for this one as well. There are other stablecoins like Binance USD…
    • Nik Kunkel: Yes, but Binance USD is just a white-labeled PAX USD.
    • Primoz Kordez: It has better traction in DeFi farming, though.
    • Nik Kunkel: On BSC or on Ethereum?
    • Primoz Kordez: I think on Ethereum, but I might be wrong.
    • Nik Kunkel: I could be wrong as well, but I feel like I am very active in the farming community, and I have barely seen BUSD used anywhere.
    • Primoz Kordez: It is a curve, and the slippage curves checked are much better. It is pretty big, at least compared to PAX USD.
    • Matthew Robinowitz: Can I just step back for a second and understand our overarching strategy and reasoning behind the PSM with other stablecoins? Is it to diversify away from how much we are using USDC? Is that correct?
    • Christopher Mooney: If we are going to have an active one, that would definitely be the case. It would be to limit collateral risk. There is a sort of benefit because this is a much cleaner type of liquidation. If we need to liquidate these, we don’t take any pricing risk or slippage risk. We also have a sin que problem. If we were to try and liquidate TUSD as Vaults, all of that would go out to Liq 2.0. It would have some undefined price curve that hopefully would fetch a price but would probably have some slippage. We probably do not have enough liquidity to trade against it, and we would probably take a loss, and it would go to MKR holders. The alternative is, you throw it into a PSM with a zero DC, and people can just come along and grab it whenever they need liquidity, and eventually, it tends to zero. It is beyond diversifying collateral risk. The PSM is a liquidation mechanism in a way.
    • LongForWisdom: I am interested in knowing if anybody from the Community has strong opinions about any of the stablecoin options beyond USDC. Does anybody think we should definitely have PSMs, or anybody that think we should definitely not have PSMs? These are the things we are trying to figure out. We did have a vote previously about moving various stablecoins to PSMs. But that is a little old now, and the environment has changed in several ways. So if anybody wants to organize further signals, as Chris said, or just wants to open the discussion, that would be helpful for the Protocol Engineering Team and the rest of us.
    • Payton Rose: Especially if we allow both ways on some of the income sources not tied to the market fluctuations, I like that. It adds stability to the DeFi ecosystem. DAI is truly at a dollar if you can redeem it for all these other dollar coins at any time. So just in general, I would support it as much as Risk thinks we could take on. It would seem like a reasonable course of action from my perspective.
    • Aaron (chat): We would love GUSD PSM.
    • LongForWisdom: I think that is reasonable. GUSD is maybe one of the contenders. Primoz, you were saying that GUSD has not great liquidity, right?
    • Primoz Kordez: It is not great compared to USDC, but it is better than PAX. It is also questionable how much we can diversify to other stablecoins because all the arbitrage is still likely to happen with USDC. We should force it, but then we need to limit PSM on USDC to push others in. I am not sure.
    • LongForWisdom: If we actually wanted to diversify, we would need some mechanism that manages the DCs of them better. You cannot have one that takes up all of the room. I do not know how that would work.
    • Primoz Kordez: It is also risky because you still want to have a high buffer available for USDC. If there is a worst-case event, a huge DAI shock…
    • Matthew Robinowitz: It is a DAI shock in the upward price direction, but if for whatever reason it happens to be a flood of DAI into the market that subsequently pushes it below one, all of the stablecoins are going to go to zero, all of our exposure would go to zero.
    • Primoz Kordez: That will not happen anytime soon. I made a post on the forum about this. There are 800 million DAI on the curve first, and then PSM starts draining. So there is 1.5 billion, which is now probably lower but 1/3 or 20% of DAI supply. So I do not believe that could happen soon.
    • David Utrobin: I think the comment that Nik made in the chat is important. That we take on the failure of any of the stablecoins, we decide to support with the PSM. Unlike Vaults with a stability fee with a risk premium priced in, PSMs have the toll fee in and out, but I do not know if that is considered from a risk perspective. If we hold on to 800-900 USDC, it is relatively safe, but maybe we are ok with doing BUSD for 50 million or 25 million. Is it really just the toll fees that are a buffer against that kind of risk? My question is, where is the risk hedging for taking on more PSMs and more stablecoins?
    • Nik Kunkel: To expand on the context of that, putting something in a PSM is very different from just onboarding it as an asset. When you onboard a collateral type, i.e., the KNC token, if the price of KNC drops, we liquidate, and that is how we get our money back. With PSM, we own those tokens, and we have complete exposure to them. The only way to limit the additional exposure we take on is with a DC for that PSM, but we take it as a loss if it goes to zero. If the thing fails, we should expect the entire DC to allocate to this thing just to be filled up by arbitraging people. The risk of loss for any stablecoin failing is the DC that we set for the PSM, and the little in and out fees are not going to matter for ensuring that loss.
    • Sébastien Derivaux: I think one of the risks is the fact that currently, our main competitor is USDC, and I am quite sure they do not care about us because we are too small, but if they start caring about us and we have 1 million USDC in a Vault, they might start to lobby to try to blacklist this PSM. That is the issue of having only one option. If we have many options, we can quit the current party risk of USDC. I understand that currently, USDC is the best option but having other options in the future if there is a need could be interesting.
    • David Utrobin: In general, given everything that everybody said, I would favor two or three PSMs for the most popular, less risky stablecoins.
    • Nik Kunkel: The problem is nothing will see as much use as USDC. You can have three PSMs, but how much are you really diversifying when you have a billion USDC and 25 million GUSD.
    • Sébastien Derivaux: It will not, but we have the facility and ability to transfer USDC to GUSD. It will not be easy, but at least we will be able to.
    • Nik Kunkel: I wonder if there is some kind of pre-configured way we can authorize to liquidate the entire USDC PSM if we needed to do it in a hurry. Some way where you can have a pre-signed transaction and then you just put it in a Bot which is checking them in a pool, and you try to front-run any blacklisting. Whoever is running that Bot is dangerous. Because you could potentially liquidate the PSM anyway, even if there were not a blacklist transaction, I am not sure if that is even viable.
    • Christopher Mooney: You would see the transaction instantly flash mint the equivalent of DAI, create an LP token and drop it in an LP pool. It is no longer in the PSM; you probably will earn fees from the LP…
    • Nik Kunkel: Even that or just trading it into DAI or something like Curve or Uniswap V3 or whatever. You will take on some loss but compared to the blacklisting risk; you avoid its ruin.
    • David Utrobin: I am curious about the design decision from Protocol Engineering about potentially adding blacklists to things like PSM preemptively so that if there is an ETH address that a government agency blacklists, we could do the proactive thing and go along with that. But I know there is a lot of ideology that might shape people’s views on this. So I would be super curious to hear the conversation.
    • Nik Kunkel: Once you have given yourself the ability to, then all of a sudden, you have the obligation. We should be cautious about implementing any censorship on top of our permissionless platform because we may find ourselves in a position where we are obligated to comply.
    • David Utrobin: That is a very good insight.
    • LongForWisdom: On that cheerfulness, let us wrap the call. Thank you all for coming!

Common Abbreviated Terms

CR: Collateralization Ratio

DC: Debt Ceiling

ES: Emergency Shutdown

SF: Stability Fee

DSR: Dai Savings Rate

MIP: Maker Improvement Proposal

OSM: Oracle Security Module

LR: Liquidation Ratio

RWA: Real-World Asset

RWF: Real-World Finance

SC: Smart Contracts

Liq: Liquidations

Credits

  • Artem Gordon produced this summary.
  • David Utrobin produced this summary.
  • Gala Guillén produced this summary.
  • Sai Teja produced this summary.
  • Everyone who spoke and presented on the call, listed in the headers.

This topic was automatically closed 91 days after the last reply. New replies are no longer allowed.